Most investors buy an S&P 500 fund to stop picking stocks. Cap weighting has quietly picked one for them: about 34 paise of every rupee in that fund now sits in seven companies running the same trade. On Saturday the people building frontier AI asked the industry to slow down, and on Wednesday the Federal Reserve is expected to raise rates into it. Knowing how a theme becomes an index is the difference between owning 500 companies and owning one bet held 500 ways. That's why we built Winvesta Crisps: to break down what's actually moving markets, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

On 12 September, Anthropic's Dario Amodei called for frontier AI development to be deliberately slowed, proposing outside evaluators inside the leading labs, common safety benchmarks, and agreed limits on how fast systems are allowed to gain capability. Sam Altman of OpenAI and Elon Musk both endorsed the idea within a day, per reporting in the Washington Post and CNBC. President Trump dismissed it on the Sunday, saying a slowdown would risk America's lead over China. Here is the part that matters for your portfolio: not one of Anthropic, OpenAI or xAI is in your index fund. The listed companies that would actually pay for a slowdown are, and per September 2026 data they are roughly 34 per cent of the S&P 500. The AI trade stopped being a sector call some time ago. It is now the index.

📉 How the most diversified product in finance got this narrow

The S&P 500 was sold to a generation of investors on a single idea. Buy all of it, own the American economy, stop guessing. For most of its history the arithmetic backed that up. The top 10 holdings averaged about 24 per cent of the index, and the previous high-water mark was 28 per cent, set in 1970.

That number now sits somewhere between 37 and 41 per cent depending on the provider and the date, per RBC Wealth Management and figures compiled by Pensions and Investments. The seven largest, the ones the market keeps calling the Magnificent Seven, are about 34 per cent on their own. The other 493 companies share what is left.

Three things got it here, and only one of them is about AI.

Cap weighting is the first and it is not a flaw. An index that weights by market value will always hold more of what the market has decided is worth more. That is the design, and for decades it produced a spread that looked like the economy.

The second is that one theme happened to be concentrated in very few listed names. Semiconductors, cloud infrastructure and the memory that goes into a data centre are industries with two or three serious players each, not twenty. When capital spending on AI accelerated, it did not spread across 500 balance sheets. It landed on about a dozen.

The third is flows. Index domestic-equity funds held roughly 18 per cent of the entire US stock market's value at the end of 2024, per Investment Company Institute data, ahead of the 12 per cent held by active domestic-equity funds. Broader definitions that count internally managed index portfolios and closet indexing put the passive share far higher, above 50 per cent in some academic estimates. The honest version is that the measurement is contested and the direction is not.

⚙️ What cap weighting actually does to your money

Work through one rupee, because the mechanics are where the risk hides.

You send 1,000 rupees into an S&P 500 index fund. The fund does not divide it 500 ways. It buys in proportion to market value, so roughly 340 rupees goes into seven companies, and the smallest 100 companies in the index share something closer to 20 rupees between them. You have not made a concentrated bet. You have accepted the one the index made on your behalf.

Now run it in reverse, which is the direction nobody models. A 10 per cent fall in those seven names takes about 3.4 per cent off the index before a single one of the other 493 has moved. A 20 per cent fall takes closer to 7 per cent. For that to be cushioned, the remaining 493 have to rally hard at exactly the moment the largest companies are falling.

They are not positioned to do that, and this is the part most concentration commentary misses. The other 493 are not independent of the trade. The biggest individual gainers in the index this year have been the AI supply chain rather than the AI platforms: Sandisk up over 700 per cent, Dell up around 245 per cent, Micron up around 243 per cent, per year-to-date figures compiled in early September. Those are not diversifiers. They are the same story with a different ticker and more operating leverage.

Then add what happens on Wednesday. August consumer prices rose 0.4 per cent on the month with core at 0.3 per cent, faster than expected, and producer prices came in at 0.4 per cent on the month and 5.4 per cent on the year. Odds of a rate rise at this week's meeting jumped to about 90 per cent from 70 per cent in a day, per CME FedWatch, and prediction markets put a 25 basis point rise at roughly 90 per cent for the 15 to 16 September meeting. It would be the first rise since 2023, and futures are pricing two by year end.

A higher discount rate hurts long-duration growth assets most. The seven companies carrying a third of the index are the longest-duration assets in it, because their valuations rest on cash flows years out. So the oil shock that pushed Brent to around 105 dollars and West Texas Intermediate to around 101 dollars is not a separate story from the concentration story. It is the mechanism that turns concentration from a statistic into a drawdown.

The dynamics covered in this article affect every US stock in your portfolio. Trade from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!

Get the app!

🏆 Who a narrower index helps and who it hurts

Concentration is not a risk that sits in one place. It redistributes, and the split follows who needs the AI capital cycle to keep running.

The useful thing to notice is that most large portfolios sit on both sides at once. An investor holding an S&P 500 fund, a Nasdaq 100 fund and a US technology fund does not own three positions. They own the same seven companies three times, at three different expense ratios.

🇮🇳 What this means for an Indian investor holding US equities

Start with the position most Indian investors are actually in, because it is not the same as an American one.

An Indian portfolio that holds a Nifty index fund plus an S&P 500 fund looks like 550 companies and two countries. It is not. The Nifty's top 10 were 53.58 per cent of that index as of 30 June 2026, with financials around 37 per cent of the whole thing, per index provider data. Reliance alone was about 9 per cent in mid-September. So the real shape is two concentrated positions: one on Indian banks and the rate cycle they trade on, and one on US AI capital spending. The number of stocks is high. The number of bets is two.

The second thing is specific to India and it is uncomfortable. You may not be able to fix your US concentration through a domestic fund even if you want to. The industry-wide overseas investment cap of 7 billion dollars, with a separate 1 billion dollar allowance for overseas ETFs, was close to exhausted again by May 2026. Axis suspended fresh inflows into select overseas schemes from 6 May, and Nippon India and Kotak are among the houses that have paused or capped subscriptions. As of the middle of this year there was no announced plan to raise the cap, and AMFI has asked more than once. A closed fund cannot rebalance you into anything. The route that stays open is direct US investing under the Liberalised Remittance Scheme, where you choose the weights yourself, including the option of an equal-weight fund.

Third, run the currency maths, because the shock absorber Indian investors have relied on for a decade was switched off three weeks ago. A falling rupee has quietly added to rupee returns on US portfolios year after year. This year roughly 5.75 percentage points of a rupee investor's gain on the S&P 500 came from the currency rather than the index. The Reserve Bank has just raised 136.37 billion dollars through its special swap window and pushed reserves to a record 740.80 billion dollars, specifically to stop the rupee falling further from its 96.82 low in May. It is trading nearer 94.50 now.

The table below runs the arithmetic both ways, holding the other 493 companies unchanged so the concentration effect is visible on its own. It is an editorial model, not a forecast.

Put those together. If the top seven correct meaningfully, the index falls, and the rupee cushion that used to soften a dollar drawdown is being actively fought by the institution best placed to fight it. Indian investors have been paid twice on the way up. They should not budget for being paid twice on the way down.

🔭 What to watch

Five things, each one checkable by anyone with a browser.

Wednesday's Fed decision and the rate path that comes with it. The rise itself is close to fully priced, so the information is in the guidance: whether the committee signals the second rise that futures already expect, and how it describes an inflation problem that is largely coming from energy.

The gap between RSP and SPY. Equal weight has been ahead of cap weight this year, with the size of the gap varying by source and measurement date, after three years of losing to it. That spread is the cleanest free breadth indicator available. Widening in equal weight's favour says participation is spreading. A sharp reversal says the index is narrowing again.

The Magnificent Seven weight itself, published monthly. It is about 34 per cent. A move through 35 per cent tells you the problem is getting worse while everything looks calm. A fall back under 30 per cent would be the first real evidence of broadening in three years.

Whether the slowdown proposal acquires any legal form. Right now it is three executives and a White House that said no on 13 September. Voluntary safety benchmarks cost the listed companies very little. Statutory limits on capability growth or on data centre build-out would cost them a great deal, and that difference is not yet priced.

Hyperscaler capital spending guidance in the late October earnings round. Capex plans are the cash flow that the semiconductor and memory names are valued on. One cautious guidance line from a large buyer does more damage to the supply chain than any safety proposal.

If this changed how you read the phrase "diversified index fund", pass it on.

Share

🏁 The bottom line

Nobody reading this chose to put a third of their US exposure into seven companies running the same trade. Cap weighting did it, gradually, and it was the correct decision every single year on the way up. The index is not broken and this is not an argument for selling it. It is an argument for knowing what it currently is, which is a concentrated position in the AI capital cycle wearing the clothes of a diversified fund.

The awkward timing is that three cross-currents have arrived in the same week. The people closest to the technology are asking for a slower build. The Federal Reserve is about to raise the discount rate that the most expensive part of the index is most sensitive to. And the rupee tailwind that has flattered every Indian investor's US returns is being deliberately held back. None of those alone would matter much. Together they are a reason to look at what you own rather than what you thought you bought.

The fix, if you decide you want one, is unglamorous. Know the number. Decide whether a third in one theme is the number you actually want. If it is not, equal weight and deliberate satellites do the job, and a position you sized on purpose behaves very differently in a drawdown from one you inherited by arithmetic.

📊 By the numbers

• About 34 per cent: the seven largest companies' combined share of the S&P 500 as of September 2026, with the other 493 sharing the rest

• 37 to 41 per cent: the top 10 weight depending on provider and date, against a long-run average near 24 per cent and a prior peak of 28 per cent in 1970

• 53.58 per cent: the Nifty's top 10 weight as of 30 June 2026, with financials around 37 per cent of the index, so Indian investors own concentration at home too

• 18 per cent versus 12 per cent: index domestic-equity funds' share of total US market value at end-2024 against active funds, per Investment Company Institute data

• 90 per cent: the odds of a 25 basis point rise at the 15 to 16 September Fed meeting after the August inflation prints, per CME FedWatch, which would be the first rise since 2023

• 7,585.73, 52,093.11 and 25,981.57: where the S&P 500, Dow and Nasdaq Composite closed on 15 September, all lower on the day

• 105.68 and 101.39 dollars: Brent and West Texas Intermediate settlement prices on 15 September, the energy shock behind the inflation the Fed is responding to

• 7 billion dollars: the industry-wide cap on Indian mutual funds' overseas investment, close to exhausted again since May 2026, which is why several international schemes are shut to fresh money

• 5.75 percentage points: the share of a rupee investor's gain on the S&P 500 this year that came from the currency rather than the index, a cushion the RBI is now spending to remove

Disclaimer: All content provided by Winvesta India Technologies Ltd. is for informational and educational purposes only and is not meant to represent trade or investment recommendations. Remember, your capital is at risk. Terms & Conditions apply. Sector impact and portfolio impact figures in this article are illustrative editorial models, not reported or verified data.

Five years ago, owning gold from India meant a locker, a jeweller's making charge and a vague sense that it would hold its value. Not anymore. Gold funds worldwide now hold a record 4,189 tonnes after taking in roughly $18 billion in August, the second largest monthly inflow ever recorded, per the World Gold Council. The metal itself topped out at around $5,600 an ounce in January and trades near $4,290 today. That's why we built Winvesta Crisps, to decode what's actually moving the funds you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Two records were set this year, nine months apart, and almost nobody has noticed they point in opposite directions.

The first was the price. Gold peaked at roughly $5,590 to $5,600 an ounce on 28 and 29 January 2026, capping a run that had already delivered more than 60% in 2025, the metal's best year since 1979. It then fell about 28% into a seven month low near $4,000 in June, the steepest quarterly correction since 2013.

The second record was the ownership. Global gold ETF holdings hit an all time high of 4,189 tonnes in August, with assets under management at $615 billion, per the World Gold Council. Indian investors did the same thing at home: gold ETF inflows reached ₹2,596.70 crore in August, up about 67% on July's ₹1,559 crore and the third straight month of net buying, per AMFI data released on 10 September.

So the largest wave of gold buying on record arrived seven months after the top, into a metal that is up roughly 0.6% for the year in dollar terms.

Most investors see a gold rally and assume they are late to a winner. What the data says is stranger. They are early to a position that has already stopped working, and the Federal Reserve meets today and tomorrow with roughly 89% odds of raising rates, per fed funds futures pricing reported on 14 September.

🎯 Meet Kabir

Kabir is 41, runs a mid-sized logistics business out of Pune, and has been investing in US markets from India since 2021. His financial assets come to roughly ₹85 lakh, split between a broad S&P 500 fund, a handful of US large caps, and about ₹6 lakh he put into an Indian gold ETF across July and August this year.

His family also owns gold. Perhaps 400 grams of it, mostly inherited, sitting in a bank locker in his mother's name. He has never counted it as part of his portfolio. That omission is the first thing worth fixing.

He bought the gold ETF for reasons that sounded sensible at the time. Domestic gold prices on the MCX rose 7.7% during August alone. The rupee was weakening. Everyone he follows was talking about central banks hoarding bullion. He was, in his own description, hedging.

Since he finished buying, gold has gone the other way. It closed August at $4,431.82 an ounce and trades near $4,287 now, a fall of roughly 3% in a fortnight. The 10 year US Treasury yield touched 5.014% on 14 September, its highest since October 2023, per CNBC.

His question is the one worth answering this week: he owns gold in two forms, he added to the noisier one at what looks like a local high, and the safest asset in the world now pays him 5% to do nothing. What should he actually do?

📊 What you are buying when you buy a gold fund

A physically backed gold ETF is close to the simplest product in finance. The fund holds bullion in a vault, issues shares against it, and charges you a fee. There is no cash flow, no earnings, no management team to get it wrong. The share price tracks the metal minus the fee, and the fee is the only thing you control.

The spread between those fees is the whole decision on the US side. SPDR Gold Shares, ticker GLD, is the oldest and most liquid at roughly $141.7 billion of assets as of June 2026, and charges 0.40%. The iShares fund, IAU, charges 0.25% on about $80.2 billion. State Street's own smaller sibling, GLDM, charges 0.10%. Aberdeen's SGOL sits at 0.17% and vaults in Switzerland rather than London, which matters to a particular kind of investor and to nobody else.

All four hold the same metal. Over a ten year hold, the difference between GLD and GLDM compounds to roughly 3% of the position, which is real money for an asset that has returned about 0.6% so far this year.

