Universities Aren’t a Public or Private Good - They’re Infrastructure

Who should fund universities? How much are tuition fees

Every few years, Britain has the same argument again. Should university be free, like the NHS, because an educated population is a public good the state has an interest in producing?

Or should it be priced, like any other service, because the main beneficiary of a degree is the graduate who earns more because of it?

Tuition fees go up. A cap gets debated. Someone proposes a graduate tax. The argument circles back to where it started, because nobody ever agrees on the premise underneath it: that a university is a good - something an individual consumes, which the state either subsidises or doesn’t.

I think that premise is wrong.

A university isn’t a good. It’s infrastructure. And once you see it that way, the entire debate about who should pay for it starts to look like the wrong question.

Nobody debates whether to fund the grid

We don’t ask whether the electricity grid should be a public or private good, because a grid isn’t something an individual consumes and then owns the benefit of. It’s the precondition for everything else to function. You can debate who operates the grid — state utility, regulated private monopoly, some hybrid - but you don’t debate whether the grid should exist, or ration it by ability to pay, because the cost of not having reliable electricity is borne by the whole economy, not just the household that goes without.

Treat a university system the same way and the funding question changes shape. The relevant output of a university isn’t “degrees for individuals.” It’s capacity - research capacity, talent-formation capacity, the physical and institutional link between a lab and the industry next door that lets a discovery become a product without a decade of friction in between.

That capacity is what an economy runs on the same way it runs on power and transport. Once you’re arguing about capacity rather than consumption, “should this be free or priced for the individual” stops being the central question, because the individual student was never the primary unit of account. The country was.

China never had this argument, and I don’t think it’s because the debate got won by one side. I think it’s because the premise was never accepted in the first place. University R&D capacity there was treated as something the state needed built, at scale, on a timeline the market alone wouldn’t produce - the same logic you’d apply to a rail network or a semiconductor fab.

Why Chinese Universities Are The Heart of China's Future Economy

And you can see the downstream effect physically, not just in policy language: research labs sitting inside industrial parks, professors running commercial ventures without institutional suspicion, campuses positioned next to the exact companies their graduates and their research feed. The infrastructure framing isn’t a metaphor for how China talks about universities. It’s a description of where they were built.

The obvious objection - and why it doesn’t quite land

The strongest case against this reframe is: infrastructure implies central direction, and research is precisely the domain where central direction historically performs badly.

Nobody can decide in advance which lab produces the breakthrough. That unpredictability is the whole argument for distributed, uncoordinated, occasionally wasteful-looking pluralism over any system where a planner declares in advance what matters. Roads are infrastructure because every driver needs the same road. Research isn’t like that, and treating it as if it were risks trading genuine discovery for prestige projects that look good in a five-year plan.

But this objects to the wrong thing.

“Infrastructure” doesn’t have to mean the state picks the research topics or the winning company. It can mean the capacity is guaranteed - the labs, the funding continuity, the administrative and physical proximity to industry - while what happens inside that capacity stays genuinely competitive.

China isn’t running one national AI lab and declaring it the winner. It’s guaranteeing that dozens of labs, sitting close enough to commercial capital and commercial problems, can compete fast, because none of them has to spend two years negotiating for the right to exist. Guaranteeing the plumbing is a different claim from picking the outcome, and it’s the one actually worth making.

It’s also fair to name what this framing costs, because a system that guarantees capacity regardless of market signal can overbuild - empty science parks, prestige labs producing output nobody wanted, capacity justified by political optics rather than genuine need. That risk is real and I’m not going to pretend the trade-off doesn’t exist. But the alternative has its own failure mode, and Britain is currently living inside a clean example of it.

What “priced good” actually costs, in practice

In 2023, the UK government commissioned an independent review into why university spinout companies - the exact handoff point where research becomes a company - were struggling to raise investment and get off the ground.

What it found was a negotiation problem baked into the funding logic. Because universities are cash-constrained institutions that have to behave commercially to survive, their tech transfer offices were negotiating equity stakes in spinouts as a counterparty, not a collaborator - and negotiating hard.

The average opening equity offer from a university was around a third of the company. Close to a third of universities opened negotiations demanding half. Two out of three academic founders never even left their university post to run the company they’d started, in part because ownership terms were still unresolved. The review’s own conclusion was blunt: the process took too long, and deal negotiation was a primary reason why.

That’s the cost of treating the university as a good rather than infrastructure, made concrete.

When the institution has to fund itself by extracting value from the very company it helped create, every spinout becomes a negotiation before it becomes a business. Compare that to a system where the lab doesn’t need the equity, because its existence was never contingent on winning that negotiation in the first place - where the professor and the company next door are already collaborators because the funding question was answered upstream, once, at the level of the whole system, instead of being re-litigated at every single handoff.

The UK review has since pushed universities toward smaller stakes, and it’s working - average equity has come down.

That’s an improvement.

But it took a Treasury-commissioned intervention to talk institutions into taking less of something they were never meant to be extracting from in the first place. That’s what happens when the underlying category is wrong: you spend policy effort correcting symptoms of a framing problem, one review at a time, instead of asking whether the framing was the mistake.

The actual question

I’m not arguing China’s version is free of cost, or that Britain should copy it wholesale - a system that answers the funding question once, centrally, gives something up in return: arm’s-length distance from state direction, a founder’s fuller claim on their own company, the friction that sometimes protects academic work from being pulled too early into commercial urgency. Those are real and worth keeping.

But I don’t think “public good or private good” is the axis worth arguing on anymore. The more honest question is whether you’re building capacity or selling a product - because everything downstream of that answer, the negotiations, the timelines, the distance between a discovery and the moment it ships, follows from which one you picked.

Britain keeps having the old argument. It might be worth having a different one.

This is a part of a series on the Platform State.