Author: Zhuifeng Trading Platform
Global markets have been pulled in three directions this summer: a blowup in leveraged AI positioning, long-dated U.S. Treasury yields hitting a 19-year high during a rate-cut cycle, and repeated ceasefire agreements around the Strait of Hormuz that were signed and then broken.
At first glance, the three stories sit in different markets and do not naturally belong together. Goldman Sachs argued on Aug. 10 that they can be read through one lens: the world may be shifting from “excess savings” to “capital scarcity.” In that reading, AI is consuming funding, governments are borrowing heavily, and the strait is constraining physical supply. All three are drawing on the same limited pool of global capital.
The article says that, unlike the past two decades, central banks now have less room to keep injecting liquidity and suppressing rates.
That does not mean the three pressures will keep worsening without interruption. It does raise a broader question: if capital is genuinely becoming more expensive and harder to access, is the market prepared for that regime? Even if one of the three stress points eases on its own, the broader scramble for capital may remain. The article argues that July’s violent swings — equities plunging and then rebounding, Treasury yields moving higher even as the Fed cut rates, and crude recovering half of a drop within three days — may have been the start of a more volatile phase rather than a one-off event.
AI leverage: the thesis held up, the leverage did not
The first shock came from AI leverage.
Leopold Aschenbrenner, described in the article as an “AI stock god,” saw his hedge fund’s net asset value jump 439% in the first half. The strategy was straightforward: a 4x leveraged bet on AI hardware and short positions in software names. In July, the Philadelphia Semiconductor Index fell 28.6% for the month, hurting both sides of the book. A public position worth $16 billion was ultimately offloaded to Citadel. Millennium and Jane Street reviewed the trade and walked away, leaving only one buyer on Wall Street.
JPMorgan later warned that institutional capacity to absorb AI-related assets had fallen sharply. If a similar dislocation happens again, the article says the market may be left with “only retail investors to provide a floor.”
What makes that more striking is that the demand backdrop did not weaken. Taiwan Semiconductor Manufacturing Co. posted a 44.7% year-over-year increase in July revenue, a record high, while the top three U.S. cloud providers had a combined backlog of $2.3 trillion. Demand for AI did not slow. The problem was that the money used to fund the trade was becoming harder to find.
The article points to Nvidia as a deeper signal. On Aug. 10, Jensen Huang did not unveil a faster chip. Instead, he worked with six large asset managers including BlackRock and Blackstone to build a financing platform. The interpretation offered in the piece is blunt: customers can no longer afford the chips outright and need Nvidia to help connect them with institutional funding.
The market did not reward that move. Nvidia shares fell that day and its CDS moved close to a historical high. Jim Chanos said it reminded him of 2008. The concern, as framed in the article, is simple: if the essence of the $500 billion plan is “lend money to customers so they can buy your own chips,” how long can that loop keep running? A sector that needs hundreds of billions of dollars in financial engineering to keep moving is, by itself, evidence of tighter capital.
Treasuries: one way to explain why rate cuts did not pull down long yields
While AI is burning cash, the government is borrowing at record scale.
The article says the Federal Reserve has cut rates six times, for a cumulative 175 basis points, yet the 30-year Treasury yield kept rising and reached 5.27% on Aug. 1, the highest level since 2007. In that setting, markets are not just pricing the Fed. They are pricing the U.S. fiscal position.
U.S. federal debt rose above $40 trillion in August. FY2026 net interest expense reached $1.039 trillion, the first time it exceeded the defense budget. The article says that for every $5 of tax revenue collected, $1 now goes directly to interest payments. Another $15 trillion of debt needs to be refinanced within a year, and every 1 percentage point increase in refinancing costs mechanically widens the deficit. The share of Treasuries held by foreign central banks has already fallen from 34% to below 24%.
Goldman’s conclusion is direct: the world has moved from “excess savings” to “extreme capital scarcity.” More than $800 billion a year is going into AI infrastructure, while reindustrialization and defense rebuilding are happening across countries and sovereign refinancing needs are hitting at the same time. Those funding demands are colliding with a Treasury market where buyers are retreating. Supply is growing. Demand is thinning. Yields move higher.
Fed officials are not speaking with one voice either. The article says Chair Warsh has sharply reduced forward guidance and said “the market has already done much of the tightening work,” signaling little urgency to raise rates and a preference to let markets do the tightening. Cleveland Fed President Hammack took the opposite line in public, saying current rates “have not become clearly restrictive” and that “multiple rate hikes may be needed.” Goldman’s read is that the Fed now looks more like a passenger than a driver.
That also helps explain why Aschenbrenner’s 4x leverage was so fragile. Concentration mattered, but so did funding costs. The capital-intensive AI buildout collided with a government borrowing wave, while central banks no longer had the same room to shield both sides.
Hormuz: why a ceasefire document no longer settles the market
Financial markets are pricing scarcity in capital. The physical world is sending a parallel message.
Since the U.S.-Iran war began in late February, Brent crude has traced a repeated pattern: $126 during the March fighting, $72 at the June ceasefire, $85 after the July breakdown, and $87.72 in August after Iran set out six conditions and Donald Trump responded with what the article called a “half-negotiation state.” Each move followed the same script. Markets priced “this may stop,” then repriced for “it has started again.” After several rounds, traders stopped trusting that the latest truce would hold.
