Market Overview November 25, 2025
Summary
The market remains in a “vacuum period” in the short term, lacking policy and earnings catalysts, but a looser liquidity backdrop will inflate a much larger AI bubble in 2026. The current surge comes with high volatility, and “the rally is not particularly healthy”; those who missed Friday’s rebound do not need to chase it. AI demand is strong, supply is constrained, and most of the companies that have rallied sharply are still being driven by earnings. “The bubble hasn’t started yet; next year there will be a fairly large bubble.”
Whether rates are cut in December is merely a tactical variable; strategically, “The train has left the station,” and the Fed is already on an irreversible path toward easier policy. The speaker sees a December cut as the most likely outcome, but the most market-friendly scenario may be no cut accompanied by a signal that cuts are coming later. Even if rates are constrained, the Fed can still provide liquidity through balance-sheet expansion, regulatory easing, or other “strange” tools.
Inflation limits the room for conventional rate cuts, but a deteriorating labor market is forcing the Fed to prioritize employment and growth. Goldman estimates tariffs have added about 53 basis points to inflation, keeping it around 2.8%-3%; real rates cannot easily turn negative. Meanwhile, employment for college graduates is approaching the weak levels of 2018, and Layoff Tracker, a coincident indicator, has risen rapidly—earlier than initial jobless claims, payrolls, and the unemployment rate. “Between growth and inflation, it will choose growth.”
Positioning does not support a replay of the April liquidation, so the pullback can continue but is less likely to turn into a deep forced unwind. Selling vol, selling gamma, and credit longs were not extremely crowded before this decline, and vol rising to around 25 does not mean a crash is imminent. CTA funds could still sell roughly $13B this week, but their positioning is two-sided; “there isn’t much kindling piled up in the market.”
Google and OpenAI are unlikely to see one side completely overwhelm the other, while Nvidia may offer better short-term risk-reward after being left out of the spotlight. The speaker compares the two camps to “the US and the Soviet Union,” with the final outcome still constrained by TSMC’s allocation decisions. Gemini 3 and potential Google TPU purchases by Meta may keep Google in focus, but Nvidia’s order visibility, continued shortages, and roughly 75% margins next year remain hard support. “Everyone will win in the end; the losers will be most people’s jobs and investors who don’t believe in AI.”
Bitcoin’s pressure this cycle looks more like a technical shock from MSTR index trading, while the US-China thaw should be treated only as a temporary truce. The market expects Strategy could be removed from MSCI in February 2026, prompting shorts to position early and weighing on BTC. If liquidity is abundant in 2026, the speaker sees little difference between buying BTC at $70,000, $75,000, $80,000, or $85,000, though AI and semiconductors remain the preferred trades. H200 sales to China and reports of reciprocal visits briefly lifted sentiment, but “every time they reach an agreement, it is a ceasefire, it is a truce.”
Deep dive
1. The Vacuum Period Allows a Pullback; the Real AI Bubble Has Yet to Begin
The market currently lacks sustained monetary easing, the rollout of tax refunds, and a fresh round of earnings validation. Nvidia has just delivered strong results, while the next reports from companies such as Microsoft are still two or three months away, leaving the market temporarily with “no catalyst” and giving shorts more incentive to press.
The speaker did not endorse the rapid rebound as a healthy trend: “When the rally comes this fast, vol is actually very high.” Those who missed Friday’s and the weekend’s moves do not need to rush; the vacuum period should still provide better entry points.
The outlook for 2026 is materially stronger. AI companies are largely funding compute investment from their own balance sheets, returns on investment are high, demand is “very, very strong,” and supply is severely constrained. Even Meta, which he considers relatively behind within the AI camp, may use Google TPU. “The bubble hasn’t started yet; next year there will be a fairly large bubble.”
2. The December Decision Is Noise; the Easing Train Has Left the Station
After the previous “hawkish cut,” several Fed officials questioned another move in December, pushing the market-implied probability of a cut down to 30%. New York Fed President Williams then signaled support for a cut, as if telling a crying market: “Don’t cry; there’s milk.”
The speaker’s metaphor is “The train has left the station.” Regardless of who becomes the next Fed chairman or how many dissents emerge in December, the broad direction of monetary policy in 2026 will be accommodative. The only question is whether the Fed cuts rates or looks for alternative liquidity tools.
He still leans toward a December cut. But at the December 10 decision, the December 16 economic data and December 18 inflation data will not yet have been released. “Do you dare not cut? Maybe not. If you don’t cut, the market will immediately die for you to see.” The more ideal combination would be a dovish pause: no cut this time, combined with a clear signal that cuts are coming later.
3. Tariffs Close the Door on the Rate Floor; Weak Employment Opens a Policy Back Door
Goldman estimates tariffs have added about 53 basis points to inflation, putting year-end inflation at 2.8%-2.9%, broadly around 3%. With equities also near all-time highs, the odds of cutting rates dramatically lower and eventually reaching zero interest rates are “very, very low,” while real rates cannot turn negative either.
The other side of the equation is the “K-shape.” The speaker summarizes the AI productivity revolution as “1% of people profit, 99% get washed out,” arguing that it will eliminate existing productivity jobs, including insurance sales, doctors, lawyers, and accountants. Students—especially those with bachelor’s degrees—will be displaced first.
