Microsoft's CTO Aims to Ditch NVIDIA and AMD for In-House AI Processors

Microsoft's CTO Aims to Shift Majority of AI Workloads from NVIDIA and AMD GPUs In a significant move, Microsoft's Chief Technology Officer (CTO) Kevin Scott has expressed the company's intention to transition most of its Artificial Intelligence (AI) workloads from Graphics Processing Units (GPUs) from NVIDIA and AMD to in-house chips. Speaking during a fireside chat with CNBC on Wednesday, Scott revealed that Microsoft hopes to use its custom accelerators for the majority of its datacenter workloads. Timeline: Microsoft has been purchasing large quantities of GPUs from both NVIDIA and AMD. The company is reportedly working on a second-generation Maia accelerator, set to be released next year. This new chip will offer improved performance in compute, memory, and interconnect capabilities. Immediate Impact: The shift towards custom chips will likely have significant implications for the tech industry. Microsoft's decision may prompt other cloud providers to reassess their GPU strategies. The move also raises questions about the long-term viability of NVIDIA and AMD's dominance in the AI market. Background Context: Microsoft has been actively developing its own custom chips, including the Cobalt CPU and Maia accelerator. These efforts are part of a broader trend towards companies creating their own specialized hardware to optimize performance and reduce costs. What Happens Next: As Microsoft continues to develop its in-house chip technology, we can expect to see increased competition in the AI market. The release of the second-generation Maia accelerator next year will be closely watched by industry experts. This development may also lead to a reevaluation of NVIDIA and AMD's business strategies. Microsoft's CTO Kevin Scott emphasized the importance of performance per dollar for hyperscale cloud providers like Microsoft. By developing its own custom chips, the company aims to meet demand while reducing costs. As the tech landscape continues to evolve, one thing is clear: the future of AI processing will be shaped by innovative companies willing to push boundaries and challenge traditional industry leaders. *This story is developing. Information compiled from Hardware reporting.*