Market Overview October 28, 2025
Summary
The speaker’s overall judgment is that AI is already in a “left foot stepping on right foot” bubble, but the probability of tail risk or a sharp crash between now and early next year is low. “Bubbles always exist.” The key question is not whether blood pressure is high, but whether a blood vessel bursts. Compared with last year-end’s “liquidity dam”—when positions were highly concentrated and rates were elevated—volatility, risk premia and CDS have already released some pressure. “It’s not that there is no risk; risk in this dimension is relatively small.”
The government shutdown has put the Fed in a “driving with its eyes closed” state, and the speaker still sees two more rate cuts as the most likely outcome this year. CPI came in about 20bp below expectations, mainly because of shelter, while the halt in manual data collection could distort October figures. Alternative data also point to weaker employment than before the pandemic. Tariffs account for about 0.5 percentage points, putting overall PCE at roughly 2.9%-3% and ex-tariff PCE at about 2.4%-2.5%. The Fed will keep driving forward, but has no evidence to pivot toward an inflation-first stance.
Room for rate cuts is limited, and markets should not expect policy rates to fall to 1%-1.5% and once again lift risk assets indiscriminately. With stocks and gold both at elevated levels and PCE still near 3%, “who would dare deliver a negative real interest rate?” Even with a new Fed chair, real-economy rates face a hard floor. The speaker added that China’s government buying is another reason behind gold’s rally. He sees 4% on the 10-year Treasury as the lower bound once supply normalizes, not the midpoint of a new normal.
If the shutdown continues to depress long-term yields, the speaker’s trade is to go long 10-year rates through swaptions. Issuance is constrained during the shutdown and supply disappears, so bond prices naturally rise. Once the government reopens, the deficit and duration supply will return. His verbal estimate is that next year’s DV01 could be around “900M” (his words were “should be 900M”), versus roughly “780M” this year. His high-conviction view: “It will be very difficult for the 10-year to remain below 4%.”
AI industry investment is not pure circular churn because current ROE and the observable earnings path remain strong. The speaker compared it to “2000 local-government debt—Shanghai was building the subway,” rather than the 2020-style local-government debt that generated no returns. His stated investment-return comparison was roughly 11 for China versus 23 for the US. Anthropic and OpenAI have use cases including model training; Meta and Google provide annual reports that show whether AI spending can quickly translate into earnings, improve margins and even lift gross margins. Circular investment alone is therefore not enough to trigger a collapse.
The credit scare since September looks more like isolated fraud combined with HYG options deleveraging than a 2008- or 2023-style banking crisis. Fifth Third was tied to a subprime car loan whose collateral had been swapped out; the speaker described it as essentially fraud. After First Brands went bankrupt, Zions Bank and Western Alliance were also implicated in or uncovered fraud, with total provisions of about $270M. Using Silicon Bank as an example, he noted assets of roughly $77B, capital of about $7B and provisions of only around $50M—yet it still recorded its second-most-profitable quarter on record. “CDS being widened gives the market more room,” reducing future crowding around the same exit.
Worsening auto credit is a concentrated repricing risk in the industrial chain, not proof that US consumer credit is collapsing across the board. Auto loan default rates are rising rapidly, but retail loans, personal loans and credit cards have not deteriorated in tandem. The speaker relayed and agreed with a friend’s explanation: China’s low-priced new cars, sold globally at near-zero margins, are disrupting the used-car markets in Europe and the US, breaking the chain that sends those vehicles on to Asia, Africa and Latin America, and ultimately feeding back into weaker new-car sales in Europe and the US. He explicitly said to “stay away from autos and automaker stocks,” citing his verbal estimates that Honda’s gross profit fell about 50%, Porsche’s fell to zero and Mercedes’ declined about 69% in the first 3 quarters.
Deep dive
1. Data distortions force the Fed to keep leaning toward growth
The government shutdown has lasted nearly a month. Its negative economic impact is coming less from layoffs than from tax refunds and government spending that should have taken place but “cannot be spent now.” Political pressure on both sides means the shutdown cannot continue indefinitely, but its duration determines the near-term fiscal drag.