Gold miners are a different instrument wearing similar clothing. The VanEck Gold Miners fund, GDX, is up about 13.2% year to date against GLD's 0.62%, because a miner's costs are broadly fixed while its revenue moves with the metal. All in sustaining costs for large producers typically run around $1,200 to $1,400 an ounce, so a move in the gold price lands almost entirely in the margin. That operational gearing works in both directions, and GDX has drawn down more than 45% inside the last year.

Read more

Last Friday this newsletter told you that you wanted the August jobs number to come in soft. It came in at 162,000. Four sessions later crude is above $100, the US 10-year yield is knocking on 5%, and the Federal Reserve is being argued into a rate rise it never planned to make. That is why we built Winvesta Crisps, to cut through the noise and tell you what actually moved markets and why. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

US stocks fell in all four sessions of this holiday-shortened week, and not one of those days was about a company. West Texas Intermediate crude went from about $93 to $102.48 a barrel in three trading days, the 10-year Treasury yield pushed above 4.95% for the first time since October 2023, and the market's odds of a rate rise at next week's Fed meeting climbed from roughly 60% to around 70%, per CNBC and CME Group futures pricing.

The August CPI report lands at 8.30am New York time on Friday, which is 6pm in India, after this article reaches you. It is the last inflation print the Fed sees before it meets on 15 and 16 September.

📊 Market recap

The week ran in one direction, and it got steeper each day. Markets were shut on Monday for Labour Day, so this was four sessions, and all four were red.

Start with the damage. Against last Friday's close of 7,718.60 on the S&P 500, 26,506.99 on the Nasdaq Composite and 53,414.25 on the Dow, Thursday left the S&P at 7,591.70, the Nasdaq at 26,081.72 and the Dow at 52,064.10, per CNBC and Investrade. That is the S&P in the high-7,500s and down roughly 1.6% on the week, the Nasdaq in the low-26,000s and down about 1.6%, and the Dow in the low-52,000s and down around 2.5%. The Russell 2000 fell from 2,960 on Tuesday to 2,890 by Thursday, so small caps took it worse than the headline numbers suggest.

Tuesday reopened the war trade, and the Dow wore it. Traders came back from the long weekend to news that Iran-backed Houthi rebels in Yemen had attacked Saudi energy facilities, per CNBC. Brent moved up towards $98 and WTI settled around $93.03, up about 1.7%. The Dow fell 626.72 points, or 1.17%, to 52,787, while the S&P 500 lost 0.58% to 7,673 and the Nasdaq slipped just 0.32% to 26,421, per Investrade. Look at the spread between those three numbers, because it explains the whole week: the industrial, energy-consuming, rate-sensitive index got hit four times harder than the tech index. Healthcare led the decliners after Novartis failed two late-stage drug trials. Canada's retaliatory tariffs on US goods also took effect just after midnight, per the Washington Post, which nobody had room to care about.

Wednesday, the bond market joined in. Brent settled up 3.36% at $101.21 and WTI gained 3.25% to $96.05, the highest settle since May for both, per CNBC. Then the Treasury Department said it would triple its buyback of longer-dated government debt to $6 billion, and the 10-year yield jumped roughly 15 basis points to about 4.85%, its highest since November 2023. The S&P 500 fell 0.48% to 7,636.36, the Nasdaq 0.64% to 26,253.34 and the Dow 404.99 points, or 0.77%, to 52,381. Energy was the only sector doing real work, up 0.63%, with technology up 0.29%. Consumer staples, consumer discretionary and industrials all fell about 1%, per Investrade. Apple held its first product event under new chief executive John Ternus the same afternoon and unveiled a foldable iPhone starting at $2,099, per CNBC and Yahoo Finance. It barely dented the tape.

Thursday was the day oil stopped being an energy story. WTI closed at $102.48, up 6.7%, its first close above $100 in nearly two months. Brent gained 5.9% to settle at $107.63 after briefly trading above $108, per CNBC. The war between the US and Iran is now in its seventh month. The 10-year yield went with it, rising about 10 basis points to roughly 4.94%, with the 30-year at 5.362%, the highest 10-year reading since October 2023. August producer prices landed in the middle of all that: PPI rose 0.4% for the month, in line with the Dow Jones consensus, but 5.4% year on year against a 5.3% estimate, with final demand energy up 4.2% in the month and 24.4% on the year, and diesel alone up 24.1%, per CNBC and the Bureau of Labour Statistics. Core PPI was the one soft spot at 0.2% against a 0.3% forecast. The S&P 500 fell 0.58% to 7,591.70, the Nasdaq 0.65% to 26,081.72 and the Dow 316.56 points to 52,064.10, a fourth consecutive losing day. Rate rise odds for next week moved to about 70%, up from 62% the day before, per Investrade. The European Central Bank raised rates to their highest since 2025 on the same day.

Friday has not happened yet. US futures were marginally higher early on, with S&P 500 futures up about 0.19%, per CNBC. Everything else is waiting on the CPI print.

The week's verdict is uncomfortable. Nothing broke. There was no crash, no credit event, no earnings disaster. The market simply spent four days quietly marking down every asset that needs cheap money, because a supply shock in the Gulf is being converted, dollar by dollar, into a tighter Federal Reserve.

😶‍🌫️ Sentiment watch

Read more

A year ago the bull case for Oracle was a number: $638 billion of contracted future revenue, up 363% in twelve months, the largest order book any enterprise software company has ever disclosed. The stock has since fallen by more than half from its high, and S&P has cut the credit rating to one notch above junk. The backlog did not shrink. The market simply started asking who pays for the machines that fill it. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors still think of Oracle as a database company that sells cloud on the side. That framing is accurate and dangerously incomplete. In the year ended 31 May 2026, Oracle reported revenue of $67.357 billion, up 17%, of which cloud infrastructure alone was $18.101 billion, up 77%, per Oracle's Q4 FY2026 earnings release. In the same year it spent $55.663 billion on property and equipment, against operating cash flow of $31.977 billion, producing free cash flow of negative $23.686 billion. Read those two facts together and the company stops looking like software. Oracle now spends 83 paise of every rupee of revenue on building data centres, which is not a software business model, it is a utility construction model with a software company's balance sheet underneath it. Oracle reports first-quarter results after the US close tonight, and none of the figures in this article are those results.

🏭 What Oracle actually is today

Oracle sells four things that behave completely differently.

The first is the legacy database and licence support business, roughly $24.541 billion of software revenue in FY2026 and shrinking slightly at minus 1%, per the Q4 FY2026 release. This is the annuity. Tens of thousands of banks, telcos, airlines and government departments run mission-critical systems on Oracle Database and pay maintenance every year because migrating off is a multi-year project nobody wants to sponsor. It funds everything else.

The second is applications: Fusion ERP, NetSuite and the industry suites, grouped as cloud applications at $15.888 billion, up 11%. Steady, competitive, unspectacular, and the part of Oracle that looks most like Salesforce or SAP.

The third is Oracle Cloud Infrastructure, the business that changed the story. OCI was a distant fourth-place hyperscaler for most of the last decade. In FY2026 it produced $18.101 billion of revenue, up 77% for the year and 93% in the fourth quarter alone. The reason is specific rather than general: Oracle bet early on renting very large, very dense GPU clusters with high-bandwidth networking to a small number of AI labs, at a moment when Amazon, Microsoft and Google were rationing capacity to their own products and their own strategic partners.

The fourth is the piece that gets almost no attention in India and probably deserves the most. Oracle has spent two years putting its database physically inside its competitors' data centres, as Oracle Database@Azure, Database@Google Cloud and, since 2026, an AWS partnership. Multicloud database revenue grew several hundred per cent year-over-year through fiscal 2026 off a small base, per management commentary on the quarterly calls, and the strategic logic is cleaner than the growth rate. Every enterprise that wanted to move to AWS but could not move its Oracle database now has a route. Oracle stops being the thing you migrate away from and becomes the thing you carry with you.

Put those four together and Oracle is a shrinking annuity funding a hyperscaler build, with an applications business paying the light bill and a multicloud strategy quietly defending the annuity. Every one of those parts is real. Only one of them is why the stock moves.

🧩 Segment breakdown

Two things stand out. Cloud is now almost exactly half of Oracle, at $33.989 billion of $67.357 billion, which was not true two years ago. And the growth is extremely lopsided: infrastructure at 77% against applications at 11%. Oracle's revenue mix is being rewritten by one line item.

The residual line matters too. Oracle's headline release breaks out cloud and software but does not separately state hardware, services and other revenue in the same summary, so the roughly $8.8 billion balance shown here is calculated from the reported totals rather than printed by the company. Treat it as arithmetic, not disclosure.

⚡ Converting $638 billion: the core strategic bet

Remaining performance obligations, or RPO, is the accounting term for contracted revenue not yet recognised. It is the closest thing enterprise software has to an order book. Oracle ended fiscal 2026 with RPO of $638 billion, up 363% year-over-year and up $85 billion in a single quarter, per the Q4 FY2026 release.

For scale: that backlog is roughly nine and a half times Oracle's entire annual revenue. Nothing in enterprise software history compares. It exists because of one contract above all others, the reported five-year, roughly $300 billion agreement with OpenAI for compute starting in 2027. On those two figures alone, a single customer accounts for something close to 47% of the entire order book, and S&P flagged that concentration as a credit risk when it downgraded Oracle in July 2026.

Here is the mechanic that decides whether this ends well. RPO is not cash and it is not capacity. To convert a dollar of that backlog into a dollar of revenue, Oracle first has to build the data centre, buy or install the GPUs, energise the site and hand over working capacity. The spending happens years before the revenue does. That is why capital expenditure ran at $55.663 billion in a year when the AI cloud business generated $18.101 billion of revenue, and why management has guided to roughly $70 billion of net capital expenditure in fiscal 2027 against a $90 billion revenue target.

Read more

Nearly a third of what an Indian investor made on US stocks this year came from the rupee, not from the stocks. The Reserve Bank has just spent four months building a wall to stop that from happening again, and it raised 136 billion dollars doing it. Knowing which half of your return is the market and which half is the currency is the difference between reading the next year and being confused by it. That's why we built Winvesta Crisps: to break down what's actually moving markets, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

On 8 June the Reserve Bank of India opened a special dollar-rupee swap window and told banks it expected to raise 50 to 60 billion dollars. By mid-August the internal estimate had been lifted to 80 billion. When the window shut on 31 August, a full month ahead of schedule, it had pulled in 136.37 billion dollars. Foreign exchange reserves hit a record 740.80 billion dollars in the week to 28 August, a ninth consecutive weekly rise, and the rupee, which had touched an all-time low of 96.82 on 20 May, was trading near 94.50 this week. Most coverage has read that as a good news story about a central bank winning. For anyone holding US equities from India it is something more specific: the machinery that quietly added several percentage points to your rupee returns every year has just been switched off, and the reason it was switched off is that somebody borrowed 136 billion dollars to do it.

🏦 How the rupee got to 96 and what the RBI pulled off the shelf

The rupee opened 2026 in the 89.85 to 90.00 range. Nothing in India's own numbers explains what happened next. Growth held up, inflation stayed inside the target band, and the RBI's August policy meeting actually raised its FY2026/27 growth forecast to 6.7 per cent from 6.6 per cent and trimmed its inflation forecast to 5.0 per cent from 5.1 per cent.

What broke the rupee was oil. When the West Asia conflict escalated in late February and US and Israeli strikes on Iran effectively closed the Strait of Hormuz, Brent went from around 80 dollars to 120 dollars a barrel in under a week. India imports the overwhelming majority of the crude it burns, and every dollar on the barrel is dollars that have to be bought with rupees. The currency broke 90 in late February, crossed 95 for the first time in late March, and set fresh all-time lows in five of seven sessions in the third week of May, bottoming at 96.82 on 20 May before closing at 96.20 the next day once the RBI intervened. That is close to an 8 per cent fall in under five months, driven almost entirely by something happening 2,000 miles away.

Central banks have two ways to defend a currency in that position. They can sell reserves, which works until the reserves run low and the market notices. Or they can go and get more dollars. The RBI chose the second, using a tool it had used exactly once before.

In 2013, with the rupee under similar pressure, Governor Raghuram Rajan opened a swap window that raised 34 billion dollars, of which about 26 billion came through Foreign Currency Non-Resident Bank deposits on three-year tenors. Banks got to swap those dollars into rupees with the RBI at a concessional 3.5 per cent a year, roughly 3 percentage points below the market rate at the time. It stabilised the currency and it is remembered as a success.

The 2026 version is the same idea, built more generously. Rather than offer a discounted rate, the RBI agreed to absorb the hedging cost entirely on fresh three to five year FCNR(B) deposits booked in the window, which works out to about a 3 percentage point subsidy against prevailing swap rates of 2.8 to 3.3 per cent for those tenors. Eligible deposits were also exempted from cash reserve ratio and statutory liquidity ratio requirements, so banks did not have to park a slice of the money with the RBI, and each carries a one-year lock-in. FCNR(B) rates on offer to non-resident depositors went as high as about 7 per cent as a result.

The response was four times what the RBI first asked for. FCNR(B) deposits alone crossed 100 billion dollars to reach 127.2 billion. Overseas foreign currency borrowings added 5.26 billion and external commercial borrowings 3.891 billion. The RBI closed the deposit window on 31 August instead of 30 September, leaving banks until 11 September to complete the swaps themselves.

⚙️ Why 740 billion dollars of reserves is not 740 billion dollars

Work through what actually happens in one of these transactions, because the accounting is where the story is.

A non-resident Indian parks 100,000 dollars in a three-year FCNR(B) deposit at an Indian bank. The bank now holds dollars but does almost all of its lending in rupees, so it has a currency mismatch it has to cover. Normally covering that costs it 2.8 to 3.3 per cent a year, and that cost comes straight out of the interest rate it can offer the depositor. Under this window, the bank hands the dollars to the RBI, receives rupees, and the RBI commits to reverse the trade at a pre-agreed rate when the deposit matures. The bank carries no currency risk and pays no hedging cost, so it can offer the depositor a far better rate. The RBI ends up holding the dollars.

Three things follow, and they pull in different directions.

The reserves number goes up, which is the point. A larger stack makes betting against the rupee more expensive, and traders price that in before the RBI has to spend a cent.

