Summary

US equities have entered a post-cut “vacuum,” and the speaker sees volatility as an entry window rather than the start of a bear market. The government shutdown created a data gap, BBB tax refunds were delayed, and fiscal, monetary and earnings catalysts are temporarily absent. Buyback flows will also largely end after December 22, leaving you with “no cards in your hand.” The pullback may last into January, but his call for tech-led gains and a US bull market in 2026 is unchanged.
Future data must deliver an almost impossibly tight Goldilocks outcome, with any deviation capable of triggering a fragile market. November nonfarm payrolls are expected to be weak, and December, January and even February data could deteriorate further. But “data starting to get worse” could signal another leg down, while “data getting very bad” would mean the market is nearing a bottom, because prices bottom well before the data.
Powell’s concern about employment leaves the door open to another January cut, and the Fed in 2026 could be more dovish than Wall Street is pricing. He said current data may understate job creation by 60,000 per month and urged preparation for net employment to turn negative—effectively leaving room for a January cut. Whether or not Kevin Hassett is nominated, the lagged effects of tariffs and the government shutdown, along with AI-driven productivity gains and related job losses, could push 2026 monetary policy in a dovish direction.
Short positioning in Oracle and Nvidia has become excessively crowded, and any fundamental catalyst from OpenAI could trigger a “very violent” rebound. Oracle CDS has approached 2008 levels, although this is company-specific risk rather than a systemic crisis. The speaker’s explicit view is that “Oracle is not going bankrupt.” Oracle’s backend demand is stable, and OpenAI’s B2B earnings have surpassed Palantir.
Nvidia is his highest-conviction name for 2026, but that conviction does not extend into 2027. From tokens and chips to wafers and substrates, the entire chain is “sold out from top to bottom,” making conventional margin-based competition arguments less relevant. With land, power and P/W constrained, data centers will continue to pursue the most efficient solution.
China and Japan are each trapped by the historical trauma of inflation and deflation, producing structural biases in their policy choices. China faces severe deflation driven by demographics, overcapacity, debt and the collapse of its property bubble, yet still fears losing control of stimulus. Japan, shaped by decades of deflation, is unlikely to hike aggressively in succession. Some positioning is already reflected ahead of the Bank of Japan’s rate decision on December 18 or 19; the speaker later argues that Japan still needs to hike, but that global carry trades may not face a full unwind.

Deep dive

1. The Post-Cut Vacuum Creates Volatility—and Entry Points

The speaker’s base case is unchanged: the macro backdrop in 2026 should remain bullish, with the US market likely led by technology stocks in particular. Token demand and the investment cycle ahead have not yet revealed any “big problems”; a pullback over the next month does not amount to a reversal of the long-term thesis.
The vacuum reflects three missing supports. With BBB tax refunds delayed, the fiscal-policy handoff is unclear; after the rate cuts, monetary policy has no new card to play; and corporate earnings have yet to arrive, leaving current disputes untested. Once buybacks largely end after December 22, the market will also lose its source of “forced money” buying.
He did not call for an inevitable sharp selloff, only emphasized that the market is unstable. Shorts can attack the Fed, AI and mega-cap technology precisely because there is no fresh information in the near term to rebut them. But as annual reports approach, it will be difficult for large short positions to remain open indefinitely.

2. The Worse the Data Gets, the Closer the Market May Be to a Bottom

Nonfarm payrolls and CPI are the most watched data releases this week after the government shutdown. Until now, the market has relied mainly on ADP, which the speaker considers “not as accurate as nonfarm payrolls.” November employment is likely to be poor, and “November is definitely not the bottom.”
His timing framework is worth preserving: “When the data starts to get worse, that could be a signal for the market to move lower; but when the data gets very bad, the market is probably close to the bottom.” December, January and even February could remain weak, but that does not mean equities must keep falling until the data bottoms.
The market is demanding an extremely narrow Goldilocks outcome: “Whether the data is good or bad, the market will be very fragile.” Only a result that is neither too hot nor too cold is likely to be accepted. Any deviation in employment or inflation could make the market highly sensitive.
Despite dissent on both sides, Powell’s remarks remained relatively soft. Existing data may understate monthly job creation by 60,000, and employment could eventually turn negative. He did not rule out a January cut, and the policy focus remained clearly centered on employment.

