Microsoft's CEO, Satya Nadella, reveals that despite launching its own AI chips, the company will continue to purchase AI chips from Nvidia and AMD. The Maia 200, Microsoft's latest AI chip, is designed to be an 'AI inference powerhouse', optimized for running AI models in production. It outperforms Amazon's Trainium and Google's Tensor Processing Units (TPU). The supply crunch for Nvidia's chips is a significant challenge for cloud giants, leading them to develop their own chip designs. However, Nadella emphasizes that Microsoft's partnership with Nvidia and AMD is strong, and the company values innovation from both partners. The Maia 200 will be used by Microsoft's Superintelligence team, which is developing the company's own frontier AI models, potentially reducing reliance on external model makers like OpenAI and Anthropic. The chip will also support OpenAI's models on Microsoft's Azure cloud platform. Despite Microsoft's advancements, securing access to the most advanced AI hardware remains a challenge, and the company's internal teams are also affected by this issue.