Rupee liquidity floods the banking system, which was not the point. Every dollar taken in released rupees into the system. Core banking liquidity was already above 8 trillion rupees by 15 August and was expected to cross 10 trillion rupees in September. Prasanna Tantri of the Indian School of Business puts the foreign-currency-driven contribution at roughly 7.7 lakh crore rupees, close to the Covid-era peak of about 9.5 lakh crore. That much surplus cash sitting in the system is an inflation problem waiting one to two years out, and it constrains what the RBI can do on rates in the meantime.

The dollars are borrowed, not earned, which is the part the headline reserves figure hides. A swap is a promise to give the dollars back. The RBI's net short forward dollar position hit a record 137 billion dollars in July, up from about 104 billion in June, and most of that jump is these longer-duration swaps. So the record 740.80 billion dollars of gross reserves sits alongside a 137 billion dollar forward obligation running against it. Tantri estimates India's external debt could rise from around 765 billion dollars towards 900 billion as a result, and the repayments cluster from 2029, when the three-year deposits mature.

Rajan's window is the useful precedent on the way out. When the 2013 deposits matured between September and November 2016, the RBI had pre-covered roughly 80 per cent of the 26 billion dollars through forward contracts, reserves had been rebuilt to 365 billion dollars over three quiet years, the net redemption after rollovers came to about 20 billion dollars, and the RBI injected 10,000 crore rupees of open-market bond purchases to smooth the liquidity squeeze. It was managed calmly and nothing broke. The FCNR(B) piece this time is nearly five times larger, and whether 2029 is as quiet as 2016 is not something anyone gets to know in advance.

The dynamics covered in this article affect every US stock in your portfolio. Trade from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!

Get the app!

🏆 Winners and losers from a rupee that stops falling

Currency defence is not a neutral act. It moves money from one set of balance sheets to another, and the split is fairly predictable once you know who holds dollars and who owes them.

The pattern worth noticing is that the same policy sits on both sides of most Indian portfolios at once. An investor who owns Infosys and also owns an S&P 500 index fund has been paid twice by a falling rupee and now faces a mild headwind on both. That is not a reason to do anything dramatic. It is a reason to stop treating the two positions as independent bets.

🇮🇳 What this means for an Indian investor holding US equities

Every unhedged US stock position held from India is two positions. There is the stock, and there is a bet against the rupee that nobody consciously placed. Most of the time nobody notices, because the rupee's drift against the dollar is slow and in one direction. In a year like this one it stops being a footnote.

Run the numbers on 2026 so far. The S&P 500 is up about 12.3 per cent on price through 8 September, or 12.9 per cent including dividends. The rupee went from roughly 89.90 at the start of January to about 94.50 now, so a dollar buys 5.1 per cent more rupees than it did. Multiply the two and a rupee investor is up around 18 per cent on an index that gained 12.3 per cent.

Nearly a third of that return had nothing to do with American companies. It came from India's oil import bill and a closed shipping strait.

Now run it the other way, which is what the RBI has just done. Had the rupee stayed at its 20 May low of 96.82 rather than recovering to 94.50, the same holding would be worth about 10.87 lakh rupees instead of 10.61 lakh. That is roughly 26,000 rupees, or 2.9 percentage points of return, that the successful defence of the rupee removed from the portfolio of someone who was already invested. There is no villain in that sentence. It is arithmetic, and it runs in reverse for anyone still buying.

Which points at the split that actually matters. If you are still adding to a US portfolio, a stable-to-stronger rupee is straightforwardly good: every remittance buys more dollars of stock. Indians have been doing exactly that at pace, with outward remittances under the Liberalised Remittance Scheme reaching 2.5 billion dollars in June, up nearly 20 per cent year on year, of which 456.7 million dollars went into equity and debt, more than double the year before. The April to June quarter alone put 1.06 billion dollars into overseas equity and debt, against 28.98 billion dollars of total LRS outflows across all of FY26. If you are done accumulating and simply holding, the same policy trims your rupee returns and the drag compounds quietly.

The one thing worth resisting is treating this as a reason to hedge the currency. Retail rupee hedges over multi-year horizons cost more than the volatility they remove for most portfolios, and the RBI's own 137 billion dollar forward book is a reminder that the party on the other side of that trade is very well informed. The better response is knowing the number: what share of your US returns to date is stock picking and what share is the rupee.

🔭 What to watch

Five things would tell you whether the wall holds, and each one is checkable.

The weekly reserves release, published by the RBI every Friday. The number to separate out is organic accumulation from swap-driven accumulation. With the swap window shut from 11 September, a reserves figure that keeps climbing means real inflows. One that stalls means the 740.80 billion dollar record was the peak of a one-off operation.

The core liquidity print and what the RBI does about it. If surplus liquidity crosses 10 trillion rupees this month, expect variable rate reverse repo auctions or open market bond sales to drain it. Aggressive draining signals that the RBI is worried about the inflation consequences, which changes the rate path regardless of what growth does.

The 94 and 96 levels on USD/INR. Analyst forecasts compiled over the past month mostly cluster near 94 over the next three to six months with a drift back toward 96 in 2027, and a 93 to 96 range for the rest of this year. A clean break below 94 says the defence is working better than expected. A move back through 96 says 136 billion dollars bought a few months.

Brent crude and the Strait of Hormuz, because that is the actual cause and nothing else has changed about it. Traffic through the strait ran at 4.9 million barrels a day in the second quarter against roughly 21.6 million before the conflict, per the US Energy Information Administration. Reopening would do more for the rupee than any swap window.

The Federal Reserve on 16 September. Markets have been pricing meaningful odds of a rise, which would widen the dollar-rupee rate gap and pull capital back towards the United States. That is the first serious test of a wall built in calm conditions.

If this changed how you read the next reserves headline, pass it on.

Share

🏁 The bottom line

The RBI has done something defensible and expensive. Faced with an oil shock it could not control and a currency setting records in the wrong direction, it chose to borrow dollars from non-resident Indians on subsidised terms rather than burn reserves it had spent years accumulating. The rupee stopped falling, reserves hit a record, and the bill is dated 2029.

For an Indian investor holding US equities, the practical consequence is narrower than the policy debate. The currency tailwind that has flattered rupee returns on US portfolios for most of a decade is being actively fought by the institution best placed to fight it, and it has 136 billion dollars of ammunition plus a 740.80 billion dollar reserve stack behind that fight. None of that is an argument against owning US stocks, which you own for the earnings of companies India does not have. It is an argument for knowing that a good chunk of what you thought was investing skill in 2026 was the rupee, and for not budgeting for it again.

📊 By the numbers

• 136.37 billion dollars: total inflows through the RBI's special swap facility to 31 August, against an initial expectation of 50 to 60 billion, later revised to 80 billion

• 127.2 billion dollars: the FCNR(B) deposit share of that, with 5.26 billion from overseas foreign currency borrowings and 3.891 billion from external commercial borrowings

• 740.80 billion dollars: record foreign exchange reserves in the week to 28 August, up 11.47 billion on the week and up 49.70 billion since the end of March

• 137 billion dollars: the RBI's net short forward dollar position in July, a record, up from about 104 billion in June

• 96.82 and 94.50: the rupee's all-time low on 20 May and its level this week, from roughly 89.90 at the start of January

• About 7.7 lakh crore rupees: the foreign-currency-driven addition to surplus banking liquidity, per Prasanna Tantri of ISB, against a Covid-era peak near 9.5 lakh crore

• 765 to 900 billion dollars: India's external debt now, and where Tantri estimates it heads as a result of this operation

• 5.75 percentage points: the share of a rupee investor's roughly 18 per cent gain on the S&P 500 this year that came from the currency rather than the index

• 2029: when the three-year deposits raised in this window come due

Disclaimer: All content provided by Winvesta India Technologies Ltd. is for informational and educational purposes only and is not meant to represent trade or investment recommendations. Remember, your capital is at risk. Terms & Conditions apply.

Five years ago, an Indian investor could treat the oil price as somebody else's problem, a line in a newspaper about the current account. Not anymore. India's crude import bill for April to July came in at $63.4 billion, up about 57% on the same months last year, and Brent is trading near $95 after a week in which it rose more than 9%. Energy is the best performing sector in the S&P 500 this year by a distance, and almost none of the Indian investors reading this own a single share of it. That's why we built Winvesta Crisps, to decode what's actually moving the funds you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Two numbers from last week, sitting side by side.

US-listed energy ETFs took in roughly $796 million of net new money across 33 funds, per ETF Action's sector report published on 7 September. The State Street Energy Select Sector SPDR, ticker XLE, took $567 million of that on its own. In the same five days, information technology funds lost about $1.99 billion, with the iShares software fund shedding $545 million and the VanEck semiconductor fund $355 million. Energy gained 2.26% on the week and is up 45.29% for the year. Technology gained 1.08% on the week and money left anyway.

That is the rotation. It has been running most of 2026 and it is not subtle.

Now the second number, the one that should matter more to you than the first. India imported $63.4 billion of crude in the four months to July, against $40.5 billion in the same period a year earlier, per oil ministry data reported in August. Volumes barely moved. The price did. Domestic production fell to 9.1 million tonnes from 9.7 million, so a larger share of what India burns now has to be bought in dollars at Brent's price.

Most investors see an energy rally as a sector they missed. What it actually is, if you earn and spend in rupees, is a bill you have already been paying for eight months without owning the offsetting asset.

🎯 Meet Ishaan

Ishaan is 36, an engineering manager at a Hyderabad SaaS company, and has been investing in US markets from India since 2020. His portfolio is about ₹48 lakh. Around 70% sits in a broad S&P 500 fund, the rest split between two US technology names and a small position in an Indian energy fund he bought in March because oil was going up and that seemed like the obvious trade.

The obvious trade lost him money. That is where this article starts.

His S&P 500 fund has done its job, up roughly 13% this year. His Indian energy position has not. He assumed, reasonably, that an index with "energy" in the name goes up when crude goes up. It did not, and the reason is worth understanding properly, because it is the single most common mistake Indian investors make when they try to hedge the oil price from inside India.

His question this week is narrow. Brent is near $95, his petrol costs more, his grocery bill is up, the rupee is under pressure, and the one position he bought to offset all of that is red. What did he actually buy?

📊 What Indian energy exposure actually does

Start with what an energy index means in India, because it does not mean what it means in the United States.

Reliance Industries carried a weight of about 29.7% in the Nifty Energy index as of 1 September, per index composition data. Reliance today is a conglomerate with telecom and retail businesses that have very little to do with the price of a barrel. Most of the rest of India's listed energy complex is refining and marketing: Indian Oil, Bharat Petroleum, Hindustan Petroleum. Those companies buy crude and sell fuel at retail prices that do not move freely. When crude rises and pump prices do not, their margins compress. That is not a hedge against expensive oil. That is a leveraged bet against it.

The scoreboard for 2026 says exactly that. Through the calendar year, HPCL is down about 22%, BPCL about 21% and IOC about 17%, in a year when Brent rose roughly 45%. Analysts at Ambit have cut their integrated marketing margin assumptions for these companies to ₹3 to ₹5 per litre for FY27 to FY30, from ₹6 to ₹8 earlier.

Ishaan bought the word "energy". He got India's fuel subsidy mechanism.

The US energy sector is built the other way round. It is dominated by producers who sell the barrel rather than buy it.

XLE holds 23 large US energy companies, charges 0.08%, yields around 2.4% and manages roughly $39 billion. Its concentration is extreme by index standards: ExxonMobil is about 20.5% and Chevron about 15.0%, so more than a third of the fund sits in two companies, and the top ten holdings account for around 73%. The Vanguard fund, VDE, holds over a hundred names at 0.09% but ends up in nearly the same place, with Exxon and Chevron just under 40% between them. The SPDR exploration and production fund, XOP, is the different animal: roughly 50 equal weighted upstream companies, no integrated majors, 0.35%, and considerably more volatile in both directions.

Read more

Five years ago a FICO score cost a mortgage lender about 60 cents per pull and nobody in the business thought about it. In 2026 the same pull lists at $10, and last quarter that one price line drove a 49% jump in FICO's business-to-business scores revenue, per the company's own earnings release. Then a federal regulator wrote a few sentences on social media and took the pricing power away. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors file Fair Isaac under analytics software: a company that sells decisioning tools and happens to own a famous three digit number. The last four quarters say something else. Scores revenue rose 41% to $458.9 million in the June quarter while the software half of the business grew 2%, per FICO's earnings release, and almost all of the Scores growth came from charging more for the same mortgage pull rather than selling more of them. On Thursday evening, 3 September, the Federal Housing Finance Agency instructed Fannie Mae and Freddie Mac to accept rival VantageScore 4.0 from every mortgage originator, effective immediately. FICO closed Friday at $932.26, down 16.68% on the day and roughly 53% below its 52 week high of $1,998.01, per stockanalysis.com. The company is still highly profitable. The thing that made it exceptional was a rule.

🧩 What FICO actually sells

Fair Isaac runs two segments, and they could not be less alike.

Scores is a licensing business. FICO does not hold your credit data. Equifax, Experian and TransUnion hold the data. FICO owns the algorithm that turns that data into a number, and it collects a royalty every time a lender pulls a score. There is close to zero incremental cost in the transaction, which is why the company reported an operating margin of 53.8% for the June quarter, per its earnings release. Scores splits into business-to-business, meaning lenders, and business-to-consumer, meaning the myFICO subscriptions sold directly to individuals.

Software is the part that looks like a normal enterprise vendor. FICO Platform sells decisioning, fraud and originations tooling on a recurring contract, alongside a shrinking book of older on premise products.

The June quarter, fiscal Q3 2026 for FICO, ended 30 June and was reported on 29 July. Total revenue was $674.2 million, up 26% from $536.4 million a year earlier. Scores contributed $458.9 million of that, up 41%. Software contributed $215.3 million, up 2%. Inside Scores, business-to-business revenue rose 49% and consumer revenue rose 5%, all per the earnings release.

Read those two lines next to each other and the shape of the company becomes obvious. Two thirds of revenue comes from the licensing segment, that segment is growing at twenty times the rate of the software segment, and mortgage originations account for more than 60% of it, per FICO's disclosures and analyst summaries of the quarter.

