The Rhyme: Warsh Holds Rates as Big Tech Bets the Farm & Its 2006 Echo

"History doesn't repeat… but it rhymes." — Mark Twain

At 2:00 p.m. Eastern today, Kevin Warsh is expected to hold the federal funds rate at 3.50%–3.75% for the fifth straight meeting. Hours later, Microsoft and Meta report quarterly earnings with a combined $148 billion in expected revenue and the only question that matters: whether $725 billion in collective AI infrastructure spending is paying off. CPI dropped to 3.5% in June from 4.2% in May, but half the Fed's own members still see a rate hike before December. The last time a new Fed chair held rates while a spending bubble swelled in the background was August 8, 2006. Kevin Warsh was in the room.

S&P 500: 7,413 | Fed funds: 3.50%–3.75% | CPI: 3.5% (Jun) | WTI crude: ~$79/bbl | MSFT exp rev: $87.7B | META exp rev: $60.2B | Hyperscaler AI capex 2026E: $725B (+77% YoY)

The meeting started at 8:30 in the morning in the Board Room on Constitution Avenue, twenty-foot ceilings, the long mahogany table, the quiet hum of the air system that never changes. Ben Bernanke sat at the head of it for only the fourth time as chairman. He had taken the gavel from Alan Greenspan six months earlier, inheriting an interest-rate machine that had been cranking in one direction for two straight years.

Seventeen consecutive hikes, starting at 1% in June 2004, pushing all the way to 5.25% at the previous meeting on June 29. Every one a quarter point, every one justified by an economy that kept running hot. Housing prices had doubled since 2000. Oil was trading near $77 a barrel, the highest in history. Core PCE inflation had crept to 2.4%. Nobody at the table thought the problem was solved. But Bernanke wanted to stop and look around. Jeffrey Lacker, the Richmond Fed president, wanted an eighteenth hike. The rest sided with the new chairman. The vote was nine to one.

Sitting along the side of that table, five months into his job as the youngest governor on the Board at age thirty-six, was a former Morgan Stanley banker named Kevin Warsh. He voted with Bernanke to pause. The S&P 500 closed around 1,271 that day and started climbing, eventually running 23% higher to 1,565 by October 2007 as investors decided the pause meant a soft landing. In the background, subprime lenders were writing mortgages with no documentation. Home prices peaked that summer and quietly started falling. The pause that looked like prudence turned out to be a prologue.

The same man who voted to pause rates in 2006 is now the one deciding whether to pause them in 2026, with a similar oil price, a comparable inflation rate, and a nearly identical question about whether the biggest spending boom of the era has gotten out of hand.

The Fed held at 5.25% for the rest of 2006. It held through the winter and spring of 2007. Home prices fell, quietly at first, then faster, but the stock market did not care. The S&P 500 rallied from 1,271 to 1,400 by year-end, then kept going past 1,500 in the summer of 2007. Investors told themselves the pause meant a soft landing. Money flowed into equities for fourteen months straight.

Then on August 9, 2007, almost exactly a year after the pause, BNP Paribas froze three investment funds because they could not value their subprime holdings. The credit markets seized. Bernanke cut by 50 basis points on September 18, the first cut in four years and a surprise half-point move that told everyone the Fed was scared. The S&P peaked at 1,565 on October 9. From there it fell 57%, bottoming at 676 on March 9, 2009.

The rhyme is in the structure: a new chair holding rates while an asset class inflates in the background, a market that reads the hold as permission to keep buying, and a risk nobody at the table wants to quantify. In 2006 it was hidden leverage in housing. In 2026 it is $725 billion in AI capex chasing revenue streams that have not materialized at scale. Meta's operating margins are narrowing even as revenue climbs 27%. The gap between what these companies are spending and what they are earning from it is the number that matters tonight.

The smart money in 2006 did not sell the day of the pause. They sold fourteen months later, when the surprise cut confirmed the Fed had been too patient. If Warsh holds today and strips the statement bare, the question is not what happens tomorrow. It is what happens when he finally moves, and whether it comes too late.

The edge: After Bernanke's August 2006 pause, the S&P rallied 23% over fourteen months before the credit crisis hit. The window between a Fed hold and a Fed panic cut was the last great buying opportunity of that cycle, and also the last chance to get out. Tonight's earnings from Microsoft and Meta will not settle the AI capex question. The September FOMC meeting will. Watch whether hike odds rise above 50% after Warsh's press conference. That is the number that turns this from a rhyme into a replay.

Apple and Amazon report Thursday after close, completing the $725 billion capex picture. If both raise spending guidance the way Meta and Microsoft have, the combined number starts looking less like an investment cycle and more like the telecom fiber-optic buildout of 1996–2000, when $500 billion in cable got laid and 95% of it went dark for a decade.