Bargaining chips: Nvidia is the bank of AI

Nvidia took three decades to reach a $1 trillion valuation, then doubled that in nine months, and crossed $5 trillion less than two years later. But the chip designer now faces a peculiar problem: investors expect that breakneck growth to continue indefinitely. To meet those expectations, Nvidia has begun acting less like a semiconductor company and more like a bank—using its colossal balance sheet to finance its own customers' ability to buy its chips. The strategy, which The Economist's Shailesh Chitnis calls "financial engineering," has generated over $350 billion in potential obligations through guarantees, equity investments, and purchase commitments. Critics worry it echoes the circular financing that fueled the dot-com bust. For now, Nvidia's cash hoard—$96 billion in liquid assets and projected $200 billion in free cash flow this year—can absorb the risk. But the more of the boom Nvidia finances, the more of the bust it may have to bear.

The bank of AI

The core problem is straightforward: AI compute is extraordinarily expensive. A single one-gigawatt data center costs nearly $50 billion to build. While hyperscalers like Amazon, Google, and Microsoft can afford that spending, smaller AI labs and startups—many bleeding cash—cannot. Nvidia's solution is to step in and enable their purchases through guarantees, backstop commitments, and equity investments.