The software story is genuinely improving underneath the flat headline. Platform annual recurring revenue grew 62% and carried a dollar based net retention rate of 148%, meaning existing platform customers spent about half again as much as they did a year before. The problem is the other half of the book: non-platform ARR fell 17% with retention of 82%. Net out the two and total software ARR grew 10%, per the earnings release. That is a decent software business being dragged by a legacy tail, and at $215.3 million a quarter it was never going to carry the valuation on its own.

📈 The growth engine was a price list

Read more

For two years, a weak jobs number was the best news a stock investor could get. It meant cuts were coming. This week the sign flipped: soft labour data rallied the market because it took a rate hike off the table, and the strong number is now the one to fear. That is why we built Winvesta Crisps, to cut through the noise and tell you what actually moved markets and why. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Last Friday, Kevin Warsh stood up at Jackson Hole and argued that the Fed should look at trends rather than "isolated data points", and that underlying inflation had not "meaningfully improved", per CNBC. Before that speech, the market gave roughly 70% odds that the Fed does nothing on 16 September. By this week, a quarter point hike was close to a coin flip.

That single repricing explains all four sessions. Stocks fell Monday and Tuesday as oil and bond yields did the hawkish work for him. Then two Fed officials spent Wednesday and Thursday quietly walking the hike back, and the market took both days with both hands.

Which leaves the whole week hanging on one number. The August jobs report lands at 8.30am New York time on Friday, which is 6pm in India, after this piece reaches you. Consensus is around 55,000 new jobs with unemployment ticking up to 4.2%, per MarketWise and TOPONE Markets. In a normal year you would want that number to be strong. This year, you want it soft.

📊 Market recap

The week ran in two halves: two sessions where the macro won, and two where the Fed talked it back.

Start with where things stand. Through Thursday's close, the S&P 500 sat at 7,747.71, the Nasdaq Composite at 26,584.06 and the Dow Jones Industrial Average at 53,686.11, per CNBC and market data. Measured against last Friday's close, that leaves the S&P in the mid-7,700s and up roughly half a per cent on the week, the Nasdaq in the mid-26,000s and up around 0.7%, and the Dow in the mid-53,000s and barely changed. A quiet weekly scoreboard hiding a very loud four days.

Monday reopened the war trade. The United States and Iran traded fire for the first time in a month, per CNBC, and equities gave up ground into the last session of August. The S&P 500 fell 0.33% to 7,686.14, the Nasdaq slipped 0.12% to 26,370.89, and the Dow dropped 374.09 points, or about 0.7%, to 53,185.90. Energy was the only S&P 500 sector to finish green. Note what still held: all three indices closed out August with a monthly gain regardless.

Tuesday was the bond market's day, and it was ugly. Fresh US strikes on Iranian targets near the Strait of Hormuz sent crude vertical. WTI jumped 5.2% to $90.22 a barrel, its first close above $90 in more than a month, and Brent rose 4.6% to $94.65, per Investrade and Yahoo Finance. Tehran announced a military response. A global bond sell-off ran alongside it, pushing the US 10-year yield to about 4.79% and the 30-year to roughly 5.27%. Equities had nowhere to hide: the S&P 500 fell 0.71% to 7,631.47, the Nasdaq lost 1.03% to 26,099.77, the Dow shed 0.79% to 52,766.88, and the Russell 2000 dropped 1.23%, with Nvidia and Amazon the heaviest weights. Higher oil feeds headline inflation, higher inflation feeds hike odds, and this week those stopped being separate stories.

Wednesday, a Fed official picked up a fire extinguisher. New York Fed president John Williams said there were "no clear signs" that a September hike was necessary, per market reports, and the three day slide stopped dead. The S&P 500 rose 0.46% to 7,666.60, the Dow added 295.07 points to 53,061.95, and the Nasdaq gained 0.45% to 26,217.83. Nine of eleven sectors finished higher and roughly 60% of the index advanced, so this was breadth rather than a two stock rescue. Nvidia rose 3.3%. The day's labour data helped in the new, inverted way: ADP showed private payrolls adding just 38,000 jobs in August against expectations near 47,000, the weakest reading since January. Two years ago that print sinks the market. This week it was a relief.

Thursday, another one did it again, harder. Fed governor Waller signalled he was willing to "sit and be patient" and said he expected inflation to moderate, per Investrade, and the market ran. The S&P 500 climbed 1.06% to 7,747.71, the Nasdaq added 1.40% to 26,584.06, and the Dow rose 624.16 points, or 1.18%, to 53,686.11, led by Goldman Sachs at about +3.4%. The 10-year eased to roughly 4.77%. The data was solid rather than scary: ISM services at 55.4 against a 54.2 consensus, jobless claims at 206,000, and a July trade deficit that widened 24.4% to $88.6 billion. Gold did something worth noticing, rising 2.84% to about $4,539.90 an ounce on the same day equities rallied hard. Not every buyer believes this is settled.

Friday has not happened yet. US futures and the entire week's verdict are sitting on the 8.30am payrolls print, which arrives after this article does. The S&P 500 is trading roughly 2% below its all time high with a Fed decision twelve days away.

😶‍🌫️ Sentiment watch

Read more

Four months ago the bull case for Broadcom rested on a $100 billion AI chip target. Management has since more than doubled it. On 2 September 2026 the company reported $16.7 billion of AI semiconductor revenue in a single quarter, up 221% year-over-year, and told investors to expect roughly $230 billion of AI revenue in fiscal 2028. The shares fell. That gap between the numbers and the reaction is the whole story. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors still think of Broadcom as a diversified chip and software conglomerate that happens to sell AI networking gear. That framing is accurate and dangerously incomplete. In the quarter ended 2 August 2026 and reported on 2 September, revenue was $29.591 billion, up 86% year-over-year from $15.952 billion, non-GAAP operating income was $20.1 billion or 68% of revenue, and free cash flow was $13.7 billion or 46% of revenue, per Broadcom's Q3 FY2026 earnings release. Guidance for the fourth quarter is roughly $34.8 billion, up 93%. None of those is the number that matters. The number that matters is $230 billion, which is what chief executive Hock Tan told analysts on the call to expect from AI semiconductors alone in fiscal 2028, and which rests on a customer list you can count on one hand.

🏭 What Broadcom actually is today

Broadcom reports two segments. Semiconductor Solutions was $20.839 billion in the quarter, 70% of revenue and up 127%. Infrastructure Software was $8.752 billion, 30% of revenue and up 29%, per the Q3 FY2026 release.

Inside the semiconductor half, one disclosed line carries almost all of the growth. AI semiconductor revenue was $16.7 billion, up 221% year-over-year and 54% sequentially, which is 56% of the entire company. It splits into two products. The first is the XPU, a custom accelerator that Broadcom co-designs with a single customer: the customer owns the architecture and decides what the chip is optimised for, while Broadcom supplies the high-speed interconnect blocks, the packaging, the physical design and the relationship with the foundry. The second is networking silicon, the Tomahawk switch and Jericho router families and the optical and PCIe components that move data between accelerators. On the Q3 call management said XPU shipments rose more than 3.5 times year-over-year and made up 73% of AI revenue, which is a meaningful shift: two years ago networking was the larger half.

The software half is VMware plus the mainframe and security assets Broadcom has accumulated, and it is run for cash rather than growth. Management put infrastructure software operating margin at 84% in the quarter, up about 650 basis points year-over-year, per commentary on the earnings call. It is the ballast. It pays the dividend and services the debt while the semiconductor side absorbs the volatility.

Everything else is a rounding error in the narrative, though not in the accounts. Subtract the disclosed AI figure from segment revenue and roughly $4.1 billion of non-AI semiconductor revenue is left: broadband, wireless components, enterprise storage and industrial. Broadcom does not report growth for that residual separately, so treat the $4.1 billion as an implied figure rather than a printed one. It has not grown meaningfully in two years, and nobody is buying the stock for it.

🧩 Segment breakdown

Two rows in that table carry the argument. AI semiconductors at $16.7 billion are now larger than the whole company was in the same quarter last year, when total revenue was $15.952 billion. A business that grew 221% year-over-year became the majority of a $1.75 trillion company in about eight quarters.

The second is the row the company does not print. Broadcom discloses AI semiconductor revenue as a single number and does not break it down by customer, by XPU programme, or between accelerators and networking beyond the percentage management chooses to mention on a call. For a business where three customers plausibly account for most of the total, that is the disclosure gap that should bother shareholders most. Everything an investor knows about concentration here comes from management commentary and analyst estimates, not from a reported segment line.

⚡ Custom silicon at gigawatt scale: the core strategic bet

The bet is that the largest AI workloads stop being general-purpose. A merchant GPU has to be good at training, inference, graphics and scientific computing, and the customer pays a gross margin in the mid seventies for that flexibility. Once a company is running one model architecture at enormous volume, the calculation changes: a chip designed for exactly that workload can deliver better performance per watt and per dollar, and the customer keeps the margin that would otherwise go to Nvidia. Broadcom sells the ability to build that chip.

What makes this quarter different from the last four is that the commitments have moved from slideware to contracted gigawatts. Google remains the anchor, with a long-term TPU agreement management has described as worth multiple tens of billions of dollars annually. Anthropic expanded its partnership with Google and Broadcom to cover what Anthropic itself describes as multiple gigawatts of next-generation TPU capacity coming online from 2027, with press reports citing Broadcom's disclosure putting the figure at about 3.5 gigawatts. OpenAI and Broadcom announced a 10 gigawatt programme of OpenAI-designed accelerators deploying from the second half of 2026 through the end of 2029, whose first chip, an inference processor called Jalapeño, is targeted for initial deployment by the end of this year. Meta's MTIA accelerator is expected to reach production shipments in the fourth quarter, per management commentary.

Read more

Most portfolios are still built for the rate cuts that were supposed to arrive this year. Futures markets now put roughly two-in-three odds on the opposite, a rise on 16 September, which would be the first since July 2023, at a moment when hiring is already slowing. Understanding why a central bank would tighten into a weakening economy is the difference between reading the next six weeks and being surprised by them. That's why we built Winvesta Crisps: to break down what's actually moving markets, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

On Friday 28 August, Kevin Warsh gave his first Jackson Hole keynote as Fed Chair and spent it on a single number: 2 per cent. He called headline PCE inflation at 3.7 per cent concerning, defended the target, and described delivering on it as the only true test of the central bank's credibility. Fed funds futures moved within the hour. Odds of a September increase went from about 35 per cent to 58 per cent that day and sat near 66 per cent by the end of the month, per the CME FedWatch tool. The two-year Treasury yield jumped to 4.35 per cent and the ten-year reached about 4.80 per cent on Tuesday, its highest since January 2025. What makes this unusual is not the hawkish speech. It is that the same week produced job openings that are drifting lower, a hiring rate that has been slowing all year, and a consensus forecast of 65,000 new jobs in August. A central bank is preparing to raise the cost of money into that.

🏛️ How a year of expected cuts became a coin flip on a hike

Start with where the policy rate actually is. The Fed spent 2024 and 2025 cutting, taking the target range down to 3.50 to 3.75 per cent, and it has now held there five meetings running, most recently on 29 July. Every forecast published in January assumed the next move was down. The question was how many cuts, not which direction.

Two things broke that assumption, and only one of them is about the American economy.

The first is oil. Disruption in the Strait of Hormuz that began in late February has proved far more durable than the first week of headlines suggested. Oil and petroleum liquids moving through the strait averaged 4.9 million barrels a day in the second quarter of 2026, against roughly 21.6 million barrels a day before the conflict, per the US Energy Information Administration. A 60-day ceasefire lapsed in mid-August and hostilities resumed. Brent peaked near 94.40 dollars a barrel on 21 August, has traded in an 86 to 91 dollar range since, and climbed back above 91 dollars on 1 September as US and Iranian strikes resumed. This is not a spike that is fading. It is now the better part of a year of elevated energy costs feeding through every supply chain that uses freight, plastics or power.

The second is what that has done to the inflation data the Fed actually watches. Headline PCE inflation rose 0.2 per cent in July to reach 3.7 per cent year on year, a tenth above the consensus forecast. Core PCE, which strips out food and energy, came in at 3.3 per cent, exactly where analysts expected. Headline CPI for July was 3.4 per cent, per the Bureau of Labor Statistics. Whichever gauge you prefer, inflation has now run above the 2 per cent target for more than five years.

Then came Warsh. Appointed by President Trump with rate cuts widely assumed to be the point of the appointment, he has instead spent his first months at the Fed rebuilding a reputation for hawkishness, and the Jackson Hole speech read as an argument that the target is not negotiable. Analysts covering the speech noted the obvious tension it creates with a Treasury that would prefer cheaper funding. Markets simply repriced.

⚙️ Why an oil shock splits a central bank down the middle

The textbook answer to a supply shock is to leave it alone, and the reasoning is worth spelling out because it is usually skipped.

Interest rates work on demand. Raising the cost of borrowing makes households buy less and firms invest less, which cools prices that are rising because too many buyers are chasing too few goods. An oil shock is the other kind of inflation. Prices are rising because a waterway is closed, and no interest rate setting anywhere in the world moves a tanker through the Strait of Hormuz. Tightening into it does not fix the cause. It simply adds a second contraction on top of the first, hitting employment to fight a price rise that monetary policy did not create and cannot reverse.

There is also an arithmetic reason to wait. A one-off jump in the price level shows up in the inflation rate for twelve months and then drops out of the comparison by itself. Do nothing and the number falls on its own next spring.

That is the case for holding, and it is a strong one. Here is why the Fed may override it anyway.

The argument turns on how long a shock has to last before it stops behaving like a one-off. Late February to early September is six months and counting, with the ceasefire already broken once. Firms that absorbed the first round of freight and energy costs have started passing them on. Workers who accepted one year of below-inflation pay rises negotiate differently in the second. At that point the shock stops being confined to energy and starts becoming the general price level, and the mechanism that carries it there is expectations.

This is the number to understand. University of Michigan survey respondents have recently put expected inflation over the coming year at about 4.8 per cent, against a five-year expectation closer to 3.2 per cent. Consumers expecting 4.8 per cent do not object to a 4.8 per cent price rise. Firms budgeting for 4.8 per cent wage demands price their products accordingly. If that continues long enough, inflation stops needing the oil shock to sustain it, and the central bank has to break the expectation rather than the shock, which historically costs far more in lost jobs than acting early does.

