The most powerful force in global markets is still not artificial intelligence. It is the price of money. The S&P 500 closed at 7,718.60 on 4 September 2026, within about 1% of its mid-August record of 7,798.99, after a month in which a 19-year high in the 30-year Treasury yield and then a hot jobs report each did more to move equities than any earnings release.
Nothing about the AI story changed in those sessions. No hyperscaler cut its spending, no chip company missed. What changed was the return on doing nothing.
Why does a bond yield decide what tech stocks are worth?
Because every equity valuation is a comparison. A 30-year US Treasury yielding 5.33% is a guaranteed government return that an investor can take instead of owning a semiconductor company at 30 times earnings, so when that yield rises, the price an investor will pay for future profits falls. The effect is largest on the companies whose profits sit furthest in the future, which describes most of the AI complex.
The mechanics showed up exactly where the theory says they should. The PHLX Semiconductor Index fell about 5% in the 18 August 2026 session, with Intel and AMD down around 6%, Micron down 5%, and Nvidia off more than 2%. The Dow, full of companies earning money today rather than in 2032, fell 0.84%, less than the Nasdaq's 0.87% but with far less drama underneath.
What actually caused the yield spike?
Two things at once, which is why it moved so fast. Persistent inflation worries and higher oil prices, driven by renewed tension around the Strait of Hormuz, collided with an enormous government borrowing calendar, and long-dated debt is where that pressure lands first. There is a third ingredient specific to this cycle: the AI build-out is increasingly debt-funded, so the bond market is now absorbing supply from the same story that is lifting the equity market.
Then the US Treasury intervened. The Treasury Department announced it would at least double repurchases of 10, 20 and 30-year debt over the coming months, which pulled the 30-year yield down more than 10 basis points to 5.184% and let the S&P 500 snap its three-day losing streak on 18 August. By 19 August the relief had partly reversed, yields climbed back, and Dow futures dropped 400 points with Walmart falling on results. Relief that depends on a buyback announcement is not the same as relief that comes from lower inflation.
Is this the AI bubble popping?
Not yet, and the distinction matters. A bubble bursts when the cash flows disappoint; this was a discount-rate event, where the same expected profits are simply worth less at a higher yield. Hyperscaler capital expenditure plans of roughly $670 billion for 2026 have not been withdrawn, and AI-linked earnings are still expected to drive around 40% of S&P 500 profit growth this year.
What the week did expose is fragility. The Magnificent Seven make up roughly 30% of the S&P 500 by market capitalisation, so an index at 23 to 24 times trailing earnings is really a concentrated bet on about seven balance sheets and one interest rate. Our AI bubble 2026 piece works through what an actual bust would look like, and the debt-funded portion of the build-out that our AI data centre capex coverage tracks is precisely the link between the two markets.
The next hard data point arrives on 26 August, when Nvidia reports its July quarter with analysts looking for revenue in the $93 billion to $95 billion range. That single call now carries both theses at once.
What does this mean for Indian markets?
The transmission was visible within hours. After the Treasury buyback plan eased bond stress, the Sensex opened 558.77 points higher at 77,468.45 and the Nifty opened at 24,225.45 on 20 August 2026, with Infosys up 1.87% and HCL Tech and TCS also gaining. Indian IT is the most directly correlated sector, since its clients are the same US firms funding the AI cycle.
The less visible channel is flows. Higher US yields shrink the return gap that pulls foreign money into Indian bonds and equities, and the rupee slipped to about 95.76 per dollar (as of 20 August 2026), a three-week low, on firmer oil and elevated US yields. A sustained 30-year yield above 5.3% is a headwind for every emerging market, India included, regardless of how good domestic earnings look. That is worth holding alongside the domestic valuation question our is the Indian market overvalued analysis takes up.
Risks to monitor
The second risk is that the Treasury's buyback fix is temporary. Repurchases change who holds the debt, not how much of it exists, and the 19 August reversal showed how quickly yields can climb back once the announcement effect fades.
The third is concentration inside the concentration. Memory and custom-chip names fell harder than Nvidia this week, which hints that investors are starting to separate the AI winners from the AI adjacent rather than selling the theme as a block. This is general information, not investment advice.
For most of 2026 the question about US equities has been whether AI earnings could keep justifying the price. The last week reframed it. The AI earnings are showing up on schedule, and the market fell anyway, because the alternative to owning stocks got better than it has been since 2007.