AI at the Fork in the Road: Boom, Bust, or Something More Complicated

As 2025 winds down, investors are staring at a question that feels both familiar and unsettling—just sharper this time. As Schroders’ Head of Global Economics, David Rees puts it¹, “investors’ focus is turning increasingly to AI-related risk: is it a stock market bubble that is about to burst and tip the US into recession? Or is it the onset of a third industrial revolution?”

Rees is quick to step away from the role of market prophet. “Economists are not best placed to decide if we are in a stock market bubble,” he admits. The real issue isn’t forecasting models or clever charts. It’s whether today’s market leaders can actually “deliver on earnings expectations,” which ultimately comes down to “the pace of AI adoption and its monetization.”

Instead of forcing a single forecast, Schroders lays out two deliberately different paths—AI Boom and AI Bust. The point isn’t to predict which one wins. It’s to pressure-test portfolios, policies, and assumptions, and to make sure complacency doesn’t quietly creep in.

Why the Bubble Hasn’t Burst—Yet

Rees starts by pushing back on the idea that AI enthusiasm has already gone too far. Even if the story eventually ends in a bubble, he argues, “we are not there yet.”

He points to three forces holding things together for now.

First, skepticism is everywhere. “Investor surveys highlight that the biggest current concern for asset allocators is an AI bubble.” Bubbles, Rees reminds us, “rarely form where everyone expects them.” When anxiety is this widespread, it usually means true excess hasn’t peaked yet.

Second, the economic backdrop doesn’t scream recession. “There is little sign of a global recession that might be the prompt for a major correction in the stock market.” In fact, Rees goes a step further, saying “we think that growth expectations are generally too pessimistic.”

Third—and this is the real wild card—comes policy risk. Rees warns that “politicization of the Federal Reserve may result in deeper interest rate cuts into a strong economy and fuel higher inflation.” History shows that when the Fed cuts while growth is still solid, “equities have performed well.” That kind of setup can drive a speculative melt-up. Still, he adds a note of restraint: “it doesn’t feel imminent.”

One Starting Line, Two Very Different Endings

Both scenarios begin from the same place. The macro backdrop is solid, and hyperscalers are planning aggressive investment. The result is strong AI-driven capex and rising equity markets through 2026. Rees notes that AI investment “has begun to make an increasing contribution to US GDP growth in recent quarters,” and Schroders assumes that contribution keeps growing.

The real break comes in late 2026, when markets pause and ask the uncomfortable question: is the hype turning into cash flow, or not?

The AI Bust: Painful, Familiar, and Cyclical

The AI Bust is “more straightforward, and perhaps more palatable,” precisely because we’ve seen versions of it before.

When investors decide AI is “not as commercially viable as hoped,” spending gets pulled back fast. Rees assumes “a two-year investment recession like that seen in the aftermath of the dot.com bubble.”

Consumption softens—not because households are cashing out stocks en masse, a narrative Rees doubts—but because “falling stock prices and rising unemployment would still have some negative impact on sentiment and spending.” That combination is “enough to tip the US into a mild recession.”

From there, the cycle resets. Higher unemployment eases capacity pressures, allowing “the Fed to cut interest rates to below neutral.” With fiscal stabilizers kicking in, Schroders expects “a cyclical, consumer-led recovery through late-2028 onwards.” Markets recover too, though leadership looks very different this time around.

One important side effect: the dollar weakens. Rees argues that “the erosion of trust in US institutions has probably weakened the safe haven properties of the dollar,” making it more vulnerable if capital flows reverse.

The AI Boom: Revolutionary—and Socially Disruptive

The AI Boom is where things get uncomfortable in a different way. Rees describes it as a compressed “third industrial revolution” unfolding “over a matter of months rather than years or decades.”

After a brief wobble in late 2026, investment surges as AI proves genuinely transformative—across large language models, robotics, and autonomous vehicles. Productivity jumps, returning to roughly 3.5% per year, levels last seen before the dot-com era unraveled.

But there’s a catch. Hold population growth and participation steady, and those productivity gains “imply an increase in the unemployment rate towards 10%.” It’s growth, but without enough jobs to support it. Consumption feels the strain.

Twin-Speed Inflation and Policy Paralysis

Inflation, in this world, doesn’t move in one direction.

Automation and rising unemployment push prices down in areas like housing and services. At the same time, the physical demands of AI push the other way. Data centres strain electricity grids, and as Rees notes, “around half of US electricity is generated using natural gas.” That opens the door to higher gas prices, more expensive fertilizer, and eventually higher food costs.

For policymakers, it’s a mess. “Twin-speed growth and inflation may cause policymakers to hesitate in cutting rates,” even as job losses mount. Rates eventually come down, but only after a prolonged period of uncertainty.

The Fiscal Fault Line Investors Can’t Ignore

One of the most overlooked consequences of an AI Boom sits in the public finances. Rees points out that “around three-quarters of Federal revenues come from the taxation of labour,” while welfare spending rises as workers are displaced.

Without serious tax reform, the likely outcome is clear: “the path of least resistance would surely be towards higher long-term bond yields as investors would demand larger risk premiums.”

Politics, Populism, and the Global Divide

Rees ends by zooming out. “Would governments allow such unfettered adoption of AI?”

Technology may be essential to offset ageing populations, especially in places like Japan and China. But rapid labour displacement risks fuelling populism—something Europe has already had a taste of. Emerging markets, unable to afford rapid adoption, could fall even further behind.

What This Means for Advisors and Investors

Rees’ closing warning is blunt: “Where potential AI scenarios are as divergent as igniting a boom or triggering a bust, complacency has to be among the bigger risks.”

Key takeaways are hard to ignore:

• AI isn’t a one-way bet. Boom and bust are both plausible—and could even overlap.
• Capex matters more than storytelling. Monetization is what counts.
• Labour disruption is the real macro shock, reshaping consumption, inflation, and politics.
• Fiscal risks are easy to underestimate and hard to reverse.
• Resilience beats prediction. Understanding scenarios matters more than betting on one.

For advisors, the message is simple: AI isn’t just another tech theme. It’s a macro regime risk that calls for humility, diversification, and constant attention.

Footnote:

1 Rees, David. “AI Economic Scenarios: Revolutionary Growth, or Recessionary Bubble?” Schroders, Dec. 2025. Marketing material for professional clients only.

Copyright © AdvisorAnalyst