The only data point that qualifies as tier one is CPI, which came in about 20bp—or 0.2 percentage points—below expectations. The decline was driven mainly by shelter, and came through relatively quickly. The speaker’s caveat: “This data may be distorted.” With manual collection halted in October, the next CPI report may not be reliable either.
His inflation breakdown is as follows: tariffs contribute about 0.5 percentage points, putting headline PCE at roughly 2.9%-3% and ex-tariff PCE at about 2.4%-2.5%. Underlying inflation is steadily moving toward 2%, with no clear “spike up,” but it remains far from a level that would permit negative real interest rates.
Alternative indicators such as job openings are of limited reliability, but they still point to employment weaker than before the pandemic. Fiscal policy is also tightening as tariffs and blocked spending weigh on activity. The Fed is therefore “driving with its eyes closed”: having already moved forward with growth as the priority, it will find it difficult to suddenly reverse course and pivot to inflation without new evidence.
2. Two cuts this year do not mean a return to ultra-low rates
The speaker sees Powell as relatively dovish. Combined with missing economic data and temporarily contractionary fiscal policy, that means the Fed will “most likely cut twice” this year. But this is simply a continuation of the current direction, not the start of an unconditional easing cycle.
The lower bound comes from the real-rate constraint. With PCE near 3%, ex-tariff inflation still at 2.4%-2.5%, and both stocks and gold rising, “anyone with a little common sense” would not push policy into negative real interest rates. A terminal rate of 1% or 1.5% is therefore “impossible.” He separately noted that China’s government buying is contributing to gold’s rally.
The current cycle is still being driven by government spending and AI demand, not unlimited monetary expansion. The speaker drew a clear distinction: the real economy has limited “bullets” for rate cuts, while the credit market has unlimited bullets. A change in Fed chair is not enough to justify betting that Bitcoin and other capital-market assets will once again be mechanically pushed higher by ultra-low rates.
3. A 10-year yield below 4% looks more like a supply dislocation
When the government shuts down, bond issuance is constrained, bonds naturally rally and yields fall. Once the budget is restored, the supply implied by the huge deficit will still have to return. “Bond supply is not about how much you issue; it is about your DV01.” Duration risk is the real pricing pressure.
The speaker’s verbal estimate is that next year’s DV01 supply could be around “900M” (his words were “should be 900M”), versus about “780M” this year. Given that scale of supply, he has high conviction that the 10-year will “find it very difficult to remain below 4%” and that equilibrium yields should sit meaningfully above shutdown-period levels.
The corresponding trade is to buy swaptions if the shutdown continues to push 10-year yields lower, using options to go long long-term rates. The bet is not on inflation suddenly spiraling out of control, but on mean reversion once the temporary supply vacuum ends.
4. AI has a bubble, but profitability still supports capex
The speaker first rejected the idea that AI is a pure bubble, then acknowledged that Nvidia and others are indeed creating a “left foot stepping on right foot” bubble. The two points are not contradictory. Valuations and circular financing can become overheated, but as long as investment continues to generate high ROE, the industry is not an empty shell.
His version of the “China local-government debt” analogy was: “It is 2000 local-government debt—Shanghai was building the subway,” not a 2020 project in which money was invested without generating returns. His stated comparison was investment returns of about 11 in China versus 23 in the US, giving the higher US valuation a return-based foundation.
There are 2 observable earnings paths. Anthropic and OpenAI have use cases including model training; the more direct case is Meta and Google’s digital media businesses. The speaker recommends reading their annual reports to determine whether AI investment can quickly generate earnings, improve margins and even lift gross margins. That is the fundamental reason he is not yet calling for the bubble to burst.
5. The true tail risk is multiple trading structures reversing at once
The danger at last year-end was not valuation alone, but too many “stone-squeezing-oil” strategies operating at once: selling black-swan insurance for small gains, selling vol, leveraged trend-following and vol-control strategies all adding risk in a low-volatility environment. Strategies that appear independent can produce selling in the same direction once an extreme scenario hits.
Under the speaker’s framework, rising volatility forces option sellers to buy back options, leaving dealers short gamma. The further the market falls, the more hedgers have to sell. When equities, CDS and credit options all trigger at once, “the dealers’ hedging books are all dumping,” turning local turbulence into a cross-asset liquidation.