So a September increase would not be an attempt to lower petrol prices. It would be a costly signal, the point being precisely that it hurts. It says the 2 per cent number is real to the people who set it, and it is aimed at the household survey and the wage round rather than at the oil market.

The uncomfortable part is where the cost lands. Policy acts on demand, and demand is the half of this economy that is already soft. Job openings have hovered around 7.2 million in the most recent JOLTS readings, with hires at 5.1 million, quits at 3.1 million and layoffs at 1.7 million. That is a labour market where firms have stopped hiring aggressively but have not started firing, which is a delicate state to tighten into. Consensus for Friday's August payrolls is 65,000 jobs with the unemployment rate ticking up to 4.2 per cent.

History does not settle the argument, and anyone claiming it does is selling something. The European Central Bank raised rates in July 2008 with oil above 140 dollars, then cut by 175 basis points before the year was out. It raised twice more in 2011 into the euro area debt crisis, and members of its own governing council later described those moves as a mistake. That is the cautionary case, and it is the one bond markets remember. The case on the other side is the 1970s, when central banks treated successive oil shocks as temporary until expectations came loose, and the eventual correction required a recession far deeper than an early move would have caused. Warsh's speech was an argument that the second risk now outweighs the first.

The dynamics covered in this article affect every US stock in your portfolio. Trade from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!

Get the app!

🏆 Winners and losers if the Fed goes

The mechanism that matters for share prices is short and mechanical. A higher policy rate raises short-dated yields, and short-dated yields are the base rate used to value every future stream of profits. Raise it, and profits expected far in the future lose more value than profits expected next year. Whether a company is helped or hurt depends mostly on when its cash arrives.

The directions below are illustrative and directional, drawn from how these assets have already traded in the past week, not forecasts.

Two rows cut against instinct. Gold, the asset most investors reach for when they hear the word inflation, fell more than 1 per cent on Tuesday to around 4,375 dollars an ounce, its lowest since 19 August, because it pays no yield and a higher yield elsewhere is its main competition. And long-dated government bonds are not the obvious loser here. A credibility hike that convinces markets inflation will be contained can pull long yields down even as short yields rise, which is the whole point of the exercise.

The market response so far has been rotation rather than a broad sell-off. On the day of the speech technology fell 1.29 per cent, led by chip names, while consumer discretionary rose 1.69 per cent and energy held its ground. Eight of the eleven S&P 500 sectors are higher so far in 2026, with energy leading, up 43 per cent. The index is not pricing a disaster. It is repricing which businesses do well when money costs more.

🇮🇳 What this means for Indian investors holding US equities

Three separate channels reach an Indian portfolio, and they do not all point the same way.

The first is the currency. The rupee has moved from roughly 85 to the dollar through 2025 to around 95.2 in early September, with the RBI reference rate at 95.451 on 31 August. Rate rises in the United States widen the yield gap in the dollar's favour, which pushes the same way. For money already invested, a weaker rupee flatters the rupee value of US holdings, and that is genuinely helpful. For money yet to be sent, it makes every future dollar more expensive. Run it in plain numbers. A US holding that falls 8 per cent in dollar terms while the rupee weakens 3 per cent leaves you down about 5 per cent in rupee terms. The currency softens the blow. It does not remove it, and it makes topping up costlier at exactly the moment prices are lower.

The second is the RBI's own position. The repo rate has been on hold at 5.25 per cent with a neutral stance since June, and the central bank has been working hard to defend the currency. Its foreign exchange forward book reached a record 137 billion dollars in July, and it has offered to fully subsidise hedging costs on fresh three to five year FCNR(B) deposits until 30 September. A Fed that raises rather than cuts narrows the room for any Indian easing, because a wider gap in India's favour is what holds the rupee steady. Indian borrowers and rate-sensitive Indian sectors feel that indirectly.

The third is flows, and it is the one that has already happened. Foreign portfolio investors pulled roughly 2.2 lakh crore rupees, about 30.6 billion dollars, out of Indian equities by early June, the largest annual outflow since India opened to foreign portfolio investment in 1993, with some counts running higher since. Higher US yields, a stronger dollar and a widened current account deficit from the oil import bill are the reasons usually given. Indian retail money has been travelling in the opposite direction, with LRS remittances for equity and debt purchases reaching 456.7 million dollars in June alone, more than double a year earlier, per RBI data.

Put those together and the position is specific. Indian investors have been buying US equities, mostly large-cap technology, with rupees that are getting weaker, into a market where the discount rate is going up rather than down. None of that is an argument to stop. It is an argument to know what the position is. A portfolio concentrated in high-multiple US technology is a bet on the direction of US interest rates whether it was built that way or not, and this is the month that becomes visible. A systematic monthly investment keeps buying through a repricing rather than trying to time the meeting.

🔭 What to watch between now and 16 September

Three dates decide this, and all of them are public.

Friday 4 September brings August payrolls. Consensus is 65,000 with unemployment at 4.2 per cent. The headline number matters less than average hourly earnings, because wage growth is the channel through which an oil shock becomes general inflation. Soft jobs with firm wages is the combination that keeps a hike alive.

11 September brings August CPI. The gap between headline and core is the whole argument in one release. Headline staying high while core stays contained supports the case for looking through it. Core drifting up says the shock has spread and strengthens the case for acting.

15 and 16 September is the FOMC meeting. If a rise comes, read the statement's language on supply-driven inflation and the projections rather than the decision itself, since those say whether this is one insurance move or the start of something.

Two live indicators sit underneath those dates. The University of Michigan one-year expectation, near 4.8 per cent, is the series that justifies the hike, so a fall in it takes the pressure off faster than any speech will. And the two-year Treasury yield at 4.35 per cent is the cleanest read available on what the market has actually priced, which means the reaction on the day depends on where it sits going in, not on the decision alone.

The wildcard is the Strait of Hormuz. Flows at 4.9 million barrels a day against 21.6 million before the conflict is the fact underneath everything above. Genuine reopening would take the inflation impulse away within months and the entire case for tightening with it.

If this changed how you read the next Fed headline, pass it on.

Share

🏁 The bottom line

A rate rise on 16 September would not be an attempt to manage the economy in the usual sense. It would be an attempt to defend a number, the 2 per cent target, at a point where six months of energy disruption has started to show up in what households expect prices to do next year. That is a defensible decision and a genuinely risky one, and the ECB's 2008 and 2011 experience is the reason bond markets are nervous about it.

For an Indian investor the practical consequence is narrower than the debate. Money that costs more makes distant profits worth less today, and distant profits are what most US technology holdings are. A weaker rupee will cushion the rupee value of what is already held and raise the cost of adding to it. Neither of those is a reason to sell into a meeting. What they do argue for is knowing how much of a portfolio sits in one long-duration bet, and having that answer before Friday rather than after 16 September.

📊 By the numbers

• 3.50 to 3.75 per cent: the current Fed funds target range, held for five consecutive meetings, most recently on 29 July

• About 66 per cent: market-implied odds of a quarter-point September increase at the end of August, up from roughly 35 per cent before Warsh's 28 August speech, per CME FedWatch

• 3.7 per cent and 3.3 per cent: headline and core PCE inflation year on year in July, against a 2 per cent target

• About 4.8 per cent: University of Michigan one-year inflation expectations in recent surveys, against a five-year expectation near 3.2 per cent

• 65,000: consensus forecast for August non-farm payrolls due Friday, with unemployment expected at 4.2 per cent

• 4.35 per cent and about 4.80 per cent: the two-year and ten-year Treasury yields this week, the ten-year at its highest since January 2025

• 4.9 million barrels a day: oil and petroleum liquids through the Strait of Hormuz in the second quarter, against roughly 21.6 million before the conflict, per the EIA

• 95.2: the rupee against the dollar in early September, from roughly 85 through 2025, with the RBI's forward book at a record 137 billion dollars in July

• July 2023: the last time the Federal Reserve raised interest rates

Disclaimer: All content provided by Winvesta India Technologies Ltd. is for informational and educational purposes only and is not meant to represent trade or investment recommendations. Remember, your capital is at risk. Terms & Conditions apply.

Five years ago, owning the Nasdaq 100 was the entire strategy and a dividend fund was what you settled for when you ran out of conviction. Not anymore. In July, US-listed technology ETFs pulled in a record amount of money in the same month the average technology fund fell about 10%, and the dividend fund most Indian investors have never opened is beating the Nasdaq 100 by roughly eight percentage points this year. That's why we built Winvesta Crisps, to decode what's actually moving the funds you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

On 28 August, at Jackson Hole, Federal Reserve Chair Kevin Warsh said the 12-month change in the PCE price index stands at 3.7% and the six-month change at 4.1%. He described financial conditions as "not broadly restrictive" and set the bar plainly: the Fed must be confident inflation is moving to target clearly and at sufficient speed, "otherwise, we have work to do." Markets read that as what it was. Rate cuts are off the table and a hike is live, with futures pricing meaningful odds of an increase at the September meeting.

Now hold that against what investors actually did in July. US-listed ETFs took in about $189 billion, per State Street's monthly flows report, with technology funds alone taking a record near $19 billion. In the same month, the average technology sector and industry ETF fell roughly 10% and close to 90% of tech-classified funds finished negative. Money went in hardest exactly where prices went down. State Street's research head described it as investors treating the pullback as a buying opportunity rather than a warning.

Most investors see that as courage. The flows say something less flattering, which is that a lot of people bought a discount rate story with a momentum reflex.

🎯 Meet Sameer

Sameer is 34, works in product at a Pune software company, and has been investing in US markets from India since 2021. His portfolio is about ₹62 lakh. Roughly two thirds of it sits in a Nasdaq 100 fund, the rest in three technology names he bought individually because he already understood the businesses.

It has worked. That is the difficulty. Every correction he has held through has been rewarded within a quarter, so his instinct in July was the same instinct the flow data captured across the market. Tech fell, he added, and he felt sensible doing it.

His question this week is not whether the AI trade is real. It is narrower and harder. If the next Fed move is up rather than down, and if the fund he has never owned is quietly beating the fund he does own, is he holding a position or a habit?

📊 What the record flows actually say

Start with what Sameer owns. The Invesco QQQ Trust holds 100 companies, and three of them do most of the work: Nvidia at about 8.0%, Apple at about 7.5% and Microsoft at about 5.7%, so roughly 21% of the fund sits in three balance sheets. That is not a criticism of the fund. It is the fund's design, and it is the reason his returns have looked the way they have.

Set that against the two funds gathering the value money.

The Schwab US Dividend Equity ETF screens for dividend reliability and balance-sheet quality, charges 0.06%, and after its March 2026 reconstitution carries an index yield near 3.6%, with healthcare around 18% and energy around 12%. It holds no Nvidia, no Microsoft, no Apple and no Alphabet. The Avantis US Large Cap Value ETF is actively managed, charges 0.15%, and combines value with a profitability screen. Per mid-August reporting, the dividend fund is up about 27% this year against roughly 19% for the Nasdaq 100, and the Avantis fund returned about 22% on NAV in the seven months to 31 July.

Read more

Five years ago, valuing Salesforce meant counting seats and multiplying. That maths broke this year. In January the market decided AI agents would eat enterprise software, and roughly a trillion dollars of enterprise SaaS market value went with it, per market commentary at the time. Salesforce spent seven months as the poster child for that trade. Then it reported a quarter in which the fastest growing product is not billed by the user at all, it is billed by the unit of work completed. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Salesforce reported adjusted earnings of $5.90 per share for its fiscal second quarter on Wednesday 26 August, against a consensus near $3.27, and the shares closed 22.6% higher the following day, per market reporting. Read the headline and you would conclude the software business had a spectacular quarter. It did not. About $2.53 of that $5.90 came from gains on strategic investments, most of it a markup on the company's stake in Anthropic, per company disclosures and analysis at The Motley Fool. Strip that out and adjusted earnings were roughly $3.37 per share, up about 16%, on revenue of $11.345 billion that grew 11%. The market bought the stock anyway, and the reason sits three lines further down the release.

🧩 What Salesforce actually sells now

The company stopped reporting the way most investors still think about it. There is no line in this quarter called Sales Cloud or Service Cloud. Salesforce now splits its subscription revenue into two buckets, and the names it chose say more than the numbers.

The first bucket, which Salesforce calls Agentforce Apps, is the business everybody knows: sales, service, marketing, commerce and Slack. It produced roughly $7.2 billion in the quarter and grew about 8% in constant currency, per the company's Q2 fiscal 2027 disclosures. The second bucket, Data 360 together with the headless platform and other products, produced roughly $3.6 billion and grew about 20% in constant currency.

Two thirds of the revenue is compounding at single digits. One third is compounding at more than twice that rate. A company that renames its entire application suite after its agent product, then reports the data layer growing 20%, is telling you where it thinks the next decade of revenue comes from.

Some of that 20% is bought rather than built. Informatica, the data management business Salesforce acquired, contributed $456 million of revenue in the quarter, $440 million of it subscription, per the company. Back that out and the data and platform line still grew, but the gap between the two buckets narrows considerably. Anyone modelling this company needs to hold both facts at once: the mix shift is real, and the acquisition flatters it.

Total subscription and support revenue was $10.820 billion, up 12% as reported and about 11% in constant currency. Total revenue was $11.345 billion. By geography, the Americas contributed $7.404 billion, Europe $2.764 billion and Asia Pacific $1.168 billion.

Read more

Four earnings reports did more for this market than four days of macro news. Nvidia, Salesforce, CrowdStrike and Okta between them rewrote a week that had been drifting on Iran sanctions and a sticky inflation print, and they did it in a single session. That is why we built Winvesta Crisps, to cut through the noise and tell you what actually moved markets and why. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Going into Friday, the S&P 500 is up roughly 0.7% on the week, and almost all of it arrived in one session. On Thursday, technology was the only one of the S&P 500's sectors to finish higher, per TheStreet, while the index itself rose 0.72%. A week that spent three days worrying about Iran sanctions and a 3.3% inflation reading was bailed out by a chip company and three software firms.

And the week is not finished. Kevin Warsh gives his first Jackson Hole keynote as Fed chair at 10am New York time on Friday, after this piece lands in your inbox.

📊 Market recap

The week told a story in three acts: a geopolitical threat that landed soft, an inflation print that stuck, and an earnings night that outvoted both.