ADP, initial jobless claims, payrolls, and the unemployment rate may all lag because companies stop hiring first, lay off workers next, and only then do the unemployed apply for benefits. Layoff Tracker, which has already risen rapidly, is a coincident indicator and will reflect changes in layoffs earlier than those data series. The speaker is “fairly pessimistic” about US data in the first quarter of 2026: even if there is no recession, it “won’t be far from one,” and the numbers will be ugly.
This defines the Fed’s decision function. Employment is its core consideration, and “ten times out of ten, it will choose employment.” Rising stocks do not necessarily create jobs, while AI may reduce employment; falling stocks, however, inevitably inflict further damage on jobs. The room for policy error is therefore very limited.
4. Private-Sector Transformation Requires Balance-Sheet Expansion; SLR Easing Could Release Liquidity First
The speaker characterizes the Republican direction as a return to a “private-sector-driven economy”: shifting away from pandemic subsidies, healthcare and Social Security spending, and short-term stimulus toward infrastructure investment and tax refunds, returning resources to the private sector. Combined with AI-driven job losses, this shift requires the Fed to backstop the economy with easier financial conditions.
Conventional rate cuts are constrained by inflation. An alternative could be regulatory easing through SLR, eSLR, and similar measures, allowing banks to expand their balance sheets. His estimate is that the changes could add about $200B in capital and release it gradually over a year, roughly between February and April 2026 based on his timeline.
The mechanism is straightforward: after 2008, regulation “tied your hands and feet.” In the past, the Fed could keep expanding the balance sheet; now, if it cannot cut rates sharply, it can “untie your hands and feet” and let private-sector banks inject liquidity into the market.
Liquidity cannot be assessed solely through TGA or reverse repo; it must also be viewed alongside indicators such as the collateralized lending rate, like a doctor combining multiple vital signs. Weakening employment or intensifying financial stress could force the Fed to expand its balance sheet. Data from Germany, Japan, and China are also weak; the speaker’s global conclusion is that “everyone’s consensus is to print money.”
5. The Trigger for a Pullback Can Be Anything, but Positioning Determines How Far It Falls
In April, equity, vol-control, short-vol, gamma, and CDS positions were nearly crowded in the same direction, so unwinding naturally created a chain reaction. Some of the longs in the credit market from the previous September have already been forced out, while short-vol and short-gamma positions are also less extreme.
Vol peaked around 25 this time, but that does not automatically mean a crash. Vol was already elevated at the start of the decline, and investment banks did not quickly fall into the kind of massive negative gamma seen previously, leaving insufficient fuel for a violent gamma squeeze.
CTA funds could still generate roughly $13B of selling this week, but their positioning is not a fully invested one-way long. It is two-sided: they buy into strength and sell into weakness. The speaker’s conclusion is that “the reason can be anything”; what determines the depth of the decline is how much kindling has accumulated. Positioning supports an adjustment, but not an April-style collapse.
6. Google Wins the Spotlight; Nvidia Wins on Short-Term Risk-Reward
The speaker has held Google for the long term and once said around $170 that “you could buy it and sleep well.” He also identified early on that its compute costs were lower than OpenAI’s. Recognition for Gemini 3 and the possibility that Meta may purchase Google TPU have simply led the market to concentrate on that trade.
Google and OpenAI are more like “the US and the Soviet Union”; neither will completely overwhelm the other, and TSMC is behind both of them allocating supply. TSMC ultimately decides who gets what, while Google is one of its true major buyers, so neither side’s position will be completely displaced.
Nvidia’s earnings and certainty around 2026 deliveries have not disappeared just because TPU has gained attention. Its products remain in persistent shortage, and what most changed the speaker’s view was the company’s statement that next year’s margin could still hold at roughly 75%. He had previously believed that this gross margin would eventually decline.
The short-term strategy is to “trade the cold, not the hot.” Google is suitable for long-term ownership, but once it becomes a crowded trade, Nvidia may offer better risk-reward. Likewise, when memory becomes hot, it may be better not to keep trading YSD and instead look for colder areas such as optical modules.
7. MSTR Explains BTC’s Weakness; US-China Cooperation Is Only a Temporary Truce
The speaker believes an important driver of BTC’s decline is the possibility that Strategy (MSTR) will be removed from MSCI. Once the market learned of the prospect, shorts positioned through an index trade. If the actual adjustment occurs in February 2026, BTC could instead rise as shorts cover. “Whoever gets the information first makes the money.”
Over the weekend, investors bought BTC to hedge MSTR shorts out of concern that US stocks would rise on Monday. But even after US equities rebounded by more than 2%, MSTR’s rebound was still below his expectation. Short interest and long-short positioning require further observation. Once the removal risk is widely known, he believes it no longer needs to be given special weight in assessing its impact on the bitcoin market.
If 2026 is indeed a high-liquidity environment, the precise BTC entry point between $70,000 and $85,000 makes “no particularly big difference.” This is not to say BTC is the strongest asset; the speaker’s ranking remains clear: “The best choice is definitely AI—definitely semiconductors.”
The possibility of H200 sales to China, along with reports that Trump may visit China and Xi Jinping may visit the US, briefly lifted sentiment but also pressured overseas compute-center companies. The White House is simultaneously preparing to address the litigation risk around IEEPA tariffs through routes including Section 301 and Section 122.
The speaker summarizes the backdrop as follows: the US leads in technology, but China has completely surpassed it in manufacturing, so the US has chosen to maintain its lead in AI and technology and go “all in” on AI. He does not believe US-China competition will end after one meeting. “Next year will definitely see a large bubble inflate”; any easing in tensions is only a truce.