Nvidia's $40B AI Bet Is Vendor Financing With Extra Steps

Nvidia's $40B AI Bet Is Vendor Financing With Extra Steps

Key Takeaways - Nvidia committed over $40 billion to AI equity investments in just the first four months of 2026. More than double its full-year 2025 total - $30 billion went to OpenAI, with Huang signaling that check may be the last before IPO - The most transparent circular deal: Nvidia invested $2.1B in IREN, which then committed $3.4B back to Nvidia in cloud services over 5 years - If you're budgeting GPU costs for 2027, build in a scenario where this loop closes

---

Here's the thing: Nvidia just put more money into AI startups in four months than most sovereign wealth funds deploy in a year. $40 billion. Four months. One company.

And the largest single check — $30 billion to OpenAI.

Might be the last before IPO, according to Jensen Huang himself.

If you're a small operator running AI workloads on GPU capacity, this matters more than it looks like it should.

Because that $40B isn't charity.

It's vendor financing with a venture capital wrapper.

The IREN Deal Tells You Everything

Nvidia invested $2.1 billion in IREN in April 2026.

IREN then committed to spending $3.4 billion with Nvidia over five years.

That's not an investment. That's Nvidia depositing money in a customer's account and getting it back plus interest in the form of cloud service contracts.

It gets worse. Nvidia's $5 billion Intel stake is now worth over $25 billion. A 200%+ return in under two years. The strategy is working so far. That validation is exactly why analysts at Wedbush and Mizuho are calling this "circular investment" and why the SEC is reportedly examining disclosure practices around these deals.

Ben Bajarin at Creative Strategies asked the right question: if the cycle turns, how much of the current GPU demand is organic versus supported by Nvidia's own balance sheet? Nobody wants that number answered.

What $40B in 120 Days Actually Means

Strip out the circular demand and what's left? That's the question that should keep every infrastructure planner up at night.

Nvidia's total 2025 private AI investments were $17.5 billion.

In 2026, they're already at $40 billion-plus with four months remaining.

The single biggest piece — $30B to OpenAI. Was scaled back from a planned $100 billion check. That's not small. That's the entire annual revenue of a Fortune 500 company going to one customer.

When your GPU supplier is also your supplier's landlord, the pricing model isn't a market. It's a subsidy scheme that lasts until it doesn't.

What This Means for Your GPU Bill

I run GPU workloads through neoclouds for client work.

I don't have a venture capital arm.

I don't have $2.1 billion to plumb into infrastructure partners. And starting later this year, I might be riding the tail end of subsidized demand that vanishes when Nvidia closes this chapter.

Here's what I think the realistic scenario is: Nvidia writes its last big check to OpenAI before the IPO.

The circular flow slows.

Neocloud operators who built capacity on guaranteed contracts start trimming. Spot capacity gets tighter. Prices. Which have been held down by Nvidia-underwritten demand. Move up.

For small businesses and solo operators, this means budgeting for potential GPU price shifts isn't paranoid. It's just arithmetic.

If you've locked in one-year reserved capacity, great.

If you're on-demand, start looking at what happens at renewal.

The specific action: audit your current GPU commitments now.

What are you paying per token or per hour?

What does that look like at 20% higher? At 40%? Run the scenario. It's a two-hour exercise that might save you a significant line item next year.

The Bottom Line

Nvidia's $40 billion AI investment spree is the most effective vendor financing operation in tech history. It keeps demand flowing, keeps chip sales up, and generates returns that validate the strategy to shareholders. The problem is that everyone downstream. The neoclouds, the AI startups, the enterprises. Is riding a subsidies bubble that has a known closing date.

If you're building on AI infrastructure this year, treat Nvidia's IPO as your marker.

After that, the circular flow slows.

Budget accordingly.

--- Sources: CNBC | TechCrunch | Morningstar