Start with where things stand. The S&P 500 closed Thursday at 7,730.99, up 0.72% on the day, per CNBC. The Nasdaq Composite added 1.57% to 26,541.35 and the Dow rose about 105 points, or 0.2%, to 53,569.44. Measured against last Friday's close, that leaves the S&P in the low-7,700s and up under a per cent for the week, the Nasdaq in the mid-26,000s and up closer to 1.4%, and the Dow in the mid-53,000s with a gain of about half a per cent.

Monday was a chip problem. Treasury Secretary Bessent used an afternoon press conference to detail expanded sanctions on Iran and on any country doing business with it, warning that such countries would be "removed from the US dollar system," per Reuters. Semiconductors took it worst. The Philadelphia Semiconductor index fell 2.64% to a three-week low, with Micron down 5.76%, AMD off 3.49%, Nvidia down 2.03% and Broadcom lower by roughly 2%, per Reuters and Yahoo Finance. The S&P 500 slipped about 0.3%. What is telling is the second-order reaction: oil fell more than 3%, because traders read the sanctions package as sparing Iran's actual trading partners, per Benzinga. The threat was maximal in language and modest in substance.

Tuesday brought relief in bonds and carnage in retail. Treasury yields fell for a second straight session as crude kept sliding, per CNBC, and the S&P 500 edged up 0.32% to 7,677.28 while the Dow logged a third consecutive advance. Underneath that calm, Dick's Sporting Goods collapsed. The retailer cut full-year adjusted earnings guidance to $11 to $12 per share from $13.50 to $14.50, blaming fewer and weaker athletic footwear launches and a reversal at Foot Locker, which it now owns. The stock fell roughly 25% on the session, its worst day on record, per Benzinga and Fox Business. The damage spread by association: Nike fell about 3.3%, with Lululemon and On Holding lower on no news of their own, per Reuters.

Wednesday was the waiting room. July core PCE, the Fed's preferred inflation gauge, rose 0.2% on the month and 3.3% on the year, per CNBC. Nobody could call that a relief. The S&P 500 finished all but unchanged at 7,675.70, the Nasdaq slipped 0.08% to 26,130.20, and the Dow lost about 114 points to 53,463.88. Meta provided the day's other headline, agreeing in principle to settle the child safety case brought by a coalition of state attorneys general. Reported figures vary across outlets, running from roughly $12 billion upfront to a total in the tens of billions once contingent payments are counted. The stock popped around 4% and then gave most of it back once an expected multi-billion dollar charge was in the frame. Everything else was positioning for one number, due after the bell.

Thursday delivered it. Nvidia reported second-quarter revenue of about $96.2 billion and adjusted earnings of $2.22 per share, ahead of consensus on both, and guided third-quarter revenue to roughly $108 billion, above the Street, per CNBC and Investing.com. Jensen Huang's framing, per prepared remarks, was that the AI infrastructure buildout is running at full steam. The stock rose about 8.7%. Then the software names landed on top of it: Okta up nearly 29%, CrowdStrike up 20.5% for its best day on record, Salesforce up 10.4% on a beat, raised guidance and an expanded partnership with Anthropic, per CNBC. Marvell and Synopsys came along for the ride.

That is the whole week's gain, concentrated in a handful of tickers on a single afternoon. The market did not resolve anything this week. It found something else to buy.

😶‍🌫️ Sentiment watch

Read more

Five years ago, analysing CrowdStrike meant counting endpoints and asking whether Microsoft would eventually bundle the product away for free. Not anymore. In the quarter it reported on 26 August 2026, CrowdStrike added a record $332.8 million of net new annual recurring revenue, up 51% year-over-year, and raised its full-year outlook for the second quarter running. Most of that growth is now sold through a single pooled contract that did not exist three years ago. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors still think of CrowdStrike as the endpoint security company that broke 8.5 million Windows machines in July 2024. That framing is accurate and dangerously incomplete. Total revenue in the quarter ended 31 July 2026 was $1.47 billion, up 26%, annual recurring revenue reached $5.84 billion, up 25%, and free cash flow hit a second-quarter record of $377.4 million, per CrowdStrike's Q2 FY2027 earnings release. The company also flipped the outage from a cost line to a recovery: the July 19 Incident and related matters produced a net credit of $14.5 million in the quarter, against a $35.7 million charge in the same quarter last year. The interesting number is none of those. It is $2.29 billion, and it sits inside a licensing construct called Falcon Flex.

🏭 What CrowdStrike actually is today

CrowdStrike sells one lightweight sensor that installs on a device and reports into one cloud platform. Everything else the company sells is a software module switched on against that same sensor and that same data: endpoint detection, identity protection, cloud workload security, exposure management, next-generation SIEM, and a managed detection service where CrowdStrike analysts do the hunting for you.

The financial statements barely reflect any of this. CrowdStrike reports exactly two revenue lines, subscription and professional services, and does not break out revenue or ARR by module. What it discloses instead is how many modules each customer runs. As of 31 July 2026, 51% of subscription customers ran six or more modules, 35% ran seven or more, and 26% ran eight or more, per the Q2 FY2027 release. Six months earlier those figures were 50%, 34% and 24%. The platform is widening steadily, one module at a time, and the module count is the closest thing shareholders get to a segment disclosure.

What has changed fastest is the shape of the product line rather than the sensor. Over the past year CrowdStrike has bought its way into the AI security stack. Pangea brought prompt-layer protection, which became Falcon AI Detection and Response. Onum brought telemetry pipeline management and Seraphic brought browser runtime security. The largest of them was SGNL, an identity company bought for about $740 million, which now ships as Continuous Identity for AI Agents. Goodwill on the balance sheet grew from $1.36 billion at the January year-end to $2.25 billion by 31 July, and the cash flow statement shows $881.4 million spent on business acquisitions in the first half, per the Q2 FY2027 release. In the same quarter the company agreed to acquire the technology assets of XM Cyber from Schwarz Digits.

Chief executive George Kurtz gave the strategy its own label in the results release, describing "the Mythos moment" as the point at which the market accepted that AI adoption needs security. His framing is that every enterprise will run on AI, and that securing it is the largest market opportunity in the company's history. Whether or not you buy the branding, the product roadmap has followed it exactly.

🧩 Segment breakdown

Two things in that table carry the argument. Subscription revenue is 95% of the total and grew 27%, while professional services grew 7%. This is a software business with a small consulting arm attached, not a services business, and the mix is getting purer.

The second is the row that says nothing. CrowdStrike does not report ARR by module, so nobody outside the company can tell you how much of that $5.84 billion is endpoint, how much is cloud, and how much is the identity and AI security work the acquisitions were meant to seed. Analysts model it; the company does not print it. That matters more than usual here, because the growth story has moved to products that are two years old at most.

⚡ Falcon Flex: the core strategic bet

Falcon Flex changes what a customer buys. Instead of purchasing a fixed number of seats of a fixed set of modules, the customer commits a pooled dollar amount across a multi-year term. They then draw that budget down across the whole portfolio, reallocating between endpoint, identity, cloud and SIEM as priorities move. There is no published rate card. Every Flex deal is negotiated.

The cohort has grown quickly. Ending ARR from accounts that have adopted Falcon Flex passed $2.29 billion as of 31 July 2026, up more than 101% year-over-year, per the Q2 FY2027 release. At the January year-end that figure was $1.69 billion, up more than 120%, per the Q4 FY2026 release. Against total ARR of $5.84 billion, Flex accounts now represent roughly 39% of the company's recurring revenue base.

Read more

Eighteen months ago the AI buildout was funded almost entirely out of retained earnings. This year roughly a third of it is being funded with borrowed money, and those bonds land in the same market that sets the discount rate on AI shares. Investors who own the trade now own both ends of it. That's why we built Winvesta Crisps: to break down what's actually moving markets, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

The AI trade has always been described as an equity story. Companies with enormous cash piles spend some of it on chips and buildings, the spending shows up as revenue at Nvidia, and the share prices go up. That description was accurate until about eighteen months ago. It is not accurate now. Amazon, Alphabet, Meta and Oracle have sold roughly 194 billion dollars of bonds so far this year, up about 79 per cent from around 108 billion dollars in all of 2025, per a Reuters analysis published on 29 July. The buildout has become one of the largest borrowers in the world, and the money it borrows comes from the same pool of long-term lenders who fund the US government. That is the part almost nobody has priced properly, because it runs the wrong way round: the debt that pays for the data centres helps set the interest rate that values the data centres.

🏗️ How the buildout stopped funding itself

For the first two years of the AI capital cycle the spending was genuinely self-financed. The Dallas Fed estimates that roughly 500 to 600 billion dollars of AI data centre investment since 2023 came out of retained earnings. Cash flow was so far ahead of capital spending that the debt question never came up.

The gap has closed. Incremental annual debt has risen from 9 per cent of capital spending in FY24 to 32 per cent in the twelve months to mid-2026, per FactSet, and aggregate debt across the five largest spenders now sits near 700 billion dollars. Moody's, which put out a note in July warning that the pace threatens credit quality at the six companies it tracks, expects capital spending of about 785 billion dollars this year and close to a trillion next year. Meta's trailing twelve month free cash flow swung to a 7.6 billion dollar outflow in the second quarter, its first negative reading since 2023. Amazon sold about 54 billion dollars of bonds in March. Nvidia, of all companies, sold 25 billion dollars in June. Alphabet went to the equity market instead and raised 84.75 billion dollars, the largest listed corporate equity transaction on record.

Bonds are only the visible part. Private credit funds have gone from near zero to over 200 billion dollars of outstanding loans to AI related companies, mostly originated by a handful of managers. Moody's counts 1.2 trillion dollars of lease commitments across the group, of which more than 820 billion dollars relates to facilities that have not started operating yet. QTS Realty raised 3.9 billion dollars of investment grade paper for a single Georgia data centre at a yield of about 7.23 per cent, which is an investment grade rating at a high yield price.

Add it up and AI financing is now close to a quarter of all gross investment grade bond issuance in the United States. August set a monthly record at 145.2 billion dollars of high grade supply, beating the 136 billion dollars sold in the same month of 2020. Issuance from January to mid-August reached 1.68 trillion dollars, about 27 per cent ahead of the same period last year.

⚙️ How a data centre bond reaches your share price

The chain has four links and each is arithmetic rather than sentiment.

The first link is duration supply. A twenty or thirty year corporate bond is not just a claim on a company. It is a block of interest rate risk that somebody has to hold, and the pool of investors willing to hold thirty years of it is finite. Pension funds, insurers and long bond funds have a fixed appetite. When new long-dated paper arrives in size, those buyers do not absorb it at yesterday's price. They ask for more yield. The Dallas Fed paper by Hugo De Vere, Srini Ramaswamy and Seth Searls, published in February, names this directly: AI issuance is now a material source of duration supply, and it partly substitutes for other investment grade borrowers.

The second link is who else is queueing. Total US federal debt passed 40.05 trillion dollars on Tuesday, and the deficit is running above last year's. The Treasury has to sell duration into the same buyers, at the same time, in an environment where inflation has been above the Fed's 2 per cent target for more than five years. Neither borrower is bidding against the other in a formal auction, but they are asking the same investors the same question, and the answer has been getting more expensive. The 30-year Treasury yield touched 5.33 per cent on 18 August, its highest level since 2007, against a Fed funds target range of 3.50 to 3.75 per cent. That gap of more than a percentage point and a half is not a forecast of rate rises. It is the term premium: what lenders charge for the uncertainty of committing money for three decades.

The third link is the one most equity investors have never had spelled out. A share price is the value today of profits a company will earn in the future, and the long government bond yield is the base rate used to discount them. Raise that rate and every future dollar of profit is worth less today. The effect scales with how far out the profits sit. A dollar expected in ten years loses roughly 9 per cent of its present value when the discount rate goes from 4.3 to 5.3 per cent. Push the same dollar out to thirty years and it loses about 25 per cent. Nothing about the company has to change.

The fourth link closes the loop. The companies whose value sits furthest in the future are exactly the ones issuing the debt. An AI data centre is a bet that revenue arrives in the 2030s, funded by bonds that mature in the 2050s, sold to investors who set the discount rate applied to that same 2030s revenue. Technology alone was 37.2 per cent of the S&P 500 as at 14 August, per S&P Dow Jones Indices, so this is not a corner of the index. It is most of it.

One honest caveat, because it matters. The long end is being pushed around by several forces at once, and AI issuance is a contributor rather than the whole story. The deficit is doing work. So is an oil supply shock that has kept headline inflation elevated. And credit spreads for companies with nothing to do with AI have barely moved, remaining close to historic lows, which tells you the crowding out is showing up in the government curve and in AI borrowers' own pricing rather than across the whole corporate market. The chain is real. It is one of three or four hands on the same lever.

The dynamics covered in this article affect every US stock in your portfolio. Trade from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!

Get the app!

🏆 Winners and losers when AI borrows

The market has already started discriminating, and it is doing it inside AI credit rather than across the market. Directions below are illustrative and directional, not forecasts.

The clearest evidence that lenders are pushing back is in the terms, not the headlines. Cover ratios on hyperscaler bond sales, which measure investor orders against the amount on offer, fell from nearly five times in February to below two times in July, per Apollo Global Management. Amazon's March dollar deal was 3.4 times oversubscribed and its July deal only 1.6 times. The new issue concession, the extra yield a borrower pays to get a deal away, went from 2.25 basis points last year to 12 basis points this year. Of 91 hyperscaler bonds issued in 2026 with comparable pricing, 78 were trading at higher yields on 28 July than at issue, a median 22 basis points worse. Oracle's five year credit default swaps have pushed past 110 basis points, the widest in three years.

Note the two rows that cut against instinct. Nvidia and the chip supply chain still benefit from all of this, because debt-funded capital spending is revenue whoever writes the cheque, and that is precisely why the revenue quality question keeps being asked. And long-dated government bond funds, the thing many investors hold as protection against an equity fall, are being hurt by the same force at the same time.

🇮🇳 What this means for Indian investors holding US equities

Indian money has been moving into US equities at pace. Remittances under the Liberalised Remittance Scheme for the purchase of equity and debt reached 456.7 million dollars in June alone, more than double a year earlier, against total LRS outflows of 28.98 billion dollars in FY 2025-26, per RBI data. Almost all of that equity money goes into the same place: large-cap US technology, either directly or through an index that is 37 per cent technology by weight.

