Amazon, Alphabet, and Microsoft are designing their own AI chips while continuing to buy record amounts of Nvidia processors

Amazon's custom chip business (Graviton, Trainium, and Nitro processors) topped a $20 billion annual revenue run rate in Q1 2026, and CEO Andy Jassy said the operation would generate a $50 billion annual revenue run rate if it sold chips to third parties like other chip companies do. Yet Amazon plans to spend about $200 billion on capital expenditures in 2026, much of it on Nvidia GPUs for AWS.

Alphabet's eighth-generation TPU systems are now available outside Google's walls: Blackstone announced a joint venture to offer TPUs as a rentable cloud service with an initial $5 billion commitment and plans to bring 500 megawatts of capacity online in 2027. Simultaneously, Google signed a multiyear SpaceX cloud deal involving access to about 110,000 Nvidia GPUs.

Microsoft's Maia 200 accelerator recently went live in some data centers, but the vast majority of AI work in Azure still runs on Nvidia GPUs. Microsoft expects to invest roughly $190 billion in capital expenditures during calendar 2026.

Amazon, Alphabet, Microsoft, and Meta are on track to spend roughly $725 billion on capital expenditures in 2026—up about 77% from last year—while Nvidia's data center revenue rose 92% in fiscal Q1 2027 (period ended April 26, 2026) to $81.6 billion, with hyperscalers making up about half of that business.