Logo
Home
About Us
All Publications
Sign Up
Logo
  • Home
  • Posts
  • The Rhyme: the Fed Puts AI on Trial & Its 1997 Echo

The Rhyme: the Fed Puts AI on Trial & Its 1997 Echo

Warsh opens his forum today with a December hike already fully priced. In 1997, Greenspan faced the same question about a different technology — and his answer sent the S&P up 33% before it fell 49%.

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

◉ THE PRESENT

At noon Eastern today, Fed Chair Kevin Warsh opens a public forum at the San Francisco Fed built around a single question: is artificial intelligence making inflation worse, or is it about to make it better? The answer matters because markets are pricing in a 100% chance of a December rate hike, with a one-in-seven shot it comes as a double. If the forum concludes that the $800 billion AI buildout is just burning through resources with nothing to show yet, the hawks win and rates keep climbing.

S&P 500: 7,561  |  10-yr yield: 5.15%  |  Fed funds: 3.75–4.00%  |  Core CPI (3-mo ann.): 3.7%  |  NVDA: $162.85 (−30% from ATH)  |  Dec hike odds: 100%

Twenty-nine years ago, another Fed chair stood at a lectern and wrestled with the same question about a different technology. His answer shaped the next three years of American markets.

◉ THE ECHO — AUGUST 25, 1987

The Chairman Who Trusted the Machines

The hearing room on the second floor of the Dirksen Senate Office Building filled up early that Wednesday morning. Alan Greenspan, seventy years old, wire-rimmed glasses, took his seat before the Senate Banking Committee for the semiannual Humphrey-Hawkins testimony. The Dow had punched through 7,000 for the first time thirteen days earlier. The S&P 500 sat at 812. Unemployment was 5.3%. And the question keeping every fund manager awake was whether the economy had gotten so good that inflation had to follow, and whether Greenspan would kill the rally to stop it.

The standard playbook said yes. Growth above 3%, unemployment falling toward 5%, wages starting to tick higher. The Fed's own staff wanted a rate hike. A substantial minority of the FOMC wanted to tighten right there. But Greenspan had a different theory. He believed that millions of personal computers flooding into offices and warehouses, connected by a commercial internet barely three years old, were expanding the productive capacity of the entire economy. Workers weren't demanding big raises because they feared technology would replace them. Companies weren't raising prices because competitors armed with the same new tools would undercut them. He told the senators that technological change had been "an important impetus behind corporate restructuring" and that inflation could stay low even as the economy ran hot.

The S&P fell six points that afternoon to 805.68. But the deeper message mattered more than the day's tape. One month later, on March 25th, the Fed raised the funds rate a quarter point to 5.50%. Then Greenspan stopped. He held at 5.50% for eighteen months while the S&P climbed 33% through the end of 1997 and crossed 1,000 by February 1998. His productivity bet was right about the economy. Revised data later showed that labor productivity had in fact accelerated just as he predicted. But the patience he showed also gave the market permission to go places no valuation could justify. The CAPE ratio climbed from 28 to 44 by December 1999. The S&P peaked at 1,527 in March 2000 and then lost 49% over the next two and a half years.

◉ THE RHYME — WHAT'S IDENTICAL

Both moments come down to the same wager: does the transformative technology of the era suppress inflation by boosting productivity, or does it fuel inflation by consuming enormous resources before the gains arrive?

◉ THE DIVERGENCE — WHAT'S DIFFERENT THIS TIME
  1. The Fed chair is leaning the other way. Greenspan sided with the optimists and held rates to let the productivity story breathe. Warsh hasn't shown his cards yet, but his colleagues have. Fed Governor Lisa Cook warned in September that AI investment was adding inflationary pressure. Fed Governor Michael Barr warned last Thursday of "serious short-term disruptions" from AI replacing human tasks. The mood inside the Eccles Building in 2026 treats AI as a cost before it becomes a benefit. In 1997, the mood was the opposite.

  2. The spending is four times larger. Hyperscalers are on track for roughly $800 billion in AI infrastructure this year, with JP Morgan projecting a trillion in 2027. At the peak of the 1990s telecom buildout, total capex hit about $120 billion in 2000 dollars. That much demand for chips, electricity, and skilled labor puts real pressure on input costs, which is the concern driving the hawks right now.

  3. Inflation is already running hot. Core CPI in February 1997 was drifting lower at 2.5%. Greenspan had room to be patient. In November 2026, core CPI has printed 0.3% monthly for three straight months, running at a 3.7% annualized pace. Annual PPI just hit 6.5%. Warsh doesn't have the luxury of waiting to see if AI bends the cost curve because prices are climbing right now.

◉ THE RECKONING — WHAT HAPPENS NEXT

Here is what happened after Greenspan placed his bet. On March 25, 1997, the Fed raised the funds rate to 5.50%. The market dipped for a week and then kept climbing. The S&P 500 finished that year at roughly 970, up 33%. Greenspan held at 5.50% through the Asian crisis, all the way until the Russian default and the collapse of Long-Term Capital Management forced three cuts in the fall of 1998. By the end of 1999 the S&P sat at 1,469. It peaked at 1,527 in March 2000, then lost nearly half its value over the next thirty months.

Greenspan's choice not to tighten aggressively in early 1997 produced the greatest wealth creation of the twentieth century and then the most brutal tech crash since the invention of the transistor. The productivity gains he believed in were real. But the patience he showed gave the stock market permission to treat every company with a website like a license to print money.

Warsh now sits in that same chair with the same question on his desk, just wearing different clothes. If his forum concludes that AI is mostly cost and heat right now, the Fed keeps hiking. That is bearish for stocks in the near term but it also means the bubble doesn't get the oxygen Greenspan gave his. If Warsh starts talking about productivity and patience, the market will hear what it heard in 1997: the chairman trusts the technology. After the March 1997 hike, the S&P was 12.4% higher in three months and 39.3% higher within a year.

In 1997, the right trade was to buy the single rate hike and hold for the pause. The eighteen months after Greenspan's March 25th move produced a 32% gain in the S&P 500. Watch Warsh's language today. If you hear patience, the 1997 playbook is in play. If you hear concern, the staircase has more steps and the landing is farther down.

◉ TOMORROW’S WATCH

Nvidia reports earnings after the bell Tuesday, its first call without Jensen Huang. On Thursday, Anthropic prices its IPO at $70-$78 a share, valuing the AI lab at up to $1.4 trillion. The last time a company went public at that scale while the Fed was actively tightening was Goldman Sachs in May 1999, less than two months before Greenspan's first pre-crash rate hike.

Recommended for you

View All Publications
caret-right

Top Story Daily

"History doesn't repeat… but it rhymes."

Mark Twain

Quick Links

Home

About Us

All Publications

Subscription

© 2026 Top Story Daily