Community Trust ScoreLikely Real
The numbers are staggering. The U.S. is sitting on an $800 billion AI investment wave, and nobody can quite agree on whether to ride it or pump the brakes.
Why It Matters
The debate over the rapid expansion of AI investments highlights the tension between innovation and regulatory caution, a dynamic that could significantly impact equity markets. As the U.S. grapples with its strategic response to AI's growth, the potential for volatility in equity valuations may arise, particularly as investors weigh the risks of overexuberance against the need for structured development. This crossroads not only affects tech-centric sectors but also has broader implications for economic stability and global competitiveness in the evolving landscape of artificial intelligence.
President Donald Trump wants to go faster. He’s announced the creation of a “Super Intelligence Force,” led by former SEC chief Jay Clayton, aimed at accelerating AI development and keeping pace with China. On the other side, Anthropic CEO Dario Amodei is pushing for a deliberate slowdown — enough breathing room, he says, for safety research to actually catch up with the technology. OpenAI’s Sam Altman and Google’s Demis Hassabis have backed that general idea. Senator Bernie Sanders goes further still, floating a permanent ban on superintelligence altogether. So the range of opinion here runs from “build faster” to “never build it at all.” That’s not a small gap.
On September 29, Trump and a group of major AI executives signed a voluntary safety accord. It focuses on internal controls and audits — not a development pause.
What a Slowdown Could Actually Cost
The financial stakes are hard to overstate. SoftBank recently ran a $10 billion bond sale to fund OpenAI — one of the largest non-financial corporate bond deals of the year. The St. Louis Fed put out a report attributing 39% of real GDP growth in early 2025 to AI-related investments. Thirty-nine percent. That’s not a rounding error. That’s the engine.
So what happens if that engine stalls? The IMF ran the numbers. A reversal in AI investment expectations could drag U.S. equity markets down by 20% and pull GDP along with it. The IMF flagged this back in January, warning that if AI productivity disappoints, the resulting market corrections could hit household wealth hard — and from there, consumption, and from there, broader economic activity. It’s a chain reaction most policymakers would rather not test.
Fitch’s scenario is even grimmer. A 35% equity shock in a slowdown scenario, the ratings agency projects, could tip the economy into recession. That’s not a fringe view anymore. It’s the kind of number that gets circulated in finance ministries.
And yet some voices aren’t panicking. David Minarsch and Shiv Shankar both see continued productivity gains from AI tools already deployed, even if frontier model development cools. The argument is basically that the existing technology is still early in its adoption curve — slowing new model releases doesn’t erase what’s already running.
The Overinvestment Problem Nobody Wants to Say Out Loud
The Bank for International Settlements has been the most direct about the uncomfortable math. Pablo Hernández de Cos drew parallels to past economic cycles — the kind where capital floods a sector faster than returns can justify, and then something breaks. A BIS paper put AI infrastructure investment at 1.5 times the socially efficient level. That’s not a minor overshoot. That’s the kind of figure that, in hindsight, tends to show up in the post-mortems.
The BIS warning echoes historical patterns. Sectors that attract this kind of capital concentration — where investor enthusiasm runs well ahead of demonstrated returns — don’t always land softly. The concern isn’t that AI doesn’t work. It’s that the investment wave has priced in outcomes that haven’t arrived yet, and probably can’t all arrive at once.
But UBS isn’t buying the doom scenario outright. The bank’s position is that slowing model development doesn’t necessarily mean capex collapses. UBS projects AI industry capital expenditure hitting $1.2 trillion by 2027, up from an estimated $900 billion this year. If that forecast holds, a paced approach to development might actually reduce overinvestment risk without gutting the industry’s financial trajectory. Maybe. It’s still a projection.
Mark Zuckerberg and Nvidia’s Jensen Huang are firmly in the accelerationist camp — no coordinated slowdown, full stop. Their argument is geopolitical as much as economic: pulling back hands China an advantage that’s hard to recover.
So the voluntary accord signed in late September sits somewhere in the middle. Internal audits, yes. A pause, no. Whether that’s a meaningful compromise or just a way to avoid a harder conversation is unclear yet.
What’s not unclear is the exposure. Thirty-nine percent of GDP growth tied to one sector. A potential 20% equity drop if expectations shift. A 35% shock scenario from Fitch. And a BIS paper saying the industry is already investing at 1.5 times the rational level.
UBS still sees capex at $1.2 trillion by 2027.
Frequently Asked Questions
What is Trump’s “Super Intelligence Force”?
It’s an initiative announced by President Donald Trump to accelerate U.S. AI development, with former SEC chief Jay Clayton tapped to lead it.
What does the Fitch slowdown scenario actually project?
Fitch projects a 35% equity shock in the event of an AI investment slowdown, a scenario the ratings agency says could trigger a recession.
