American exceptionalism, and what it changes for buyers.

The phrase has been everywhere this cycle. Cathie Wood frames it as a structural thesis on US equities: the combination of AI capital expenditure, energy buildout, robotics, and dollar dominance keeps American innovation compounding at a pace no other economy can match. Marc Andreessen, in A16z’s American Dynamism thesis and the Techno–Optimist Manifesto, treats it as a builders’ obligation — that the United States is the place where the next century’s foundational companies will be built, and that policy should bend in that direction. Jamie Dimon’s recent shareholder letters carry the same shape with a different ending: America’s hand is the best in the world, and we are at risk of fumbling it.

I find the consensus useful precisely because it cuts in three directions: a bull, a builder, and a banker. When that many seats agree on the diagnosis and disagree only on what to do about it, the diagnosis is worth taking seriously.

This is not a piece about whether to be long the US. That is not my lane. It is about three things this thesis changes for the people I work with — operators sitting on enterprise IT decisions, private capital sitting on illiquid commitments, and boards sitting on five-year plans.

1. Compute is being repriced. Your usage curve is not.

Every operator I talk to is being asked to underwrite a multi-year cloud commitment, an AI co-pilot rollout, or a GPU procurement. The pitch from the vendor side reads the same way: “American exceptionalism in AI is structural. You don’t want to be the company that under-built.” That framing is half-right.

The structural part of the thesis lives in the hyperscaler capex curve, the energy buildout, and the workforce pipeline. The contingent part lives inside your business: do your users do enough work-with-AI to justify the run-rate cost?

The honest answer for most enterprises is “not yet, but soon, and we don’t know the shape.” That is a different procurement decision than “buy now, optimize never.” The macro thesis doesn’t do the unit-economics work for you. It changes the tailwind, not the engagement math. The renewals we have repriced in recent months consistently leaned on the macro pitch while the actual usage curve was running 30–40% below committed minimums.

2. Private capital is the natural buyer of the second derivative.

The headline AI names trade at multiples that already price in the consensus. But the thesis, taken seriously, implies a much wider set of plays than the Mag 7: the power grid, the industrial software stack, the rare-earth and processing chain, the workforce-development layer, the specialty manufacturers re-shoring under IRA and CHIPS. Many of these sit in private markets, in mid-market PE, or in operator-led search funds.

For private-capital allocators, the operational question is not “do we believe the thesis.” Most already do. The question is whether the firm is set up to underwrite the second-derivative bets — which generally requires a different bench (industrial operators, M&A counsel, financial modeling that pairs technical diligence with policy diligence) than what most allocators carry on payroll. That is a senior-counsel problem, not a permanent-hire problem.

3. Boards should ask a different question this year.

The most common question I see in board packets is “what is our AI strategy?” That question is a year out of date. The better question, given where the thesis lands, is:

If American exceptionalism plays out the way Wood, Andreessen, and Dimon describe, which of our five-year plan assumptions break first?

In practice that decomposes into four sub-questions. Does our energy and compute cost curve assume a world that no longer exists by 2027? Does our talent model assume a workforce that AI augmentation is about to reshape, and if so, are we hiring against the old shape? Does our customer concentration mask a coming reset in pricing power, as our customers themselves get repriced by the same forces? Does our capital plan keep us optionality-rich enough to ride the next eighteen months without being forced into a rights issue at the wrong moment?

Boards that ask those four questions get better steering-committee output than boards that ask about an AI strategy.

A note on the dissent.

Plenty of careful observers — Howard Marks and Ray Dalio prominent among them — argue the thesis is partially priced and partially wrong. The fiscal arithmetic, the demographic curve, and the political risk premium all push the other way. Dimon’s own letter acknowledges most of this. I take the dissent seriously.

The operator move is not to bet the firm on the consensus. The operator move is to identify which of our decisions are robust to the consensus being half right.

Most of them are not.

— Nolan Stark, Managing Director
Inquiries: nolanstark@milonadvisory.com