3. Wall Street Is Underestimating the Fed’s Dovish Resolve

The market sees roughly 6 votes for holding rates steady. The 2026 dot plot’s median rate did not change, but the depth of internal dissent has led investors to price the Fed as having “no ammunition.” The speaker’s contrarian view is that “everyone is underestimating the Fed’s dovish resolve.” Even with a new chair, the Fed would not cut rates to zero as Trump wants, but it would likely remain dovish.
The first reason is the lagged impact of policy shocks. The transmission of tariffs and the government shutdown into the economy and employment will not end immediately; the effects are more likely to emerge in the first quarter of 2026. Second, a sharp increase in AI-driven productivity will itself eliminate some jobs, adding pressure for easier policy.
On reserve management, the speaker cited differing figures at different points—four hundred billion, four billion and 40 billion per month. His later explanation was that the Fed’s roughly $40B monthly operations can stabilize short-term liquidity and be increased if necessary. Banks can borrow short and hold two-year, five-year and longer-duration bonds, expanding their balance sheets and leveraging up to transmit liquidity into the long end without counting it toward the SLR.
Fiscal spending and long-duration debt supply mean the 10-year yield will not fall too low, while bank balance-sheet expansion means it “probably won’t get too high.” The arrangement began a trial run on December 1 and, in the speaker’s recollection, should enter a more definitive phase around April 1 of the following year. The result is an easier overall policy backdrop for 2026, even if the next 2 weeks do not feel easy.

4. Oracle’s Credit Panic Could Become a Short Trap

Wall Street is worried that Oracle’s capex is too large and that OpenAI’s ability to keep paying and sustaining its investment is uncertain. Investors have consequently crowded into Oracle shorts and CDS trades. Its CDS has approached 2008 levels, but the speaker cautions that 2008 was a market-wide crisis; this is primarily company-specific pricing.
His view leaves no room for ambiguity: “Oracle is not going bankrupt.” Oracle’s backend demand is stable, and OpenAI’s B2B earnings have surpassed Palantir. On OpenAI, he also said Hugging Face was approaching $1T (“if I remember correctly”) while OpenAI was at $500B. For the consumer business, he repeated his earlier estimate of monthly maintenance costs at roughly $1.5, arguing that neither the revenue nor the loss side is large enough to support a bankruptcy narrative.
The key to the trade is not immediately proving every criticism wrong, but the concentration of positioning. A better OpenAI model, stronger B2B or consumer revenue figures, or simply a recovery in confidence could rapidly reverse the pricing. The more concentrated the positioning, the more dangerous the shorts become.
He used Intel as a comparison. The industry already knew that 18A and 18A-P were sound, yet the stock still fell to $18 and traded as though bankruptcy were imminent. Once earnings stopped deteriorating, long-standing shorts reversed. Oracle does not need to go through the lengthy validation process required of a foundry, so its catalyst cycle could be shorter.

5. Nvidia’s Supply Constraints Outweigh Conventional Margin Concerns

The speaker called Nvidia at least the highest-certainty investment for next year, while adding an explicit time limit: “There is indeed very little certainty the year after next.” This is not a claim of a permanent moat, but a view on next year’s supply, demand and delivery visibility.
He pushed back on the narrative that “GPUs are cheap.” Their lower price may fundamentally come at the expense of P/W. Once the roughly 50% margin attributed to “Alphago” is adjusted for front-end manufacturing and the cost of sending wafers to TSMC, the economics are not as far apart as they appear.
Land and approved power are both scarce resources, so compute centers will ultimately optimize for efficiency per unit of site capacity. He therefore still sees Nvidia’s chips as the most efficient option. Meta appears to want to buy roughly 200,000 TPUs, but actual delivery will take time; OpenAI could develop its own ASICs, but that “isn’t happening overnight.”
Another challenge to Meta Seven is margin. Over the past 3 years, revenue rose about 100% while margin increased roughly 10%. The market accepts that revenue will continue to grow but questions whether margins can rise further. His answer is that this is not a normal competitive market: tokens, chips, wafers and substrates are all in shortage. Nvidia’s margin should increase next year, and the semiconductor sector is at least likely to perform well.

6. China and Japan Represent Two Sides of Historical Fear

The speaker cited “adapting policy to local conditions” and “real, substantive growth without water,” interpreting them as a push to reduce local-government construction starts and hollow projects. The problem is that tightening government spending would make China’s already severe deflation even harder to control.
He had previously used Chernobyl as an analogy: to lift inflation, “they removed all the control rods,” yet deflation remained unresolved and could suddenly run out of control once the reaction finally began. The policy system also struggles to achieve Goldilocks, swinging between excessive liquidity provision and shutting down the mechanism entirely.
The three roots of deflation are demographics, overcapacity, and massive debt combined with the collapse of the property bubble. Rural social security and targeted consumption policies are merely “a drop in the bucket.” In his observation, the monetary base M0 has nearly doubled since last September without materially changing the underlying deflation.
Japan developed the opposite bias after living with deflation since the 1990s. The speaker’s comments on a rate cut versus a hike around the December 18 or 19 decision were inconsistent at different points; later, he argued that Japan still needs to hike, but is unlikely to suppress inflation through consecutive, aggressive moves. Some of the rate action and its impact on carry trades have already been priced into the earlier selloff. He also said Japanese government debt has climbed above 300% of GDP and toward 400%, making Japan a major global source of carry trades.

Finding The 1% of Stocks That Matter | Henry Ellenbogen Interview When Do We Buy, Global M2, Does Crypto Need A Catalyst, & Owning Your L’s