The Last Week of the World Powell Built
Friday, May 15. Circle it. Powell's out. Warsh is in.
Eight years of one man's hand on the most powerful monetary lever in human history, and it ends not with a crisis, not with a triumph, but with a calendar date and a Senate committee vote that went 13-11. The narrowest margin in history for a Fed Chair confirmation. Thirteen to eleven. On the eve of a hot inflation print. With Brent crude above $100 and the Strait of Hormuz running at partial capacity. This is the context in which the United States hands the keys to a Hoover Institution fellow who thinks the dot plot is a crime against intellectual honesty.
You could not write it. You'd be told it was too on the nose.
Let's slow down on Warsh for a moment, because the market is pricing him as a binary — either a Trump puppet who cuts into a 3.8% CPI print and blows up the Fed's credibility for a generation, or a closet hawk who wraps his inflation-fighting instincts in AI-disinflationary rhetoric and basically continues Powell's hold with better branding. Neither picture is complete and both are dangerous to trade on.
Here is what is actually known: Warsh wants to shrink the balance sheet. The Fed's balance sheet currently sits near $6.7 trillion. Before the financial crisis it was $800 billion. That gap — the chasm between those two numbers — is the accumulated legacy of every QE program, every emergency facility, every "whatever it takes" moment since 2008. Warsh has called this "unhelpful." He wants it gone. Or at least much smaller.
The bond market knows what that means. Fewer Fed buyers of Treasuries means more supply relative to demand means higher yields. The 30-year has already crept above 5% — the "Warsh trade," some are calling it — and if the balance sheet rundown accelerates, the move isn't over. Equity risk premiums don't get richer at 5%-handle long bonds. The S&P at 20.9 times forward earnings is a valuation built for a world with $6.7 trillion of central bank support behind it. Remove the scaffolding and the building may still stand. Or it might not. Anyone telling you confidently which outcome occurs is selling something.
Then there's the other thing. The strange, almost philosophical position Warsh has staked out: that AI is a disinflationary force powerful enough to justify looser policy even when prices are running hot. He said as much in his confirmation hearing. The tech industry absorbed this like manna. Nvidia touched an all-time intraday high on Tuesday morning, briefly, before the CPI hit and the chip complex fell apart with an almost theatrical synchronicity.
Qualcomm down 12%. Worst session since 2020. Intel off 9%. AMD and Micron bleeding mid-single digits. The SOX dropping 5% on the day. This after a month in which the index had gained 29%. After Intel, of all companies, had run 430% in a year. When the reversal comes in a market like this, it doesn't arrive with a warning. It shows up in the middle of a quiet Tuesday morning alongside a CPI report and a war update and just starts erasing numbers.
The irony is that the SOX is still up 60% year-to-date. What happened Tuesday was a tremor. The question — the one nobody can fully answer — is whether the "AI Class of 2026" trading at roughly 39 times earnings is a justified premium or a late-cycle hallucination. Dot-com devotees were paying 152 times. Bulls find comfort in that comparison. Bears note that 39 times is still almost double the historical S&P average multiple. The argument goes on. The chips go up. The chips go down. The data centers keep getting built.
Somewhere in all this noise, the actual consumer is doing considerably less well than the market's mood would suggest.
For the first time in three years, wage growth is losing the race to inflation. Gas at $4.50 nationally. Beef up 14.8% year over year. Airline fares up 20.7%. The University of Michigan survey now shows households expecting 4.5% inflation over the next year — a number that, if it embeds further into wage negotiations, becomes the Fed's next problem on top of its current ones. Consumer sentiment has hit all-time lows while the S&P 500 sits near all-time highs. These two facts coexist because the stock market is not the economy, and the economy is not the stock market, and 2026 is making that point more brutally than any economics textbook.
Powell, at his last FOMC, sat with four dissents — the most fractious committee since 1992. He held rates and said nothing particularly memorable. That's his legacy in miniature: a man who walked tightropes for eight years without falling, who sometimes moved too slowly and sometimes too fast, who was never beloved by anyone but was trusted by enough people to keep the system functional. History will probably be kinder to him than the room was on Tuesday.
Warsh's first press conference will come this week. Every word will be parsed for signals, syllables mapped onto rate trajectories, adjectives interrogated for hidden hawkishness. "Regime change" was his phrase from the confirmation hearings — not a term you deploy lightly when you're about to chair an institution whose primary asset is institutional credibility. Either it was a performance for Trump, or he means it. If he means it, the bond market is not done moving.
Barclays data shows the S&P 500 averages drawdowns of 5%, 12%, and 16% in the one, three, and six months following a new Fed Chair taking over — going back to 1930. That's an average. Averages mean nothing in any individual instance. But the number exists, and it exists because transitions are windows of genuine uncertainty, not just political theatre.
GameStop tried to buy eBay this week. eBay's board sent back a letter calling the proposal neither "credible nor attractive." Ryan Cohen, who turned a dying video game retailer into something that still somehow commands institutional attention, got publicly rejected by a digital auction platform. This is also a data point, in its way: a symbol of how strange and crowded and restless capital has become when it can't find a clean directional bet.
Everyone is waiting. For the Strait to open or not open. For Warsh to signal or not signal. For May CPI to either confirm or break the inflation trajectory. For Cerebras to price and tell us whether the IPO window is still wide open or beginning to close.
Friday is very close. The world Powell built is almost over. Whatever Warsh builds in its place starts now.