The Massive Demand for GPUs
The NVIDIA chips, particularly the H100 and H20 models, are essential for artificial intelligence technologies. They power data centers running models from OpenAI, in which Microsoft is a leading partner and investor. These GPUs are used to train and deploy increasingly sophisticated AI systems, meeting the rising demand for services such as ChatGPT and other AI-based tools.
With this purchase, Microsoft spent approximately $30 billion in 2024 to bolster its AI infrastructure. In comparison, Meta purchased 224,000 GPUs during the same timeframe, while Amazon and Google acquired 196,000 and 169,000 GPUs respectively.
A Well-Oiled Strategy
Microsoft’s approach goes beyond just bulk buying. The company is also developing its own in-house chips, known as Maia, announced at the end of 2023. The goal is to decrease its long-term dependence on NVIDIA, while still maintaining a competitive edge over its rivals. This strategy allows Microsoft to diversify its technological capabilities while sustaining a dominant position in the AI sector.
Rising Competition
Other tech giants are not sitting idle. Meta is also developing its own chips (MTIA), and Amazon is betting on its Trainium processors. Yet even so, their investments still fall short of Microsoft’s 2024 expenditures, which doubled down impressively.
In China, companies like ByteDance and Tencent each purchased around 230,000 GPUs, but their options are constrained by U.S. restrictions on exporting certain advanced technologies.
AI as a Top Priority
With such a scramble for GPUs, it’s clear that AI has become a central focus for all major tech companies. By purchasing twice as many GPUs as its closest competitors, Microsoft has positioned itself as a pivotal leader. Its partnership with OpenAI and its internal developments indicate it is already preparing for the next phases of this technological revolution. It remains to be seen if this lead will hold against competitors who are actively gearing up.