Entrepreneurs Don't Have a Flag When the Chips Are Down
There’s a conversation happening in boardrooms and congressional hearings about which AI models Americans should be allowed to use. The implicit assumption is that entrepreneurs will follow the rules, pick the approved tools, and compete accordingly.
That assumption is wrong.
When a founder is six months from running out of runway, they don’t care where the model was trained. When a small business owner is trying to cut costs before payroll, they don’t ask what country the API is hosted in. When a startup is staring down a competitor who just shipped a feature twice as fast, they use whatever tool closes that gap.
Entrepreneurs don’t have a flag when the chips are down.
The Game Theory Is Simple
Game theory doesn’t care about policy intentions. It cares about incentives and outcomes.
If your competitor uses a Chinese model that’s cheaper, faster, or better at a specific task, and you refuse to use it out of patriotic principle, your competitor wins. They capture the customer. They get the funding. You don’t.
This isn’t cynicism. This is how markets work.
The same logic applies at the national level. Countries that restrict access to powerful AI tools are not making their entrepreneurs more patriotic. They are making them less competitive. And less competitive entrepreneurs lose to foreign competitors who have no such restrictions.
The country that limits model power doesn’t get more loyal founders. It gets fewer successful ones.
We’ve Seen This Movie Before
This isn’t the first time the world has gone through a moment where national pride and competitive survival collided.
American manufacturers learned Japanese production techniques in the 1980s. Not because they admired Japan. Because Toyota was eating their lunch and they had to adapt or die.
American companies outsourced to Indian engineering talent for decades. Not because of ideology. Because the talent was exceptional and the economics made sense.
American firms used Soviet-era mathematical methods in finance and physics. The best tools don’t carry passports.
AI models are just the latest version of this. The math doesn’t care who wrote it.
The Enforcement Problem
There’s also a practical issue that policymakers tend to underestimate.
You can restrict a chip sale. You can sanction a hardware manufacturer. You can prevent a large enterprise from signing a contract with a foreign AI provider.
You cannot stop a founder with a laptop and a VPN.
The friction of restriction falls hardest on large, visible, compliance-heavy organizations. Startups, freelancers, and small businesses operate at a speed and scale that makes enforcement nearly impossible. By the time a regulation is written, tested, and enforced, the tools have changed three times.
Restrictions don’t eliminate access. They just add friction for the people least able to absorb it.
The Real Risk
The danger isn’t that American entrepreneurs use Chinese models. The danger is that restrictions give them a reason to.
American AI has a real advantage right now. OpenAI, Anthropic, Google, Meta are world-class labs producing world-class tools. Most entrepreneurs prefer to use them. They’re integrated, reliable, and improving fast.
But that preference is not unconditional. It’s conditional on those tools being the best option for the job.
The moment a Chinese model is meaningfully better at a task that matters to a founder’s survival, that founder will use it. No law will change that calculus. The only thing that changes it is making sure the American tools stay better.
That means more investment in domestic AI, not less access to foreign tools. It means competing, not restricting.
The Bottom Line
Nationalism is a luxury good. It’s something you can afford when everything else is equal.
When the chips are down — when the business is on the line, when the competitor is gaining, when survival is the question — founders optimize for survival. That is not a character flaw. That is exactly the instinct that makes entrepreneurship work.
Countries that understand this will build policy around it. They will compete to have the best tools, the best infrastructure, and the most attractive environment for building.
Countries that don’t will find their entrepreneurs winning anyway — just with someone else’s flag on the model.
Get essays like this in your inbox
New essays on AI, education, and the future of work. No spam, unsubscribe anytime.