# The Federal Reserve Has Officially Entered the AI Era: What Warsh’s Jackson Hole Speech Reveals About the Economy’s Next Phase
## A New Framework for an Old Institution
When Kevin Warsh stepped up to the podium at Jackson Hole for his first address as Federal Reserve Chair, most financial analysts braced for clues about interest rate decisions. He gave them almost nothing on that front. Instead, he delivered something arguably more consequential: a full-throated acknowledgment that artificial intelligence has fundamentally altered how the American economy operates—and that the Fed needs a new playbook to keep up.
Warsh used the phrase “hinge point in history” to describe the current moment, and it’s hard to argue with the weight of what he laid out. For years, the Fed has built its models around traditional factors of production—labor, capital, and land. Warsh explicitly confirmed that AI now belongs in that same category, calling it “potentially a new factor of production” that will reshape both economic output and the way monetary policy is formulated.
This isn’t a casual observation from a policymaker brushing up on tech trends. It’s a structural admission that the economic models the Fed has relied on for decades may no longer be sufficient. And that has sweeping implications for everything from rate decisions to how trillions of dollars in capital are allocated across the global economy.
## The Numbers Behind the Hype
Warsh didn’t rely on abstract optimism. He anchored his argument in hard figures that even skeptics would struggle to dismiss.
Business capital expenditures—what Warsh described as the “seed corn of future economic growth”—are climbing at their fastest pace since 2021, rising approximately 9% over the last four quarters. Strikingly, more than half of that growth is directly attributable to AI infrastructure buildout. But Warsh said the real metric he’s watching isn’t the spending level itself. It’s the “second derivative”—whether the pace of spending is still accelerating or beginning to decelerate.
That distinction matters enormously. A steady rate of capital investment can be absorbed by an economy gradually. An accelerating rate of investment is a different beast entirely—it signals a feedback loop where growing demand for AI capabilities drives more spending, which in turn drives more capacity and more demand. If that loop is real and sustained, it could reshape productivity metrics in ways no central banker has ever had to account for.
## The $100 Billion Token Economy
Perhaps the most jarring figure in Warsh’s speech was his breakdown of the token market. He noted that annualized token sales at the two leading AI laboratories now exceed $100 billion—a more than 500% increase from just one year prior.
Tokens, in this context, are the fundamental units of exchange in the AI economy. Every time a user queries a large language model, generates text, or runs an inference, tokens are consumed. Warsh’s decision to cite a specific revenue figure—rather than speaking in generalities about “the AI boom”—signals that the Fed now views token transactions as a measurable economic activity, not a niche curiosity confined to the technology sector.
This reframing is significant. By treating token revenue as a trackable line item in the national accounts, the Fed is implicitly acknowledging that AI-driven services are no longer peripheral to economic output. They are central to it.
## Rewriting Moore’s Law
Warsh also introduced a concept he described as a “hyper-Moore’s law,” claiming that AI capability is compounding even faster than the famous observation that computing power doubles roughly every two years. For a historically hawkish central banker to invoke such language is notable—it is the furthest he came to endorsing the idea that AI alone justifies the massive capital inflows flooding the sector.
He framed the acceleration this way: “Ever-expanding pools of capital are pouring into AI-related infrastructure of all sorts.” The implication is clear—the cycle of investment is self-reinforcing, and the returns on capability improvements are compounding at a rate that defies traditional economic forecasting.
## The Unanswered Questions
Warsh was refreshingly honest about the limits of what even the Fed can predict. He posed two questions he declined to answer: whether AI will drive a “significant, sustained rise in productivity across the economy” and whether “token usage” will prove “complementary or competitive to labor.”
A Fed task force dedicated to productivity and employment is studying both questions, but Warsh was blunt that its findings would have “no bearing on decisions we make in the current policy conjuncture.” That gap—a central bank admitting AI is macro-relevant while simultaneously lacking the analytical tools to integrate it into policy—is perhaps the most important disclosure of the entire speech. The Fed is making rate decisions today using frameworks built for an economy that no longer fully exists.
## Who Captures the Value?
Warsh also raised the question of economic distribution in the AI era. He wondered whether the surplus generated by AI will flow broadly across the global economy or concentrate in specific sectors and asset classes. “How much of the surplus goes to owners of scarce assets—AI labs, chipmakers, energy producers, and cloud providers?” he asked.
