Nvidia is back in the news only days after shocking everyone with a $4 trillion market capitalization. This time, it’s the focus of a $40 billion data center megadeal that includes Microsoft and BlackRock.
The three companies are working together to buy Aligned Data Centers. If the transaction goes through, it will speed up the building of AI infrastructure in the U.S. and Europe. Nvidia sees this as yet another strong indicator that its chips will be important for AI in the future.
At the same time, Nvidia’s power throughout the globe is also being tested. China’s prohibition on GPU imports and AMD’s improving competitiveness are making the AI arms race even more heated.
Nvidia’s method of controlling everything from hardware to software, on the other hand, keeps getting results that few other companies can match.
That could be why CEO Jensen Huang is talking more and more about missed opportunities.
He said in a recent interview that Nvidia should have put more money into CoreWeave, an infrastructure company, and Elon Musk’s xAI.
The only regret I have … is that I didn’t give him more money.
Huang’s comments demonstrate that Nvidia recognizes the stakes are becoming higher for a company that is already redefining what AI can achieve.
Nvidia CEO Jensen Huang said the company should have put more money into CoreWeave and Elon Musk’s xAI.Image source: Kevin Dietsch/Getty Images
Nvidia’s mastery over the whole stack, not just processors, is what makes it the leader in AI.
The H100 and H200 GPUs are currently the best in the business for training and using large AI models. But its lead over the competition goes beyond silicon.
Nvidia is creating what experts are calling an “AI operating system” with its CUDA software ecosystem and tight developer lock-in. This platform is so complete that even trillion-dollar corporations like Microsoft and Amazon now rely on it.
It’s why the $40 billion purchase of Aligned Data Centers is more than simply a real estate deal.
Microsoft needs the physical space. BlackRock wants to invest in infrastructure. But Nvidia’s processors will run the AI workloads that will make it all worth it.
This isn’t just a theory. Microsoft recently released the GB300 NVL72 cluster on Azure, which connects more than 4,600 Nvidia GPUs into a single supercomputer-style accelerator, a record for Azure.
That’s the type of processing power that OpenAI and other cutting-edge laboratories require to run their next-generation models.
And it’s not only the U.S. becoming bigger. In Europe, Nscale, which is located in the UK, will send 200,000 Nvidia GPUs to Microsoft data centers.
This shows that demand is still high, and that Nvidia is still the best option for AI at scale.
Nvidia is the best at AI because it knows how to work with the complete stack, not just CPUs. The H100 and H200 GPUs are the finest in the market right now for training and using big AI models. But it has an edge over the competition that goes beyond silicon.
Experts say that Nvidia’s CUDA software ecosystem and severe developer lock-in are building a “AI operating system.” Even trillion-dollar companies like Microsoft and Amazon now depend on this platform because it is so comprehensive.
More Nvidia:
Microsoft has unveiled the GB300 NVL72 cluster on Azure. It unites more than 4,600 Nvidia GPUs into one supercomputer-style accelerator, which is a record for Azure. That’s the kind of computing power that OpenAI and other cutting-edge labs need to run their newest models.
It’s not only the U.S. becoming larger, either. Nscale, which is based in the UK, is planning to ship 200,000 Nvidia GPUs to Microsoft data centers throughout Europe. This proves that Nvidia is still the best choice for AI at scale and that demand is still strong.
Huang confesses that xAI and CoreWeave made mistakes as competitors close in.
Nvidia is succeeding, but CEO Jensen Huang is openly thinking about what the company didn’t accomplish. Noting in a recent interview that Nvidia was part of Elon Musk’s latest xAI fundraising round, Huang said he wished his company had put in more money.
“I want to be involved in everything Elon does,” Huang said. He realized that xAI’s ambitions and Musk’s potential for growth made it too big to ignore. The company is currently building its own big GPU clusters and employing the greatest AI professionals to compete with OpenAI.
Huang’s honest moment didn’t just address xAI. He also claimed he was sorry he didn’t put more money into CoreWeave, a cloud service for GPUs that works nicely with Nvidia’s ecosystem.
CoreWeave is a big name in high-density AI compute and is garnering business from the same Big Tech companies that Nvidia works with.
These disclosures came just days after Nvidia said it will give OpenAI up to $100 billion over time, using its Vera Rubin platform to pay for at least 10 gigawatts of additional data centers. Huang said that 4 to 5 million Nvidia GPUs make up one gigawatt.
This is a huge quantity of electricity that highlights how crucial infrastructure has become.
Companies other than Nvidia are moving. OpenAI and AMD have signed a deal worth billions of dollars to employ six gigawatts of MI450 GPUs. It also gave out warrants for up to 160 million AMD shares, but only if specific targets were met.
The message is clear: Nvidia may still be in the lead, but others are catching up fast.
The game around Nvidia is evolving as it becomes the best company in AI infrastructure. Competitors now want more than just its processors; they also want its connections, ecology, and even its access to international politics.
China’s tougher rules on Nvidia chip imports might cause problems for the corporation in other countries. At the same time, AMD’s return tells Wall Street that the AI war is far from finished.
Even the threats from authorities at home are evolving. The Trump administration recently backed off on probable antitrust lawsuits and is now backing a pro-AI policy strategy that explicitly calls for developing additional chips and data centers.
Nvidia is still the most significant aspect of it all. It offers AI the tools, software, and infrastructure it needs to function in the future, even as it takes more and more energy to do so.
But Jensen Huang’s most recent comments, in which he said he was worried about xAI and CoreWeave, suggest something else. The CEO of the world’s most valuable company knows that in this market, taking risks could be the only genuine risk.
