Nvidia lines up $500 billion from Wall Street giants to build AI's physical backbone
Six of the world's biggest investment firms are treating AI chips and data centres as infrastructure on a par with roads and power grids, and Nvidia is at the centre of the deal.
Key points
- Nvidia announced a $500 billion capital-raising partnership with six major investment firms on Monday.
- Partners include Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR.
- The money will fund data centres (the large warehouse-scale buildings that house AI computers) and chip factories.
- BlackRock separately struck a deal last month to finance and own a majority stake in a Meta data centre in Texas.
- AI companies including Google, Meta, Amazon and OpenAI have collectively spent over $1 trillion on AI infrastructure in three years.
Nvidia, the company that makes the specialised computer chips powering almost every major AI product you have heard of, has pulled together half a trillion dollars in planned investment from six of Wall Street's biggest names. The partners are Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR.
The goal is straightforward: build more of the physical infrastructure that AI depends on.
What exactly is being built?
The money targets two things. First, new data centres, the vast warehouse complexes packed with computer hardware that run AI services like ChatGPT, Google Search and Claude. Second, new factories to manufacture the AI chips themselves, making them less scarce and easier to buy.
Nvidia chief executive Jensen Huang put it plainly: "In AI, compute is revenue." Compute, in this context, just means computing power, the raw ability to process information at speed.
Jim Zelter, president of Apollo, which manages more than $1 trillion in assets, called modern computing power "a scarce, mission-critical asset class" positioned to drive long-term economic growth.
Why are banks suddenly interested in chips and data centres?
Investors are treating AI hardware the way they have long treated toll roads and electricity grids: as essential kit that generates steady, predictable returns.
This is a shift. As KKR co-chief executives Joe Bae and Scott Nuttall noted: "Compute has become a critical infrastructure asset."
Nvidia's chips, known as GPUs (graphics processing units, the specialised processors that handle the number-crunching AI requires), are used by Google, Meta, Amazon, Microsoft, SpaceX, Tesla, OpenAI and Anthropic, among others. That near-universal adoption has driven Nvidia's stock market value up roughly fivefold in three years.
What does this mean for ordinary people?
More investment in data centres and chip factories should, over time, ease the bottleneck that limits how quickly AI tools improve and how reliably they run.
Anthrophe recently said demand for its Claude chatbot had grown so fast that "demand requires significant new compute." More supply means fewer outages, faster responses, and potentially lower costs for the businesses that build the apps people use every day.
For now, this is a story about very large organisations moving very large sums. The BBC Technology desk first reported the announcement. But the downstream effect, more chips, more data centres, more competition for AI capacity, will eventually reach anyone who uses a phone, a laptop, or a search engine.
| Partner | Known scale |
|---|---|
| Apollo | Manages over $1 trillion in assets |
| BlackRock | Separate Meta data-centre deal, Texas |
| Blackstone | Major alternative asset manager |
| Brookfield | Specialist in real-asset infrastructure |
| Goldman Sachs | Top-tier global investment bank |
| KKR | Large-scale digital infrastructure investor |
Common questions
Does this mean Nvidia is becoming a bank or a property company?
No. Nvidia remains a chip designer. This arrangement lets outside investors fund the buildings and factories that use Nvidia's technology, freeing Nvidia to focus on designing the next generation of chips.
Will this make AI cheaper or faster for consumers?
Potentially yes, over time. More supply of chips and data-centre space tends to reduce the scarcity premium that currently drives up costs for companies building AI products, and those savings can eventually flow to users.
Is $500 billion actually guaranteed to be spent?
Not all at once. This is a planned capital commitment across multiple projects and partners over an extended period, not a single lump-sum transaction.