The real-world timing of his speech offered a stark illustration. Just days before Warsh took the stage, Nvidia reported record quarterly revenue of $96.2 billion and disclosed $366 billion in forward AI infrastructure commitments. The same period saw a move to acquire Hugging Face, the open-source AI community hub, for roughly $12.9 billion. Meanwhile, independent analyses from the previous year indicated that 95% of generative AI companies are failing.
The contrast is striking. A handful of dominant players are capturing an outsized share of the AI buildout’s value, while the broader ecosystem of startups struggles to survive. Warsh’s question about market structure already has a leading candidate for an answer.
## What This Means for the Economy Going Forward
Warsh’s speech effectively draws a line in the sand. The Fed recognizes AI as an economic force on par with the industrial revolution in its potential impact, but it has no established framework for incorporating that recognition into policy. The institution that once moved markets with a single quarter-point rate change now finds itself navigating an economic landscape where the most transformative force of the decade operates largely outside its traditional models.
For investors, businesses, and policymakers, the message is clear: the rules of the economic game are changing, and the pace of that change is accelerating faster than anyone predicted.
—
## Frequently Asked Questions
**Q: What did Kevin Warsh mean by calling AI a “new factor of production”?**
A: Traditionally, economists classify the inputs that drive economic output into three categories: labor, capital, and land. By describing AI as “potentially a new factor of production,” Warsh is saying that AI has become so integral to economic activity that it can no longer be treated as a subset of existing categories. It is now an independent driver of output that warrants its own analytical framework within the Fed’s policy models.
**Q: What is the “second derivative” of AI spending, and why does it matter?**
A: The first derivative of spending refers to the rate at which capital expenditures are increasing—the absolute growth number. The second derivative refers to whether that growth rate itself is speeding up or slowing down. Warsh is signaling that what matters most is not just whether AI investment is growing, but whether the acceleration of that growth is sustained. An accelerating investment curve suggests a self-reinforcing cycle that could have outsized economic consequences.
**Q: Why is the token market significant enough for the Fed to track?**
A: Tokens represent the fundamental unit of value exchange in the AI economy. With annualized sales at the two leading AI labs exceeding $100 billion—a more than 500% year-over-year increase—token transactions now constitute a measurable economic activity at a scale that rivals or exceeds many traditional industries. The Fed treating this as a trackable economic line item reflects the growing centrality of AI services to overall economic output.
**Q: What is the “hyper-Moore’s law” concept Warsh referenced?**
A: Moore’s law is the long-standing observation that computing power approximately doubles every two years. Warsh’s “hyper-Moore’s law” describes a scenario where AI capabilities are compounding at an even faster rate than traditional computing power has historically followed. This suggests that improvements in AI are accelerating in a way that could outpace the capacity of existing economic models to predict their impact.
**Q: How does the 95% failure rate of generative AI companies coexist with the massive growth Warsh described?**
A: This paradox reflects the nature of AI investment as a winner-take-all dynamic. While a vast number of AI startups fail, the capital and talent that don’t succeed are being absorbed by a small number of dominant players who are scaling rapidly. Nvidia’s record quarterly revenue and massive infrastructure commitments illustrate how value is concentrating rather than dispersing across the AI ecosystem.
**Q: Will the Fed’s interest rate decisions be affected by AI?**
A: Warsh did not make any specific commitments regarding how AI will influence rate decisions. However, by classifying AI as a new factor of production, he has opened the door for AI-driven productivity and investment trends to eventually inform monetary policy. The gap between recognizing AI’s importance and having a framework to integrate it into policy decisions remains the central tension the Fed now faces.
—
## Conclusion
Kevin Warsh’s Jackson Hole speech marks a turning point not just for the Federal Reserve, but for how the global financial system thinks about artificial intelligence. By placing AI alongside labor, capital, and land as a factor of production, Warsh has effectively rewritten the playbook for the world’s most powerful central bank. The numbers he cited—$100 billion in annualized token sales, capital expenditure growth at its fastest since 2021, and more than half of that growth tied to AI—are not speculative projections. They are measurable realities that demand a new policy response.
Yet Warsh’s most candid admission may be the most important one of all: the Fed does not yet have the tools to fully understand or integrate AI into its decision-making framework. Until it does, every rate decision, every economic forecast, and every policy pronouncement will be made with incomplete information about the force that is reshaping the economy in real time. The hinge point Warsh described is not just about AI’s potential. It is about the institution itself—its ability to adapt, to learn, and to govern an economy that is changing faster than its own models can keep up.
Thank you for reading
