Nvidia will stop investing in OpenAI and Anthropic

Nvidia poured some $40 billion in OpenAI and Anthropic over the years.

Now, CEO Jensen Huang said the company is done.

That comment tells investors something important. The "venture phase" of the AI boom may be coming to an end.

For the past few years, Nvidia helped fund some of the most important AI startups in the world. The strategy was straightforward. Support young AI labs early so they build their systems on Nvidia chips. If those companies succeed, they turn into massive long term customers for Nvidia's AI chips.

OpenAI’s latest funding round is a good example. Nvidia committed about $30B as part of the company’s $110B raise. That’s far below earlier speculation that Nvidia could invest as much as $100B. That number never made much sense. OpenAI is expected to spend a similar amount buying Nvidia AI hardware anyway.

Huang said Nvidia typically stops investing once a company gets close to public markets.

Both OpenAI and Anthropic are widely expected to pursue IPOs this year.

Nvidia stepping back now means it's time for it to start collect the profits of its investment.

The disruption behind the news: Nvidia gets to make a lot more money.

Nvidia's goal to speed up demand for its GPUs has largely been achieved. The generative AI boom pushed almost every major model developer onto Nvidia hardware.

OpenAI alone is expected to spend tens of billions on computing infrastructure. If the company puts even half of its $110B funding into data centers, Nvidia captures a large portion of that through GPUs that cost roughly $25,000 to $40,000 each.

That’s the business model shift. Nvidia doesn’t need ownership stakes in AI labs if those labs are already locked into buying its hardware.

Once AI companies go public, they also stop behaving like research labs. They start behaving like infrastructure companies.

Public companies have to grow revenue quickly. That usually means larger models. Bigger computing clusters. And constant spending on compute dominated by Nvidia.

There’s also a less obvious financial incentive once these labs become public. Compute spending becomes a signal to investors.

Here’s a simple way to think about it. If OpenAI spends $60B of its $110B funding on GPUs at about $30k each, that implies roughly 2M high end accelerators deployed. With a typical 3 to 4 year depreciation cycle, that equipment creates about $15B to $20B in annual compute amortization.

That puts pressure on management to scale usage and revenue quickly to justify the investment.

This dynamic lines up nicely with Nvidia’s interests. The fastest way for AI labs to show growth to shareholders is to deploy more compute and train larger models. That reinforces dependence on Nvidia hardware.

What to watch next

The AI labs are about to become infrastructure companies.

Their largest expense will be compute.

And Nvidia controls that cost curve today.

There are three numbers worth watching over the next 24 months.

First, how much of OpenAI’s $110B actually turns into data center spending. Even $60B in GPU purchases would reshape the semiconductor market.

Second, watch IPO timing. Once OpenAI or Anthropic becomes public, investors will finally see their spending patterns. That will show exactly how dependent these companies are on Nvidia hardware.

Third, watch the companies trying to break Nvidia’s hold. AMD, custom chips from hyperscalers, and internal silicon projects from Microsoft, Google, and Amazon are all attempts to reduce GPU dependence. But switching costs are very high. AI models, training frameworks, and software tools are deeply optimized for Nvidia’s architecture.

That’s why Huang can step back from investing now.

The ecosystem phase did its job. Nvidia helped turn the AI boom into a cycle where companies depend on its hardware.

Once an industry needs your chips just to run, you don’t have to keep making venture bets. At that point, you’ve already won the AI chip war.

Nvidia (NVDA) has a Disruption Score of 4. Click here to learn how we calculate the Disruption Score.  

Nvidia is also part of the Disruption Aristocrats, our quarterly list of the world’s top disruptive stocks.