According to sources quoted by the American news outlet Business Insider, Microsoft intends to expand its GPU inventory to 1.8 million, mostly from NVIDIA. Having additional processors on hand will allow Microsoft to release AI solutions that are more efficient, faster, and cost-effective.
The source does not specify particular potential applications for these chips, but obtaining a huge number of chips allows Microsoft to widely deploy them across its own products, such as cloud services and consumer electronics.
According to insiders mentioned in the same report, Microsoft intends to invest USD 100 billion in GPUs and data centers by 2027 to improve its present infrastructure.
Microsoft's large stockpile of AI chips demonstrates the company's efforts to preserve a competitive advantage in the AI industry, where substantial computing capacity is critical for innovation.
NVIDIA, on the other hand, recently announced that the AI machine it is developing with Microsoft will run on Microsoft's Azure cloud platform and will use tens of thousands of NVIDIA GPUs, including the H100 and A100 chips.
NVIDIA declined to disclose the contract value for this partnership. However, industry insiders quoted in the paper estimate that each A100 chip costs between USD 10,000 and $12,000, whereas the H100 is substantially more expensive.
Microsoft is also in the process of designing the next generation of chips. Not only is Microsoft working to lessen its dependency on NVIDIA, but OpenAI, Tesla, Google, Amazon, and Meta are also investing in their own AI accelerator processors. These businesses are likely to compete against NVIDIA's flagship H100 AI accelerator processors.