The 10-year was around 4.5% last year, while the equity risk premium had been compressed to a very thin level—effectively “a huge liquidity dam, with the water level still high and the dam not strong enough.” At that point, all that was missing was a trigger. Today, the gap between equity earnings yields and interest rates is roughly 50bp, leaving more room overall. The dam is built differently.
VIX is now clearly below last year’s level, while short-end vol is no longer at the extremely low 8%-10% levels seen previously. Trump’s repeated changes in his China messaging have also lifted volatility. The speaker believes that even if VIX rises to 25, option sellers would not necessarily be wiped out across the board. His summary of the US-China relationship is that they “will never negotiate well, and will never go to war”; the persistent disturbance keeps risk pricing from falling asleep.
6. Regional-bank events look like neither 2008 nor 2023
The September trigger included a subprime car loan involving Fifth Third. Its collateral had been swapped out, which the speaker described as essentially fraud. After auto-supply-chain company First Brands went bankrupt, Zions Bank and Western Alliance were also implicated in or uncovered fraud, with related provisions of about $270M. The credit market is “a market where the swimmers are naked,” so isolated cases quickly revive the PTSD of 2008 and 2023. But the loss on this loan itself was not especially large.
The 2008 mechanism was completely different. Banks broadly held subprime mortgages, then used CDO tranching to repackage the roughly 8%-10% riskiest slices as AAA. Once defaults across the pool exceeded that range, the supposedly senior tranche could also be wiped out. Regulation is tighter today. Regional banks may have weaker risk controls, but the speaker said collateral ratios on commercial and car loans are both “above 40.”
2023 was a duration shock. Treasury yields rose from roughly 1% to 4%-4.5%; a 3-4 percentage-point move multiplied by about 8 years of duration produced an asset-liability loss approaching 20%, enough to erase bank capital. Current losses on individual auto loans are far smaller.
Using Silicon Bank as an example, he cited a balance sheet of about $77B, capital of roughly $7B and provisions of only around $50M. The bank still recorded its “second most profitable quarter.” This does not prove that future real-estate risk is impossible, but in the current state there is “definitely no” systemwide banking-credit shock.
7. Wider CDS spreads release crowding; the auto supply chain is the localized mine
CDS spreads have widened even as the S&P has continued to rise, creating an apparent divergence. The speaker’s explanation is technical: HYG, the most liquid high-yield ETF, had an overly concentrated options-selling community. A one-way move forced dealers to sell hedges, sharply lifting put skew and volatility on the protection side.
This widening may not amplify equity risk; it may instead function as an early evacuation. “The bridge is still just as wide, but some people crossed first—they have already blown up.” CDS is no longer extremely cheap, which means some tail risk has already been repriced and fewer people will have to squeeze across the bridge when a real shock arrives.
The decoupling of auto loans from other forms of consumer credit is central to the speaker’s rejection of a macro-recession explanation. If the economy were deteriorating across the board, retail loans, personal loans and credit cards should all be “a complete mess.” For now, the rapid deterioration is concentrated in auto defaults, pointing to an industry structure problem rather than a comprehensive collapse in household balance sheets.
The speaker relayed and agreed with a friend’s view that China’s new cars are being dumped globally at cost or near-zero margins, disrupting used-car markets in Europe and the US while also wiping out India and Asia, Africa and Latin America as destinations for those vehicles. Once the used-car export chain is damaged, new-car sales in Europe and the US also become more difficult. Xiaomi SU7 may be targeting the $20K-$30K price range. On the new integrated auto-manufacturing model, he cited Tesla and said its iteration and cost competition versus traditional automakers is almost “like slicing vegetables.”
Honda’s gross profit was said to have fallen about 50%, Porsche’s to zero—an implied 100% decline—and Mercedes’ to have fallen about 69% in the first 3 quarters. Traditional CEOs also find it difficult to cut capacity and revenue guidance proactively. The speaker’s clear stock view is: “Do not touch autos or automaker stocks under any circumstances.” He is especially bearish on STMicroelectronics, but this does not imply that the entire market is in trouble.