What Nvidia Is Really Building

Nvidia has spent years making artificial intelligence more powerful. It is now helping finance the infrastructure required to use it.

The chipmaker has joined Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to develop financing platforms that could mobilise more than $500bn for AI infrastructure. Nvidia may itself backstop as much as $125bn of the eventual transactions.

This is more than an expansion of customer finance. It reflects a shift in where Nvidia sees the principal constraint to its growth.

The challenge is no longer simply producing more advanced processors. Their deployment requires data centres, energy, cooling systems, networks and vast amounts of capital. Goldman Sachs estimates that annual investment in AI infrastructure could rise from about $765bn in 2026 to $1.6tn by 2031.

When the constraint moves beyond the product, companies often follow it.

Aircraft manufacturers support customer financing. Telecommunications companies subsidise devices to expand their networks. Manufacturers invest in charging infrastructure to accelerate the adoption of electric vehicles. In each case, the boundary between supplying a product and creating the conditions for its use becomes less distinct.

Nvidia’s move raises a wider question for education. When the greatest barriers to educational participation and success sit beyond the traditional boundaries of the institution, how far should those boundaries extend?

For much of their history, schools and universities have defined their role through teaching, research and qualifications. Yet the value of that core offer increasingly depends on conditions that institutions do not entirely control.

A student may secure a place at university but struggle to find affordable housing. Another may have access to a course but lack the digital infrastructure, childcare or financial flexibility needed to complete it. A graduate may leave with strong academic knowledge but without an obvious route into employment.

These are not simply peripheral issues. They influence access, persistence and outcomes. But nor are they all problems that an educational institution can, or should, solve alone.

This is where the comparison with Nvidia becomes useful. The company is not becoming an energy provider, construction group and bank in the conventional sense. It is building partnerships that address the constraints surrounding its product while preserving its central role within the ecosystem.

Education is already moving in a similar direction.

Universities are expanding accommodation partnerships, developing mental health provision, working more closely with employers and offering more flexible financial arrangements. Schools are building relationships with health services, community organisations and technology providers. The institution is becoming less self-contained, not because its educational purpose has diminished, but because delivering that purpose increasingly depends on a broader network.

The strategic question is therefore not whether institutions should do more. It is where they should lead, where they should partner and where responsibility properly belongs elsewhere.

Extending too far can disperse resources and blur institutional purpose. Extending too little can leave the core educational offer dependent on conditions that prevent some learners from benefiting fully from it. The appropriate boundary will differ by institution, population and context.

What matters is that the boundary is designed deliberately.

This requires institutions to distinguish between activities that are central to their purpose and conditions that are essential to achieving it. They may not need to own student housing, build technology platforms or employ every specialist directly. But they need to understand how the availability and quality of those services affect the educational outcomes for which they are responsible.

Partnerships then become more than a way to outsource delivery. They become part of institutional design.

That changes the questions leaders must ask. Not only what should the institution provide, but what must exist around it? Which constraints can it address directly? Which require shared investment? Which partnerships expand institutional capacity, and which create dependencies that need to be governed?

Nvidia’s financing initiative suggests that competitive advantage can increasingly lie beyond the product itself. A powerful chip has limited value without the infrastructure to run it. An educational place has limited value if the conditions required to take it up are absent.

The future institution may not be the one that tries to provide everything. It may be the one that understands most clearly what must surround its core offer for that offer to work.