Amazon and Google Challenge Nvidia With 81% AI Chip Share — Goldman Sachs Nvidia Stock Analysis
Goldman Sachs Nvidia stock analysis turns on one number: Nvidia controls an estimated 81% of the AI data center chip market, even as Amazon, Google, Meta Platforms and Microsoft keep building their own processors. For investors, that means the demand pool is still enormous, but more of it is now being contested by the company’s largest customers.
81% And Nvidia’s AI Lead
81% is the share IDC assigns to Nvidia in the AI data center chip market, a level built on graphics processing units that split complex calculations into smaller parts and run them at the same time. That architecture has fit the work behind large language models, which is why major hyperscalers and AI companies have leaned on Nvidia hardware for years.
$1 trillion is Nvidia’s forecast for total sales across its Blackwell and Vera Rubin architectures in 2026 and 2027. That projection shows the company is still expecting a very large revenue pool from AI infrastructure, even as its biggest customers widen the field of suppliers around it.
Amazon Trainium Reaches $20 Billion
40% sequential growth is what Amazon’s chip business posted in the first quarter of 2026, lifting its annual revenue run rate to more than $20 billion. Amazon says the run rate would be closer to $50 billion if sales of chips to itself for use in AWS data centers were included, a sign that internal demand remains a major part of the business.
$225 billion is Amazon’s purchase commitment for Trainium AI chips, and demand for the chips is fully booked. Amazon launched Trainium in December 2020, and the processors are now being deployed by Anthropic, OpenAI, Uber and Meta Platforms, showing that the business has moved beyond internal use into external sales.
Google TPU And Meta Demand
2015 was the year Google launched the first generation of its Tensor Processing Unit, and the company now has deals in place with Meta Platforms and Anthropic for TPU deployment. Sundar Pichai sees the TPU business as one of Alphabet’s key growth drivers, putting a formal growth target around a chip line that once existed mainly to support Google’s own workload.
Meta Platforms uses Amazon’s in-house Graviton CPU to support agentic AI applications, which adds another layer to the competitive pressure on Nvidia. Amazon and Google are no longer only customers buying chips; they are also selling custom silicon to third parties, and that shift is what can change supplier leverage inside AI infrastructure. If those internal chip businesses keep expanding, Nvidia can still sell into a huge market, but it will be selling into one that is becoming less dependent on a single vendor.
Across 2026 and 2027, the most important watchpoint is whether Nvidia’s $1 trillion sales forecast keeps pace with the growing deployment of Trainium and TPU chips at Amazon, Google and their partners.
