Remarks at the iConnections TomorrowX Summit, published November 18, 2026
Stan Druckenmiller said at the iConnections TomorrowX Summit that he has largely exited the portfolio he described nine months ago, and that he now has less clarity on the 12-month outlook than at any point in a 50-year career. Asked how much of his January positioning he still holds, he answered: "Basically almost none." He said the firm is in capital preservation mode, trying "to not lose money as opposed to make money," and that the pause "could last two days or six months. I don't know."
A January Setup That Has Flipped on Every Variable
Druckenmiller said the January backdrop was one of unusual clarity: a strong economy, Fed funds at 3.5%, nominal growth of 6%, and what he called "ridiculously loose" monetary policy. The Fed dot plot pointed to more cuts, commodities including copper were breaking out, the advance-decline line was making new highs, and credit spreads were very tight. On top of that, Claude Code had energized AI insiders, which made him bullish on capital spending. A recession within 6 to 12 months seemed nearly impossible to him.
He said a war then began about a month later. He admitted he "panicked" and sold a bunch of positions, but described the outcome as lucky because on buying back quickly it became "more apparent than ever" that AI infrastructure was headed for a complete liftoff, and the exercise cleaned out holdings that were not going to rise strongly.
Today, he said, nearly everything else has reversed. The economy remains strong, but the Fed is hiking rather than cutting following what he called a regime change, led by a figure who sat next to him for 15 years and who is unhappy about 15 years of free money and five years of above-target inflation. Druckenmiller said commodities look like they have topped, and the firm is now short copper. The advance-decline line peaked on August 14 even as the averages made their highs in October, and credit spreads are widening. "I think you're getting the joke here," he said. "Everything is the exact opposite of the way it was then."
The firm is neither long nor short equities. He said he likes to go short, but "I wouldn't dare go short the innovation and everything going on here," adding that much of that area is oversold and that positioning has been washed out tactically. The firm holds some small positions in the dollar, and its equity book is basically flat, with offsetting longs and shorts. He also flagged that cyclical stocks led defensives a year ago while defensives are leading now, and said he is "really alert to the fact that the second half of 27 could be economically disappointed," which he is considering in the context of interest rates.
Where He Would Look Next: AI Use Cases, and Labs Selling Tasks Rather Than Tokens
Druckenmiller said the next leg of any equity advance is more likely to come from AI use cases than from infrastructure. Infrastructure could be "a great tactical trade" given how washed out positions are, but he does not expect it to be the engine it was over the last four years.
He said the most interesting use case may be the AI labs themselves, because he does not believe they will remain in the token business for much longer. "My own belief is that they're not going to be releasing models, and if you have the number one intelligence in the world, you can start selling tasks instead of tokens and you can create your own businesses. Doesn't take a genius." He cited cybersecurity as an example: "Who's going to compete with these guys in cyber security? They created the problems. So now they're the ones that could have swarms of agents defending us from them." He also pointed to healthcare as an area with "all kinds of use cases." Of the idea, he said: "Selling tasks, not tokens. You heard it here first."
Asked for the AI bear thesis he respects most, he named politics and financing, and said both concern him.
How the Risk Process Works: P&L as the Signal
Druckenmiller described a risk framework that rejects standard tools. The firm does not use VAR or stress tests. He said that when he ran the Quantum Fund for Soros, banks would not lend because the fund did not use VAR, so he supplied it for them. His actual method is to watch whether his P&L is behaving and correlating as it should, and to cut back when it is not. "The P&L is my true signal." His reasoning is that in chaotic or bear periods, correlations break down, and he said some talented people learned this the hard way with shorts in software and AI infrastructure earlier this year....