Everything About AI
NextFin News - US stocks slipped for a second straight session on Tuesday as the 10-year Treasury yield pushed to its highest level since 2007, driving home a market reality that has upended a generation of portfolio construction: equities are once again taking their orders from the bond market, and the old rule that government bonds cushion stock losses is not holding.
The benchmark 10-year yield closed at 5.24% on September 28 after touching an intraday high of 5.274%, a level last seen in June 2007, while the 30-year yield reached 5.62%, its highest since 2002. The S&P 500 fell 0.2% and the Nasdaq Composite dropped 0.1% as investors braced for August personal-consumption data due Wednesday and for quarter-end positioning pressure.
Why Bonds Stopped Being the Hedge They Were Supposed to Be
The tape tells a story of a market caught between two forces. On one side, the S&P 500 sits less than 3% below its August 13 record close of 7,798.99 and is still up about 12% in 2026, lifted by AI-driven earnings expectations. On the other, the entire US Treasury curve is trading within 5 basis points of 5%, a configuration not seen in nearly two decades, and the third quarter has already delivered the biggest quarterly slump in corporate bonds since 2022 alongside a 40% rebound in oil prices.
The immediate trigger is inflation. The Federal Reserve raised its benchmark rate by 25 basis points on September 16 to a range of 3.75%-4.00%, the first increase since July 2023, after core inflation held at 3.3% year over year in July - well above the central bank's 2% target. The Fed's own projections show 16 of 18 officials expecting at least one more quarter-point increase this year, with the median dot plot pointing to a 4.1% funds rate by year-end 2026.
The market reaction has been asymmetrical. Seven of the 11 S&P 500 sectors fell on Tuesday, led by energy's 0.9% decline, while utilities gained 1.1% - a rotation into defensive income that typically appears when investors start discounting cash flows at a higher rate. Apple was the Dow's biggest loser, down 2.7%, while Carnival surged 13% as the index's biggest gainer - a dispersion that suggests stock-picking is beginning to matter more than beta.
But the deeper story is not Tuesday's tape. It is that the relationship between stocks and bonds that held for much of the past two decades has broken down, and the breakdown is now actively shaping prices.
The Valuation Math Is Getting Less Forgiving
For most of the 2000s and 2010s, rising Treasury yields were read as a sign of strong growth, and stocks rose with them. When yields fell, they signaled panic, and stocks fell too. Either way, the correlation between stock and bond prices was positive, and 60/40 portfolios worked.
That regime ended when the shock hitting the system changed. D. E. Shaw Group research has long argued that the negative correlation between equity and Treasury prices - the environment in which bonds act as an effective hedge - persists when the dominant shocks are about growth strength or risk appetite. When the shocks are instead driven by unstable inflation expectations, the relationship flips.
That is precisely the shock mix of 2026: a war disrupting oil supplies, an AI capital-expenditure boom demanding financing, a Federal Reserve that just restarted a hiking cycle, and a federal debt trajectory that has pushed net interest expense to a record 3.3% of GDP in the second quarter, compared with 1.7% of GDP when the 10-year yield last sat around 5% in 2007.
The consequence is that higher yields no longer signal growth confidence; they signal a higher discount rate applied to future earnings with no offsetting growth reassurance. Scotiabank analysts put the threshold plainly: "When U.S. 10-year yields exceed 5%... the correlation tends to be negative, meaning that higher bond yields are generally associated with lower stock prices. This was broadly the case from the late 1960s through the late 1990s."
The 10-year yield crossed that 5% line in this cycle in mid-September and has not looked back.
The AI Trade Is Now a Bond-Market Trade
The transmission mechanism runs through the equity risk premium - the excess return stocks promise over risk-free government bonds. When that premium compresses, equities have less cushion against higher discount rates.
Societe Generale calculates the US equity risk premium has fallen to about 3.5%, close to the 3% threshold where it sees equities starting to struggle against more attractive bond returns. The bank's model keeps the US cost of equity above 7.8% this year while Treasury yields have climbed more than 75 basis points since the start of 2026.
JPMorgan's proxy for the premium - the gap between the S&P 500's equity discount rate and the real 10-year Treasury yield - has fallen to just 2.2%, a new low for the post-financial-crisis period since 2007 and 90 basis points below the bank's long-term average.
"While it is still some way above its 2000 trough, this implies that there is currently more limited room before a further rise in bond yields starts becoming a problem for the equity market," JPMorgan strategist Nikolaos Panigirtzoglou wrote.
The asymmetry matters. When the equity risk premium is wide, stocks can absorb a rise in yields because the starting expected return is high. When the premium is thin, every additional basis point of yield comes directly out of equity valuation with little buffer. At 2.2%-3.5%, depending on the model, that buffer is close to exhausted.
The Counter-Thesis: A Cyclical Squeeze, Not a Regime Shift
The second-order channel is where this story departs from a simple "higher yields hurt growth stocks" narrative. The AI investment boom that lifted the S&P 500 to record highs is itself a major driver of the bond selloff.
Data center construction, chip fabrication plants, and power infrastructure require enormous upfront capital. Much of that capital is being raised in the bond market. When corporations compete with the US Treasury for a finite pool of savings, the price of duration rises for everyone - including the technology companies whose valuations depend most on low discount rates.
This is the feedback loop the market is only beginning to price: AI earnings growth requires cheap capital, but AI capital demand is making capital less cheap. The tension surfaced in the third quarter, when corporate bonds posted their worst quarterly performance since 2022 even as AI-linked equities marched toward records.