The question is no longer simply whether fighting continues. It is how long any ceasefire can last. The article cites Zolghadr, secretary of Iran’s Supreme National Security Council, as saying the strait would remain closed if the U.S. did not change its behavior. Trump said he was prepared to increase economic pressure. Talks continue, but neither side has stepped back.
OVX, the oil volatility index, is around 60, or 3.2 times VIX. The options market message, according to the article, is clear: the biggest tail risk is not in U.S. equities but in the waterway.
The buffer is also thinner than before. The U.S. Strategic Petroleum Reserve has fallen to 298.7 million barrels, the first drop below 300 million since 1983. Before the war, it stood at 415 million barrels. After a one-time release of 172 million barrels in March, the reserve kept falling on a net basis. The International Energy Agency described it as the “largest supply disruption” in the history of the global oil market. If Hormuz is hit again, the article argues, Washington’s ability to calm prices by releasing reserves is at its weakest point in four decades.
Where the three lines meet: from excess savings to capital scarcity
The article argues that the three events may not be coincidences.
The AI leverage blowup happened because 4x leverage is hard to sustain when rates are rising. Long-end Treasury yields surged because governments and companies are competing for the same investor money. The Hormuz problem appears geopolitical on the surface, but underneath it is about the sudden scarcity of a form of physical capital: oil. Equities, bonds and commodities are moving in different ways, yet they may share the same root cause — a structural reversal in the balance between global capital supply and demand.
Goldman calls that reversal a shift from “excess savings” to “capital scarcity.” In the post-2008 world, savings were abundant, investable projects were relatively scarce, rates stayed low, and central-bank QE acted as a backstop while governments borrowed cheaply. The article says that framework lasted close to two decades.
By 2026, it says, the framework had been broken from both sides. On the supply side, global central banks moved from net buyers of Treasuries to net sellers. On the demand side, AI infrastructure, reindustrialization, defense spending and the energy transition all released trillion-dollar-scale capital needs at once. Goldman specifically says AI capital expenditure exerts the strongest upward pressure on rates among the four because it pushes up both risk-free yields and credit spreads.
Even if the Strait of Hormuz reopens tomorrow, and even if leveraged AI positions are fully cleared, the structural pressure on rates would remain. The root problem, in this framing, is not the individual event. It is the change in regime.
Another way to frame it: a triple hit to liquidity
The article also recasts the story as three forms of liquidity pressure arriving at the same time.
Trading liquidity was hit first. The forced unwind of 4x leveraged AI funds drove the Philadelphia Semiconductor Index down 28.6% in a month and set off a sequence of decline, liquidation, deeper decline and more liquidation. Funds with related exposure were forced to cut risk.
Bond-market liquidity is the larger reservoir under pressure. With $15 trillion of debt to refinance within a year and roughly $2 trillion of additional net issuance needed in FY2026, the Treasury is absorbing hundreds of billions of dollars from markets every month while buyers step back. Yields move higher, interest costs rise, deficits widen, and issuance grows again — a self-reinforcing loop.
Physical liquidity is the third blow. Some 27% of global seaborne crude routes have been cut off, there is no alternative route, and the SPR buffer has been reduced to a 40-year low.
The timing overlap matters. The article says the three liquidity pressures formed a rare resonance, a “triple kill” in liquidity. July’s extreme market action expressed that directly in prices: the Nasdaq fell for six straight sessions and then rebounded 2.8% in a single day, the 30-year yield reached a 19-year high, and Brent dropped 14% in two weeks before recovering half the decline in three days.
The framework is not presented as a prediction tool. It explains why July was so violent, but it does not say what August must look like. The three pressures also ease on different schedules: AI positioning depends on when Citadel distributes the assets back into the market, Treasuries depend on the Treasury Department’s issuance pace, and Hormuz depends on the pace of geopolitical bargaining. They do not have to move together. Still, their overlap helps explain why this summer felt more volatile than any single indicator suggested.
After this summer: a preview of higher volatility as a norm
On the equity side, second-quarter earnings have already done a lot to confirm AI demand. Cloud providers posted fresh highs in growth and order backlogs reached records. Yet the July unwind exposed the other side of the trade. In a crowded, leveraged sector, being right on the thesis is not the same as surviving long enough to see the thesis play out. The article singles out JPMorgan’s warning that if institutional balance-sheet capacity keeps shrinking, the next correction may have even fewer natural buyers than July had.
On the Treasury side, Goldman’s line that the Fed is “a passenger, not a driver” is repeated as a core message. If the fiscal path does not change, structural pressure on long-end yields will not disappear after a few rate cuts. Whether the 10-year can fall back below 4% depends less on the next FOMC and more on whether global savings are willing to flow back into Treasuries. That sits outside the Fed’s control.
On oil, the risk premium tied to Hormuz has changed from a one-off shock to an ongoing discounting mechanism. Even if another temporary ceasefire is reached, the large gap between OVX and VIX is unlikely to disappear quickly. With the SPR at a 40-year low, the market would be more fragile and the government less capable if another supply-side disruption arrives. Goldman’s base case, as cited in the article, is that “neither side appears willing to pursue large-scale escalation and energy prices should remain bounded,” but the article adds that this view is fragile in itself: if markets move too far the other way, concerns can reappear quickly.
The article closes on a broader point. The real legacy of the summer of 2026 may not be the price move in any single asset. It may be that markets have started to pay a premium for a new regime defined by capital scarcity. AI leverage, Treasury financing and Hormuz supply risk are three slices of that story. They collided this summer, but the structural forces beneath them have not finished playing out.