Microsoft’s AI buildout is showing up in one of the clearest places possible: the chips it bought. Across 2024, the company purchased 485,000 Nvidia Hopper chips, according to reporting from the Financial Times, which cited data from tech consultancy Omdia.
That figure puts Microsoft far ahead of major rivals mentioned in the same reporting. Meta bought 224,000 of the same flagship Nvidia chip this year, while Microsoft’s 2024 purchase more than tripled the number of Nvidia chips it bought in 2023.
A chip order that signals priority
The Nvidia Hopper purchase is not just a procurement detail. It is a practical signal of how heavily Microsoft has leaned into AI infrastructure during 2024.
AI systems require large amounts of computing power, and flagship chips sit at the center of that equation. The source article does not break down exactly where each chip will be used, but the scale of the purchase makes one point difficult to miss: Microsoft is investing aggressively in the hardware base needed to support AI work.
The comparison with Meta helps frame the size of the move. Microsoft bought more than twice as many Nvidia Hopper chips this year than Meta did. The source also notes that Microsoft bought more than twice as many as any of its biggest rivals.
For a company already associated with enterprise software, cloud services, and AI partnerships, the chip purchase shows that strategy is being backed by physical infrastructure. The AI push is not only about models, products, or announcements. It is also about securing enough computing capacity to run demanding systems.
How Microsoft’s AI year fits together
The Nvidia Hopper purchase sits alongside several other AI-related moves mentioned in the source article. Microsoft is also building its own custom AI chips, called Maia. The company announced Maia at its Ignite conference in late 2023.
That matters because it shows Microsoft is not relying on a single route for AI hardware. On one side, it bought 485,000 Nvidia Hopper chips across 2024. On the other, it is developing Maia as its own custom AI chip effort.
Microsoft also deepened its partnership with OpenAI during the year. In October, it participated in OpenAI’s $6.6 billion funding round. The source describes the round as colossal, and the number alone makes clear that the relationship remains a major part of Microsoft’s AI position.
Taken together, the moves point to a company building across several layers at once:
- Nvidia Hopper chips for immediate AI computing capacity.
- Maia as Microsoft’s own custom AI chip project.
- A deeper OpenAI partnership connected to a $6.6 billion funding round.
- Data center power planning tied to a long-term energy deal.
Each part supports the same broader direction. Microsoft is not treating AI as a narrow product feature. It is building around chips, partnerships, and the infrastructure needed to keep data centers supplied.
Power is part of the AI infrastructure story
The source article also points to another major piece of Microsoft’s AI strategy: electricity. In September, Microsoft signed a deal to reopen the infamous Three Mile Island nuclear plant and purchase all of the power that the plant creates for 20 years.
The purpose, according to the source, is to help juice Microsoft’s data centers. That detail connects the chip purchase to a larger operational challenge. Buying advanced chips is only one part of scaling AI. Running data centers that can use them is another.
The Three Mile Island agreement shows how far the infrastructure conversation can extend. Microsoft’s AI effort is not limited to chips or software. It reaches into long-term planning for power supply, because data centers need dependable electricity to operate at the scale the company is pursuing.
The source does not say how the power will be allocated across Microsoft’s operations. It does, however, make the connection clear: the deal is meant to support data centers, and those data centers sit behind the company’s broader AI activity.
What the numbers reveal
The clearest number in the story is 485,000. That is how many Nvidia Hopper chips Microsoft bought across 2024. The second key number is 224,000, the number Meta bought of the same flagship Nvidia chip this year.
The gap between those two figures is central to the story. Microsoft’s purchase was more than twice Meta’s. It also more than tripled Microsoft’s own Nvidia chip buying from 2023.
Those comparisons matter because they show momentum, not just size. Microsoft was already active in AI before 2024, but the reported chip buying suggests a much larger hardware commitment this year than the year before.
The same pattern appears in the other moves cited by the source. Maia points to in-house chip development. The OpenAI funding round points to a continued strategic partnership. The Three Mile Island deal points to the power requirements behind large data center operations.
For Microsoft, the AI race is therefore not just about having access to models or launching AI features. Based on the facts in the source article, it is also about owning or securing the building blocks that make large-scale AI possible: chips, custom hardware, partnerships, and power.