HomeCategory › Article

Nvidia Spent $40 Billion Buying Its Own Customers. Wall Street Called It a Strategy.

Autonomous AI agents are replacing traditional software and human workflows across every industry.

Image credit: Startups World News

TL;DR

Nvidia dropped $40 billion on AI investments in five months, with $30 billion going to OpenAI alone. The catch: nearly every company getting Nvidia cash also buys Nvidia chips, creating a circular capital loop that’s equal parts brilliant strategy and structural conflict. Founders taking Nvidia money should know exactly what they’re trading beyond equity.
How much has Nvidia invested in AI companies in 2026?
Over $40 billion in the first five months of 2026 alone. The single largest investment is $30 billion in OpenAI. The rest includes $3.2 billion in Corning, $2.1 billion in IREN, and roughly two dozen private startup rounds including CoreWeave and Nebius.
What is the circular investment concern with Nvidia's strategy?
Nvidia invests capital into AI companies, those companies spend much of that capital buying Nvidia GPUs, and the GPU revenue flows back to Nvidia’s bottom line. Critics argue this artificially inflates both Nvidia’s venture returns and its hardware revenue, since the money essentially circulates within the same ecosystem.
Should founders accept investment from Nvidia?
It depends on your leverage and your alternatives. Nvidia capital often comes with compute commitments, architecture alignment, and strategic visibility that goes beyond a typical VC relationship. Founders should negotiate compute terms separately from equity terms and maintain a diversified chip strategy.
Are regulators paying attention to Nvidia's dual role as supplier and investor?
Yes. Both the SEC and EU regulators are beginning to examine whether Nvidia’s simultaneous role as the dominant GPU supplier and the largest AI investor creates anti-competitive dynamics. The disclosure regime around these arrangements is evolving but hasn’t yet caught up to the scale.
What alternatives do AI startups have to Nvidia GPUs?
AMD’s MI300X, Google’s TPUs, and Amazon’s Trainium are the main alternatives. None currently match Nvidia’s full H100/B200 stack performance, but they’re improving. Diversifying compute suppliers before you’re locked in is the safest strategic play for founders building AI-dependent products.

Last Updated on May 12, 2026 by Taya Ziv

There’s a business model so old it doesn’t even have a fancy name. You lend someone money so they can buy your product. Car dealerships do it. Mattress stores do it. And now, apparently, the most valuable chipmaker on earth does it too.

Nvidia just crossed $40 billion in AI equity investments in 2026. Not over a decade. Not as part of some slow-drip corporate venture fund. Forty billion dollars in five months. And the biggest chunk, $30 billion, went to a single company: OpenAI.

Let that sink in for a second. Nvidia, the company that makes the GPUs that OpenAI can’t function without, just handed OpenAI $30 billion. OpenAI, in turn, will spend a significant portion of that money on… Nvidia GPUs.

If a mattress company did this, we’d call it a financing scheme. When Nvidia does it, analysts call it “ecosystem building.”

The Numbers Are Absurd

Here’s the full picture: beyond the $30 billion OpenAI stake, Nvidia poured up to $3.2 billion into Corning (which makes the glass and optical interconnects for Nvidia’s data center cables), $2.1 billion into IREN (a data center operator that runs Nvidia hardware), and participated in roughly two dozen private startup rounds. CoreWeave, Nebius, and a long tail of AI infrastructure companies all got checks.

This comes on top of 67 venture deals Nvidia closed in 2025. The company has quietly become the most active investor in AI, and it’s not even a venture fund.

The pattern is consistent: almost every company Nvidia invests in is also a customer. Some of those investments come paired with multi-year compute commitments and joint architecture agreements. It’s not just money. It’s money plus a roadmap lock-in.

Why This Should Bother You

Here’s where it gets uncomfortable for founders.

Nvidia isn’t just investing in the AI ecosystem. It’s investing in companies that are structurally dependent on Nvidia hardware. And those investments often come with strings that aren’t immediately obvious: capacity reservations, silicon roadmap alignment, and in some cases, board observer seats.

Think about what that means. Nvidia sees your cap table. Nvidia sees your compute budget. Nvidia sees your architecture decisions before you announce them. And Nvidia is also the only company that can sell you the chips you need to survive.

