Microsoft Outspends Rivals on Massive Nvidia Chip Purchase

Microsoft has significantly outpaced its tech rivals in its acquisition of Nvidia's advanced AI chips, highlighting the company's substantial investment in artificial intelligence infrastructure, reports the Financial Times.

Analysts at Omdia estimate that Microsoft purchased 485,000 of Nvidia's "Hopper" chips this year, more than double the number acquired by any other major US or Chinese tech company. This substantial purchase puts Microsoft well ahead of Meta, its closest competitor in the US, which acquired 224,000 Hopper chips.

This dominance in chip acquisition has given Microsoft a substantial advantage in the race to develop next-generation AI systems. The company's Azure cloud infrastructure played a key role in training OpenAI's latest language model, o1, as Microsoft and OpenAI compete with Google, Anthropic, and Elon Musk's xAI for AI dominance.

Omdia estimates that ByteDance and Tencent, two of Nvidia's largest Chinese customers, each acquired around 230,000 chips this year, including the less powerful H20 model. Amazon and Google, both increasing their use of custom AI chips to reduce reliance on Nvidia, acquired 196,000 and 169,000 Hopper chips, respectively.

The soaring demand for Nvidia's most advanced GPUs, fueled by the rapid expansion of AI, has driven a significant increase in Nvidia's valuation this year. However, recent concerns about slower growth, competition from companies' custom AI chips, and potential disruptions in the Chinese market have led to some moderation in Nvidia's stock price.

Microsoft, which has invested $13 billion in OpenAI, has been particularly aggressive in expanding its data center infrastructure to support both its own AI services, such as Copilot, and external customer needs through Azure.

Microsoft's Nvidia chip orders this year are more than triple the number of the same generation of chips it acquired in 2023.

"Good data centre infrastructure, they’re very complex, capital-intensive projects," Alistair Speirs, Microsoft’s senior director of Azure Global Infrastructure, told the Financial Times. "They take multi-years of planning. And so forecasting where our growth will be with a little bit of buffer is important."

Despite Microsoft's substantial investment in its own AI accelerator, the company still relies heavily on Nvidia's chips. Microsoft has deployed only about 200,000 of its own Maia chips this year, while its use of Nvidia chips requires significant investment in complementary technologies to provide a differentiated service to customers.

"To build the AI infrastructure, in our experience, is not just about having the best chip, it’s also about having the right storage components, the right infrastructure, the right software layer, the right host management layer, error correction and all these other components to build that system," Speirs added.