The cracks are visible in the financing edifice. Bonds issued to fund data center construction have come under pressure, and Anthropic's IPO prospectus showed revenue growing twelve-fold last year alongside operating losses exceeding $8 billion and net losses above $40 billion, with the company targeting a valuation of more than $2 trillion. That combination - explosive revenue growth, deep losses, and a multi-trillion-dollar price tag in a 5%-yield world - is the clearest illustration of the regime change.
Michael Burry, the investor known for his early bearish bet on subprime mortgages, has moved up his timeline on the AI trade. "Fundamentally, I am moving timelines up," Burry wrote in his Monday investment newsletter, explaining his switch from short positions to put options on key AI stocks because options are "relatively cheap due to exceptionally tight volatility measures such as the VIX."
The strongest case against the bearish read is that the stock-bond correlation breakdown is a cyclical episode, not a structural one, and that equities have already proven they can look through it.
The S&P 500 sits less than 3% below its record despite a summer bond-market rout. Investors have bought every dip this year, and the earnings backdrop remains solid: the US economy is expanding at an annual pace of around 8% to 9% in nominal terms, the unemployment rate is 4.1%, and productivity growth is strong. BlackRock's chief investment strategist Wei Li captured the benign interpretation: "What looks like a breakdown in the historical correlation between them may simply reflect a different macroeconomic regime" - one that investors can adapt to rather than flee from.
There is also a mechanical argument. Rolling stock-bond correlations have swung before, ranging from deeply negative to moderately positive across the post-2002 period. A two-day equity decline on yields near 5.25% does not, by itself, prove that the year-to-date equity rally is over. If the August core PCE print due Wednesday comes in at or below the expected 3.4% year over year, rate-hike expectations could ease and the bond selloff could pause - giving stocks room to recover, just as they did after the brief 5% touch in October 2023, when the one-day episode was accompanied by only a mild selloff.
The counter-thesis has force, but it depends on one assumption: that inflation is a cyclical problem the Fed can talk down with one or two more hikes. If inflation proves structural - driven by deglobalization, energy supply shocks, and the capital intensity of the AI build-out - then the Fed's 2% target becomes a moving horizon rather than a destination, and yields stay higher for longer.
What Would Prove the Bearish View Wrong
The falsifying signal is specific: if core PCE prints at or below 0.2% month over month for two consecutive months - taking the year-over-year rate back toward 2.5% - and the 10-year yield falls back below 4.75% on that evidence, the structural-inflation thesis breaks and the correlation likely flips back toward its growth-driven, positive form. Under that scenario, the September selloff would be a cyclical air pocket, and the record-high equity market would be justified.
The signal that would confirm the bearish view is the mirror image: core PCE at or above 0.3% month over month, the 10-year yield holding above 5.25%, and the equity risk premium compressing below the 3% threshold Societe Generale flags. That combination would mean bonds are no longer just volatile - they are the binding constraint on equity valuation.
Outlook: Who Wins and Who Is Exposed
The mechanism has a clear set of beneficiaries and exposed parties. Higher-for-longer yields favor the parts of the market that produce current cash flow rather than distant promise: utilities, quality large-caps with strong balance sheets, and short-duration credit. They hurt the long-duration end of the market - unprofitable growth stocks, speculative AI financing, and the corporate-bond issuance that funds the AI build-out. The rotation into utilities on Tuesday, up 1.1% while energy and the broader market fell, is the first visible sign of that reallocation.
Split by time horizon, the picture differs sharply. In the short term, sentiment and quarter-end positioning dominate; a benign August PCE print on Wednesday could spark a relief rally and push the S&P 500 back toward its record. Over the medium term, the third-quarter earnings season will test whether AI revenue growth can keep pace with the rising cost of capital - and whether margins can hold as financing costs roll higher. Over the long term, the question is structural: whether the inflation shocks of the 2020s represent a permanent regime change or a cyclical overshoot that mean-reverts.
The base case is that the correlation stays negative while yields remain above 5%, capping equity upside and raising volatility without necessarily triggering a deep bear market. The upside case requires inflation to cool faster than the Fed's own 3.4% year-end core PCE projection, pulling the 10-year yield back below 4.5% and reopening the path to new equity records. The downside case is a sustained break above 5.25% on the 10-year alongside sticky inflation - the 2007 playbook, when a sustained rise above 5% preceded the S&P 500 shedding nearly 5% over two months and much more afterward.
Positioning surveys already lean toward the cautious camp. Transamerica's year-end 2026 outlook calls for an S&P 500 level of 7,400 - below the current index - alongside a 10-year yield of 3.75% and a federal funds rate of 3.00%-3.25%, implying the tightening cycle is already closer to its peak than its start. That view is far from consensus, but it quantifies the risk: if the Fed's median projection of one more hike this year is right, the bond market's pain may not be over.
All three cases hinge on Wednesday's inflation print and the Fed's reaction function. The central bank has said inflation "remains elevated" and has pledged to deliver price stability. The bond market is now asking whether that mandate can be met without breaking something in the asset base that was built on the assumption that bonds would always be there to catch falling stocks.
The market is not pricing a cyclical dip in yields. It is pricing a regime in which bonds are no longer the safe asset they were assumed to be - and until that assumption is proven wrong, equities will keep trading in bonds' shadow.
Explore more exclusive insights at nextfin.ai.