That’s not a venture investor. That’s a landlord who also runs the only hardware store in town.

The circular investment pattern is what’s really raising eyebrows. Capital flows from Nvidia to a startup. The startup spends most of that capital on Nvidia GPUs. The GPU revenue flows back to Nvidia’s income statement. Nvidia’s stock price rises. Nvidia uses the increased market cap to invest more. The cycle repeats.

Wedbush analyst Matthew Bryson acknowledged the pattern but argued it could help Nvidia build a lasting competitive moat. And honestly, he might be right. That doesn’t make it less weird. It just makes it effective.

The Founder’s Dilemma, 2026 Edition

If you’re a founder building anything that touches AI infrastructure, you’re probably already thinking about this. The question isn’t whether Nvidia’s money is available. It is. The question is what it costs you beyond the equity.

Taking Nvidia capital often means your compute stack gets locked to their ecosystem. Your roadmap gets visible to your chip supplier. Your negotiating position on GPU pricing gets… complicated. If your biggest investor is also your biggest vendor, every procurement conversation has a shadow.

And the alternative isn’t great either. Founders who don’t take Nvidia money report that GPU allocation timelines are a real competitive disadvantage. When Anthropic needed 220,000 GPUs and ended up renting from Elon Musk’s xAI, it wasn’t because they couldn’t afford to buy. Access to compute is the new currency, and Nvidia controls the mint.

This is a Hobson’s choice dressed up in a term sheet.

Actually, We’ve Seen This Movie Before

The closest historical parallel isn’t in tech. It’s in oil.

In the early 1900s, Standard Oil didn’t just sell fuel. It invested in railroads, refineries, pipelines, and distribution networks. It controlled the supply chain from wellhead to gas pump. The government eventually broke it up because controlling both the infrastructure and the money flowing through it was too much power for one entity.

Nvidia isn’t at Standard Oil levels yet. But the structural shape is similar. When you control the only GPU architecture that matters for training frontier models, and you’re also the biggest investor in the companies training those models, and your investments come with architecture lock-in agreements… regulators will eventually notice.

The SEC and Wall Street are already starting to ask whether the disclosure regime around these arrangements is keeping pace. Because Big Tech is already spending five times more on AI infrastructure than on payroll, and Nvidia sits at the center of both the spending and the investing.

What Founders Should Actually Do

So you can’t avoid Nvidia and you can’t ignore the structural risk. What do you do?

First, separate the money from the chips. If you take Nvidia investment, negotiate compute terms independently. Don’t let the equity conversation and the procurement conversation happen in the same room. The moment they’re linked, you’ve lost leverage forever.

Second, diversify your compute strategy before you need to. AMD’s MI300X is real. Google’s TPUs are real. Amazon’s Trainium is improving. None of them are as good as Nvidia’s H100/B200 stack today, but “good enough” compute from three suppliers beats “best available” compute from one supplier who’s also on your cap table.

Third, understand what you’re actually worth to Nvidia. If you’re spending $10 million a year on GPUs, you’re a small customer. If you’re spending $100 million, you have leverage. Know which bucket you’re in before you take the meeting.

And fourth, watch the regulatory landscape. Both the FTC and the EU’s DMA are paying attention to vertical integration in AI supply chains. When Founders Fund deploys $4.6 billion into just seven companies, that’s a concentrated bet. When Nvidia deploys $40 billion into its own customers, that might be something else entirely.

The Uncomfortable Truth

I’ll be honest: I’m not sure this is a problem anyone can solve right now. Nvidia has the best chips, the most capital, and the deepest relationships in AI. Saying “don’t take their money” is like telling a desert traveler to skip the only oasis because the water might have strings attached.

But founders should at least walk in with their eyes open. The AI funding landscape in 2026 isn’t a free market in any traditional sense. It’s an ecosystem where your chip supplier, your investor, and your competitive intelligence leak are increasingly the same entity.

The $40 billion isn’t venture capital. It’s not philanthropy. It’s the most expensive customer acquisition strategy in history. And it’s working.

Enjoyed this analysis?

Get stories like this in your inbox every Monday morning.

You Might Also Like