Microsoft is emerging as a power player at the end of 2024 through its volume purchases of NVIDIA AI chips, the number greater than that of any other tech company currently. The source of this news is a report by the Financial Times (FT). We have the statistics, or at least close enough estimates, to establish context so it’s easier to understand this development. Let’s go~
Microsoft Spent 31 Billion To Acquire 485,000 AI Chips From NVIDIA In 2024
There is growing uncertainty in the tech industry due to the tariffs by the incoming US President Donald Trump. To counteract this, the tech giants, including Microsoft, HP, and Dell, are pressing suppliers to speed up production. On the local side of things, Microsoft, in particular, is seeking to secure the lead procurement of AI chips. The growing reliance on artificial intelligence is nothing new, as it is the natural progression of technological advancement. Interestingly, Microsoft’s staunchest rival in the gaming space, Sony, is collaborating with AMD for AI advancement applications in gaming. However, it’s worth mentioning that the gaming niche only makes up a fraction of Microsoft’s entire business. But that’s beside the point, as we’re highlighting the emphasis on artificial intelligence chips and their surging demand.
This is quite a curious situation for Microsoft. That’s because, in addition to the large order of NVIDIA chips, they’re also developing their own version of AI chips. The official title of the custom artificial intelligence chips from Microsoft is Maia. It’s certainly an interesting move and grounds for a lot of speculation as to whether ‘Maia’ is still not ready for use or not. Let’s circle back to the reliance on artificial intelligence and how quickly the company is adopting it. Microsoft has invested over $1.6 billion to secure nuclear energy for 20 years to power its AI data centers. This is because AI data centers are energy sponges requiring constant high-energy sources to function. With its penchant for enormous investments for long-term gains, Microsoft is seeking to jump ahead of the curve for AI progression.
How Many AI Chips Have Other Companies Purchased?
We’ve discussed Microsoft’s foray into AI and its ambitious aims to succeed with it. But what about the competitors? Well, the same report disclosing Microsoft’s flex also revealed the other tech giants’ orders for NVIDIA chips. We’ll discuss them individually.
While nowhere near the Microsoft figures, Tencent and ByteDance, two Chinese technology giants and conglomerates, share second place. They are in a similar range, with a purchase of 230,000 NVIDIA AI chips each. Tencent is another titan of the tech industry with branches in several other ventures. One recent such venture includes a buyout of Ubisoft, a video game company.
After Microsoft’s colossal nearly 500K chips order, Meta Platforms Inc., of Facebook fame, comes in third place. This is quite the shocker, but Meta’s orders stack up at 224,000 NVIDIA chips. After that, xAI holds the fourth position by purchasing 200,000 NVIDIA GPUs.
Within the same analysis, we also discover that Amazon is far down the charts, only purchasing 196,000 NVIDIA chips. This is another move that raises a lot of questions. However, it also proves that perhaps relying solely on NVIDIA chips isn’t Amazon’s only move.
In a drastic turn of events, the final contender of the Financial Times report was another big name. The last place in this AI chip battleground belongs to Google, with only 169,000 NVIDIA chips to its name.
A big company absent from these sales comparisons is Apple. However, that’s nothing surprising as most of these companies are collaborating with third-party partners to develop more custom AI chips. Some are developing their own versions that are due to launch in the coming years and are relying on NVIDIA for their present needs.
NVIDIA A Key Modern Player
The sky’s the limit at the moment for NVIDIA’s potential revenue growth. Several tech giants alongside Microsoft are tapping it to produce its in-demand AI chips, the Hopper GPU. Interestingly enough, the Hopper GPU has been in the market since 2022. NVIDIA is all set to launch its successor, Blackwell, in 2024. However, several complications and necessary adjustments during testing to fix those problems were factors for withholding its launch. Blackwell is widely marketed to have even greater capability than the Hopper. In fact, according to an interview with DCDS (Data Center Dynamics), the head of data center product marketing at NVIDIA, Mr Dion Harris, goes on the record to say:
“When Grace Blackwell comes to market next year (2025), I think a lot of those [Hopper applications] will carry forward, and you’ll see a lot more excitement just in terms of performance and benefits. But I think Grace Hopper is really having a tranformative impact on how some of those applications are being developed and run.”
A simple analysis of these words is that NVIDIA is still banking on the ‘key value’ of the Hopper. In fact, with Blackwell’s delay, there is no immediate competitor for Hopper. Those that are in development will need time and resources.
Final Thoughts
Well, there it is, a who’s who and where everyone stands in the ever-expanding AI-chips buy rush of NVIDIA chips and Microsoft’s dominance in standing ahead of the pack. However, it’s still not a done deal, as there are several emerging factors, like in-house custom AI chips, that several tech companies are developing. Who do you feel will emerge as the top tech company once the next wave of AI technology launches? We’re curious to know what you all think of this report by the Financial Times.
In case you’re wondering more about the stiff competition in the AI chip market, check out our coverage of AMD losing market share to Marvell and Broadcom. It gives insight into how a potentially challenging situation can transform into an opportunity if the leadership plays to the company’s strength.
Stay tuned for more similar tech and finance industry updates that we will bring you next time.
One thought on “Microsoft Doubles AI Chip Purchases Remaining Nvidia’s Top Client For GPUs”