That leaves an Indian portfolio holding one half of a two-sided trade. The AI buildout is now financed with a stack of debt and equity, and the debt sits senior. Bondholders and private credit lenders get paid before shareholders, they have covenants, and they have been repricing their risk in public since February through cover ratios and concessions. Equity holders have the residual claim and no such mechanism. If the depreciation schedules on this hardware turn out to be shorter than assumed, or the lease commitments that have not started yet begin on worse terms, the loss lands on the junior claim first. The credit market is telling you what it thinks of the risk. Indian retail investors are almost entirely on the other side of that conversation, and the pricing signal sits in an asset class they cannot easily buy.

The currency layer compounds rather than cancels. The rupee has traded in the mid 95s to the dollar this week and is roughly 10 per cent weaker over twelve months. A weaker rupee flatters the rupee value of existing US holdings, which is the good news. It also raises the cost of every subsequent dollar remitted under LRS to add to those holdings, and higher US long yields tend to support the dollar, so the two forces in this article push the same way on the exchange rate.

The practical response is not to trade the bond market. It is to know that a portfolio built on high-multiple US technology is a leveraged position in long-term US interest rates whether it was constructed that way or not, and this is the year that becomes visible. If the entire US allocation is in the AI complex, that is a concentration decision rather than a diversification one. A systematic monthly investment into broad index exposure keeps buying through a repricing instead of trying to guess where it ends.

🔭 What to watch from here

The 30-year Treasury yield is the transmission channel, not the Fed funds rate. Watch the gap between the two. A term premium that keeps widening while the policy rate sits still is the signal that supply, not the Fed, is setting the price of long money.

Treasury buyback operations start their expanded run on 9 September. Scott Bessent's department is raising the maximum size of liquidity support buybacks for longer-dated nominal coupons from 2 billion to at least 4 billion dollars per operation in the 10 to 20 year and 20 to 30 year buckets, running through 4 November. The first attempt at this on 19 August pulled the 30-year down about 9 basis points to 5.196 per cent, and the move was gone within two sessions, with the yield back at 5.234 per cent on Thursday 20 August and 5.273 per cent on Friday 21 August. Whether the larger operations hold better is the cleanest live test of whether debt management can offset supply.

Cover ratios and new issue concessions on the next round of hyperscaler deals will tell you more than any analyst note. Lenders commit money rather than opinions. A concession drifting back towards last year's 2 basis points means appetite has returned. Wider means the buildout is getting more expensive to fund at exactly the moment it is getting bigger.

Capital spending guidance is the other side of the same question. Goldman Sachs expects the five hyperscalers to issue roughly 250 billion dollars of bonds this year and 400 billion dollars next year. If third quarter results raise capex again without raising the cash flow that covers it, the borrowing forecast goes up with it.

Two policy dates frame the autumn. Kevin Warsh gives his first Jackson Hole keynote as Fed Chair on Friday, three weeks before the 15 to 16 September FOMC meeting, and the July decision was a 9-3 hold with three regional presidents dissenting over inflation. How the statement treats supply-driven inflation matters more for the long end than the decision itself.

If this changed how you read an AI headline, pass it on.

Share

🏁 The bottom line

The AI buildout has quietly changed from a cash story into a credit story, and that changes what owning it means. When capital spending came out of retained earnings, the only question was whether the projects would earn a return. Now that a third of it comes from borrowed money, a second question sits on top: what does the market charge to lend for thirty years, and what does that rate do to the value of profits promised for the 2030s. The same investors answer both questions, and their answer has been getting more expensive since February. Indian investors buying US technology are buying the junior claim on that trade while the senior claim reprices in public. Watching the 30-year yield is not a bond market hobby. For a portfolio concentrated in long-duration technology, it is the most direct read available on what those shares are worth.

📊 By the numbers

• 194 billion dollars: bonds sold by Amazon, Alphabet, Meta and Oracle through 7 July 2026, up about 79 per cent from roughly 108 billion dollars in all of 2025, per Reuters

• 9 per cent to 32 per cent: incremental annual debt as a share of hyperscaler capital spending, FY24 against the twelve months to mid-2026, per FactSet

• 785 billion dollars: Moody's estimate of 2026 capital spending across the six companies it tracks, heading towards about a trillion dollars in 2027

• About a quarter: AI financing's share of gross US investment grade bond issuance, against record August supply of 145.2 billion dollars

• Near 5 times to below 2 times: cover ratios on hyperscaler bond sales, February against July 2026, per Apollo Global Management

• 2.25 to 12 basis points: the new issue concession paid by hyperscaler borrowers, 2025 against 2026

• 5.33 per cent: the US 30-year Treasury yield on 18 August, its highest since 2007, against a Fed funds target range of 3.50 to 3.75 per cent

• 40.05 trillion dollars: total US federal debt as at Tuesday 25 August

• 1.2 trillion dollars: hyperscaler lease commitments counted by Moody's, of which more than 820 billion dollars has not started yet

• 456.7 million dollars: Indian LRS remittances for equity and debt purchases in June 2026, more than double a year earlier, per RBI

Disclaimer: All content provided by Winvesta India Technologies Ltd. is for informational and educational purposes only and is not meant to represent trade or investment recommendations. Remember, your capital is at risk. Terms & Conditions apply.

Five years ago, buying a biotech fund meant accepting that you would never know what was inside it and hoping the sector drifted up. Not anymore. On 19 August a single holding rose about 144% in one session on the first successful Phase 3 trial of a personalised mRNA cancer vaccine, and most investors holding biotech through a fund would have struggled to see it in their portfolio value. How a fund is weighted decides how much of a result like that actually reaches you. That's why we built Winvesta Crisps, to decode what's actually moving the funds you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Moderna and Merck announced on 19 August that their personalised mRNA cancer vaccine, given alongside Merck's Keytruda, met its primary endpoint in a Phase 3 trial of 1,137 patients with surgically removed high-risk melanoma, per STAT and CNBC. It is the first positive Phase 3 result for an individualised neoantigen therapy and for any mRNA-based cancer treatment. Moderna rose about 144% on the day and Merck about 10%. The obvious next thought, for anyone who does not want to own a single clinical-stage company, is to buy the sector through a fund. That thought runs into arithmetic almost immediately, because the two largest biotech funds available to Indian investors are built on opposite principles and would have handed you very different amounts of that day.

🎯 Meet Devika

Devika is 34, works in product at a Bengaluru software company, and has been investing in US equities from India for about four years. Her portfolio is roughly ₹9 lakh: a broad US index fund, three large technology positions, and cash she has not deployed since the market got choppy in August.

She read about the Moderna result on Wednesday evening, and her reaction was the correct one. She does not want to own one biotech company whose value depends on a regulator's decision. She does want exposure to a field that just produced its first genuine proof point after a decade of promises. So she looked up biotech ETFs, found two obvious candidates, and stopped, because their one-year returns were 40 percentage points apart and she could not work out why.

Her question is the one this article answers: given what just happened, and given that long-term US interest rates are at their highest in nearly two decades, should she buy biotech, and if so, through which fund?

📊 What the two funds actually give you

The two funds are the SPDR S&P Biotech ETF (XBI) and the iShares Biotechnology ETF (IBB). They cover the same industry and are constructed in opposite ways.

That table explains the 40-point return gap without any reference to stock picking. IBB is market-cap weighted, so its top ten positions are about 58% of the fund and it behaves like a portfolio of large, commercial-stage biotechnology companies with approved products and revenue. XBI uses a modified equal weight, so its top ten are only about 11% of the fund and roughly 80% of its holdings are mid and small caps, which in this industry means companies whose value rests on trials that have not finished.

Now Devika's arithmetic. In a modified equal-weight fund, a typical position is around 1% of assets. A holding that rises 144% in a day on a roughly 1% weight adds something in the order of 1.5 percentage points to the fund. In a cap-weighted fund, unless the company happens to be among the largest constituents, the same result adds meaningfully less. That is the trade in one line: a basket converts a spectacular single-stock event into a modest fund move, in exchange for making sure that no single failed trial can take a large bite out of your capital.

Read more

Five years ago, owning Nvidia meant taking a view on how many graphics cards gamers would buy. Not anymore. The company now sells complete AI data centre systems, and in its most recent quarter the networking business alone grew 199% year-over-year. The part nobody talks about is what the market has done with that: the shares are up about 20% over the past year while revenue grew 85%, and the price-to-earnings multiple has fallen to roughly a third below its own five-year average. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors still argue about whether Nvidia is a bubble. The market stopped agreeing with them some time ago. In the quarter ended 26 April 2026, Nvidia reported revenue of $81.6 billion, up 85% year-over-year, with data centre revenue of $75.2 billion, up 92%, per the company's Q1 fiscal 2027 results. Over the twelve months to 19 August 2026 the shares rose about 19.5%, to $217.56. A business growing at 85% whose shares grow at 20% is being derated, and the numbers say so plainly: the trailing multiple is about 33 times against a five-year average near 62, and the forward multiple of roughly 22 sits below the semiconductor industry median of about 30, per market data. On Wednesday 26 August the company reports again. The question is no longer whether Nvidia can beat. It is what a beat is still worth.

🧩 What Nvidia actually sells now

The gaming company is gone. In the last reported quarter, data centre was 92% of revenue.

Two rows in that table carry the argument. Compute, which is the accelerators themselves, was $60.4 billion and grew 77%. Networking was $14.8 billion and grew 199%, more than doubling its share of the company in a single year. Everything else, which Nvidia now groups as edge computing, was $6.4 billion.

Networking is the line most investors under-read. It covers the switching, interconnect and optics that tie thousands of accelerators into a single machine, and its growth rate is roughly two and a half times the growth of the chips themselves. That is what happens when the unit of sale stops being a chip and becomes a data centre. A customer buying accelerators can shop around. A customer buying a rack-scale system with Nvidia's fabric inside it has bought an architecture, and architectures do not get swapped out at the next procurement cycle.

The margin profile tells you how much pricing power sits behind that. GAAP gross margin was 74.9% in the quarter, on $81.6 billion of revenue. Semiconductor companies do not earn 75% gross margins on hardware unless the customer has no alternative that works.

⚙️ The growth engine, and the thing it depends on

Read more

Watching US markets week to week without context is just noise. This week the noise had a single source: the long end of the US Treasury market. The 30-year yield touched its highest level since 2007, and the part of the equity market that broke was not the part most people expected. Memory and storage stocks, the most extended trade in semiconductors, fell hardest, while a biotech readout produced the largest single-day gain in a major index constituent this year. That's why we built Winvesta Crisps, to cut through the noise and tell you what actually moved markets and why. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Three sessions into the week the market had already made its point. Long-dated US government bonds sold off hard enough to drag the S&P 500 down for two straight days, and the stocks that fell most were not the expensive software names everyone worries about but the memory and storage makers that had been the best performers in semiconductors all year. Then on Wednesday the Treasury announced a larger buyback of longer-dated debt, yields eased, and the index closed higher. The week's lesson is that the bond market is currently setting the price of equities, and it is doing so day by day.

📊 Market recap

A note on the window. This recap covers Monday 17 to Wednesday 19 August, the last three completed sessions before it went to press. Thursday's session, which included Walmart's second quarter results, and Friday's trading fall outside it.

Monday. The week opened with the bond market in charge. The 30-year Treasury yield rose nearly six basis points to about 5.31%, passing the previous month's high to reach its loftiest level since 2007, per Bloomberg. The move was not about the Federal Reserve. It reflected a July federal deficit that was the largest monthly total since March 2021, a heavy calendar of long-dated bond sales, and inflation that has now sat above the Fed's 2% target for five years while the Strait of Hormuz standoff keeps oil elevated. The same pattern showed up abroad, with Canadian 30-year yields at their highest since 2010 and German rates back at 2011 levels. Equities drifted lower, with elevated oil prices and global yields the stated drag, per CNBC.

Tuesday. The pressure became a proper risk-off session. The 30-year topped 5.33%, a fresh 19-year high, per CNBC. The S&P 500 fell about 0.6% and the Nasdaq about 1.3%. The selling concentrated in semiconductors, and inside semiconductors it concentrated in memory and storage: Micron fell about 5%, Western Digital about 7%, SanDisk about 6% and Seagate about 6%, per 24/7 Wall St. The uniformity across the group is the tell. No single company reported anything. The entire complex was repriced at once, which is what a discount rate reset looks like rather than a company problem.

Wednesday. The Treasury Department announced an increased buyback operation for longer-term debt, yields fell back, and the S&P 500 ended a three-day losing run, per CNBC. The index rose about 0.21% to around 7,708, the Dow about 0.22% to around 53,463, and the Nasdaq about 0.16% to around 26,331. The gains were modest and the composition was odd: the day belonged to a melanoma trial result and a takeover report rather than to anything macro.

😶‍🌫️ Sentiment watch

CNN's Fear and Greed Index has been sitting in the low-to-mid 50s, having slipped out of greed and into neutral territory during the week as bond yields climbed, per readings reported by Benzinga. Sources differ by a few points on the exact level, so treat it as a zone rather than a number.

The VIX closed Wednesday at 14.89, down about 6% on the day. That is a striking reading for a week in which the long bond hit a 19-year high and the Nasdaq had a 1.3% down day. Equity volatility is not pricing much stress at all. Either the bond market is wrong about inflation, or the options market has not yet decided that a 5.33% 30-year yield is an equity problem. Both cannot hold.

The stocks moving markets this week are all tradeable from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!

Get the app!

🏆 Best performers

Read more

Five years ago, analysing Target meant counting store openings and asking whether the home and apparel aisles would clear without heavy markdowns. Not anymore. In the quarter it reported on 19 August 2026, the three smallest lines on Target's income statement, advertising, membership fees and marketplace commissions, grew four times faster than the merchandise on its shelves, while a $994 million cheque from US Customs doubled reported profit. Only one of those two things happens again next year. That's why we built Winvesta Crisps, to decode what's actually driving the companies you own, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors still think of Target as a discount chain fighting Walmart for the same weekly shopping trip. That framing is accurate and dangerously incomplete. Second quarter net sales were $26.5 billion, up 5.3%, and diluted earnings per share came in at $4.11 against $2.05 a year ago, per Target's Q2 2026 earnings release. Almost exactly half that earnings jump was a refund of tariffs the US Supreme Court ruled the government never had the authority to collect. Strip it out and earnings per share still rose about 20%, which is the number worth studying. Underneath it, merchandise sales grew 5.0% and non-merchandise sales grew 20.1%, and the second figure carries margin the first one has never had.

🏭 What Target actually is today

Target runs 2,019 stores across 253.8 million square feet of retail space with more than 400,000 staff, per the Q2 2026 earnings release. Roughly half the merchandise it sells is sourced outside the United States, and Target is the importer of record for most of its owned and exclusive brands, which is why a customs ruling landed in its profit and loss account rather than a supplier's.

The merchandise business is organised into six categories, and in the second quarter every one of them grew. Food and beverage is the largest at $5,991 million, household essentials next at $4,617 million, then apparel and accessories, hardlines (which Target now calls Fun 101), home furnishings and décor, and beauty. The spread between them is the whole turnaround story: hardlines grew 10.6% and beauty 7.2%, while apparel grew 0.1% and home furnishings 0.2%.

Sitting beside that is a business Target does not give its own segment. Advertising revenue, credit card profit sharing and other non-merchandise income together made $592 million of net sales in the quarter, up from $492 million. That looks trivial against $26.5 billion until you read the footnote attached to it. Target discloses that revenue from Roundel, its retail media arm, is classified as net sales, or as a reduction of cost of sales, or as a reduction of SG&A expenses, depending on how each advertising arrangement is structured. The $279 million of advertising revenue on the sales line is a fraction of what Roundel actually generates. Target has said publicly that Roundel produces more than $2 billion of value for the company, and that it aims to double that between 2025 and 2030.

That is the unusual thing about this business. The fastest-growing, highest-margin part of Target is deliberately scattered across three lines of the income statement, so no single number on the page tells you how big it is.

🧩 Segment breakdown

Two things in that table carry the argument. Food and beverage plus household essentials are now 40% of net sales, which is a grocer's mix, and they grew at 7.2% and 4.4%. This is what a traffic-led recovery looks like: comparable transactions rose 3.6% while the average basket rose 0.2%, per the earnings release. People came back more often. They did not spend more per visit.

The other is the bottom of the table. Advertising revenue grew 28.6% year-over-year, other non-merchandise income 23.4%. Those two lines together are 1.8% of net sales and contributed roughly 8% of the entire year-over-year increase in the top line. Their importance has nothing to do with size and everything to do with what they cost to produce.

⚡ The business that does not sit on a shelf

Target's core strategic bet is that it can grow profits faster than sales by selling things that never move through a warehouse.

There are three of them. Roundel sells advertising against Target's own shopping data, working with more than 2,000 vendors and drawing on a Target Circle loyalty base the company puts above 100 million members. Target Circle 360 is the paid membership tier, sold on same-day delivery. Target+ is a curated third-party marketplace where Target takes a commission on somebody else's inventory, somebody else's warehouse and somebody else's working capital.

Read more

Five trading days ago the S&P 500 closed at a record. Since then Brent crude has risen for four straight sessions, the 30-year US Treasury yield has hit its highest level since 2007, and the Nasdaq has fallen three days running. Nothing changed at Nvidia or Microsoft in those five days. What changed was the traffic through one shipping lane in the Gulf. That's why we built Winvesta Crisps: to break down what's actually moving markets, in plain language, before the consensus catches up. 60,000+ investors from all over India are already in. What about you?

Subscribe now!

🔔 Don't miss out!

Add winvestacrisps@substack.com to your email list so our updates never land in spam.

Most investors think of the oil price as an energy story. It shows up in the pump price, it helps oil producers, it hurts airlines, and if you own neither it is somebody else's problem. That framing has stopped working. Brent traded near 92 dollars on Wednesday, up about 37 per cent over twelve months per Trading Economics, and the position it damaged most in the past week was not an airline. It was the part of your portfolio you probably think of as the safest long-term bet: large-cap US technology. The route from a blocked shipping lane to a compressed price-to-earnings multiple runs through the bond market, and it is shorter than most people realise.

🛢️ What is actually happening in the Gulf

The Strait of Hormuz is the highest-volume oil chokepoint on the planet. In the first half of 2025, roughly 20.9 million barrels a day moved through it, about 20 per cent of global petroleum liquids consumption and around a quarter of all seaborne oil trade, per the US Energy Information Administration.

That flow has been broken since late February 2026. During the first quarter of this year, crude and petroleum liquids through the strait fell almost 30 per cent year on year to 14.6 million barrels a day, per EIA data. The position now is worse than the quarterly average suggests. Commercial transits have fallen to a handful a day against the hundred-plus that used to pass before the conflict, and roughly 3,200 vessels, including around 800 tankers and cargo carriers, are idling west of the waterway waiting for it to clear.

The insurance market tells the same story in money. War risk cover for a transit ran at about 0.25 per cent of hull value before the war. It has since been quoted in the range of 3 to 10 per cent, which turns a 100 million dollar tanker's premium from roughly 250,000 dollars into several million per trip. At that price, a shipowner does not need the strait to be formally closed to stop using it.

The immediate trigger for this week's move was diplomatic. The memorandum of understanding signed in June, intended to give both sides 60 days to negotiate, expired on Monday, and President Trump said he was not interested in extending it. Iran's foreign ministry has said the US naval blockade must be lifted before the strait reopens fully. Iran and Oman are talking about an arrangement for managing shipping, without US participation. None of that is a settlement, and the oil market has priced it accordingly.

⚙️ How a barrel of oil reaches a technology stock

The chain has four links, and each one is mechanical rather than a matter of sentiment.

The first link is the obvious one. Oil is an input to freight, plastics, fertiliser, power generation and the cost of moving every physical good, so a sustained rise in crude raises headline inflation. The July US CPI report showed this working with a lag: energy prices fell 1.5 per cent for the month, but were still up 14.7 per cent over the year, with petrol up 24.6 per cent and fuel oil up 39.1 per cent, per the Bureau of Labor Statistics.

The second link is where it gets interesting. Higher and more volatile inflation changes what lenders demand for parting with money for thirty years. That extra compensation is called the term premium, and it is the piece that has actually moved. The Federal Reserve's target range is 3.50 to 3.75 per cent after it held rates on 29 July. The 30-year Treasury yield touched 5.327 per cent on 18 August, its highest since 2007. The long end sits well above the policy rate, and the gap is not a forecast of Fed hikes. It is the price of uncertainty about inflation over decades, and an oil supply shock is exactly the kind of event that widens it.

The third link is the one most retail investors have never had explained to them. A share price is the value today of profits a company will earn in the future, and the long government bond yield is the base rate used to discount them. Raise that rate and every future rupee of profit is worth less today. The arithmetic is unforgiving for anything long-dated. A dollar of profit expected in ten years is worth about 9 per cent less when discounted at 5.3 per cent instead of 4.3 per cent. Push the same dollar out to thirty years and it loses about 25 per cent of its present value. Nothing about the company has to change for that to happen.

The fourth link is portfolio concentration. The companies whose value sits furthest in the future are high-multiple software and semiconductor businesses, and those are precisely what Indian retail portfolios in US equities are concentrated in. Technology alone accounted for 37.2 per cent of the S&P 500 as at 14 August, per S&P Dow Jones Indices. The sector that benefits from expensive oil, energy, was 3.4 per cent of the index at the end of July. The natural hedge is roughly a tenth the size of the exposure.

There is a fifth point that is not a link in the chain but explains why it keeps operating. The Federal Reserve cannot fix this. A supply shock raises prices and lowers output at the same time, so cutting rates worsens the inflation and raising them worsens the growth hit. That is why the July meeting produced a hold with dissents rather than a decision, and why markets are split, with CME FedWatch pricing roughly a 30 per cent chance of a September hike while prediction markets put the odds of at least one hike this year near a coin flip. When the central bank is boxed in, the oil price sets more of the discount rate than the central bank does.

The dynamics covered in this article affect every US stock in your portfolio. Trade from India on the Winvesta app. No US bank account needed!

🚀 Join 60,000+ investors, become a paying subscriber or download the Winvesta app and fund your account to get insights like this for free!

Get the app!

🏆 Winners and losers when oil sets the discount rate

The pattern is not the one the textbook suggests. Directions below are illustrative and directional, not forecasts.

Two of those rows deserve a note. Owning energy stocks as an oil hedge sounds sensible and works badly at index level, because 3.4 per cent of the portfolio cannot offset a repricing of 37.2 per cent of it. And long-dated government bonds, the asset most people hold as protection against equity falls, are being hurt by the same force at the same time. On 18 August the S&P 500 fell 0.6 per cent, the Nasdaq fell 1.3 per cent, and the 30-year yield rose to a 19-year high. Stocks and bonds fell together, which is what happens when the shock is to inflation rather than to growth.

🇮🇳 What this means for Indian investors holding US equities

Every Indian investor holding US equities is exposed to this twice, and the second exposure is the one nobody accounts for.

The first is the one described above. Your Nasdaq holdings get repriced by the US long bond yield, which is being pushed up by an oil shock. The second is that India is the most oil-exposed large economy in the world on the import side. India's crude import dependency ran at 88.7 per cent in FY 2025-26 against consumption of about 5.5 million barrels a day. The crude oil import bill rose 61.2 per cent year on year to 49.8 billion dollars in the April to June quarter alone. When the strait closed, the Indian crude basket went from around 70 dollars to above 120, reaching 113.57 dollars on 11 March per petroleum ministry data.

That flows into the rupee, which is why the currency layer compounds rather than cancels. The rupee sat around 95.7 to the dollar on Wednesday, a three-week low, down about 0.7 per cent over the month and roughly 10 per cent over the year, with higher oil, higher US yields and a risk-off mood all pulling the same way, per Trading Economics. A weaker rupee flatters the rupee value of your US holdings, which is the one piece of good news here. It also raises the cost of every subsequent dollar you remit under LRS to buy those holdings, and it feeds domestic inflation through the same import bill. Estimates of the sensitivity vary by method: a 10 dollar move in crude is put at 30 to 40 basis points of GDP on the current account deficit, and somewhere between 20 and 50 basis points on CPI depending on how much is passed through to retail fuel prices.

India has done more about the supply side than most people give it credit for. Sourcing has moved from around 27 countries a decade ago to more than 40, and the share of crude arriving outside Hormuz has risen from roughly 55 per cent to about 70 per cent, per the petroleum ministry in March. What diversification cannot change is the price. Crude is a global market, so rerouting protects delivery, not cost. LPG is the harder exposure, with close to 90 per cent of imports historically transiting the strait, and more than half of LNG imports as well.

The practical response is not to trade this. Positioning a portfolio for a ceasefire is a bet on a negotiation between two governments that have missed their own deadline. What the situation does argue for is knowing your actual duration. If your US exposure is concentrated in high-multiple technology, you own a leveraged bet on long-term interest rates whether or not you think of it that way, and this is the year to find out how much. Broad index exposure through a systematic monthly investment keeps buying through a repricing rather than guessing at the end of it. And if the entire US position is unhedged tech while the rupee also sits on the wrong side of the same shock, that is a concentration decision, not a diversification one.

🔭 What to watch from here

Transit counts through the strait are the cleanest real-time signal available, and they are public. Lloyd's List Intelligence, Windward and NBC's tracker all publish traffic data. A sustained return towards pre-war levels would do more for equity multiples than any earnings season.

War risk insurance premiums matter more than the headlines about talks. Shipowners commit money, not statements. Premiums easing from the current 3 to 10 per cent of hull value back towards 1 per cent would tell you the market believes a settlement before any communiqué does.

The 30-year Treasury yield is the transmission channel to watch, not the Fed funds rate. Yields have now spent their longest continuous stretch above 5 per cent since before the financial crisis. A move back below 5 per cent while oil stays high would mean the term premium story is fading. Both staying high together is the combination that keeps compressing multiples.

The September FOMC meeting is where the boxed-in problem gets tested. Pay attention to how the statement treats energy prices. A Fed that describes oil inflation as transitory has room to wait, and one that does not has to choose between two bad options.

For the India side, the crude basket price and the monthly import bill are the numbers that decide how much of this reaches domestic inflation and the rupee. The RBI has been intervening daily through dollar sales, which limits volatility but does not change the underlying arithmetic.

If this changed how you read an oil headline, pass it on.

Share

🏁 The bottom line

The oil price is no longer just an input cost. With the Fed unable to offset a supply shock, crude is doing much of the work of setting the long bond yield, and the long bond yield sets the discount rate on every future profit your portfolio owns. That makes high-multiple US technology, the most popular holding among Indian retail investors in US equities, the most oil-sensitive position most of them own, and the energy exposure that would offset it is a tenth of the size. Five trading days took the S&P 500 from a record close to three consecutive falls, without a single company reporting anything. The lesson is not to trade the Gulf. It is to know that a portfolio built entirely on profits arriving in the 2030s is a bet on interest rates, and that a shipping lane currently gets a vote on those.

📊 By the numbers

• 20.9 million barrels a day: oil flows through the Strait of Hormuz in the first half of 2025, about 20 per cent of global petroleum liquids consumption and a quarter of seaborne oil trade, per the EIA

• 14.6 million barrels a day: the same flow in the first quarter of 2026, down almost 30 per cent year on year, per the EIA

• 3 to 10 per cent of hull value: war risk insurance for a Hormuz transit, against about 0.25 per cent before the war

• Near 92 dollars: Brent crude on 19 August, a fourth straight session of gains and up about 37 per cent over twelve months, per Trading Economics

• 5.327 per cent: the US 30-year Treasury yield on 18 August, its highest since 2007, against a Fed funds target range of 3.50 to 3.75 per cent

• 37.2 per cent versus 3.4 per cent: technology's weight in the S&P 500 as at 14 August, against energy's weight at the end of July, per S&P Dow Jones Indices

• 7,798.99: the S&P 500's record close on 13 August, followed by three consecutive falls

• 88.7 per cent: India's crude oil import dependency in FY 2025-26, on consumption of about 5.5 million barrels a day

• 49.8 billion dollars: India's crude oil import bill for April to June 2026, up 61.2 per cent year on year

• Around 95.7: rupees per dollar on 19 August, a three-week low, roughly 10 per cent weaker over twelve months, per Trading Economics

Disclaimer: All content provided by Winvesta India Technologies Ltd. is for informational and educational purposes only and is not meant to represent trade or investment recommendations. Remember, your capital is at risk. Terms & Conditions apply.