Nvidia built a near-monopoly on AI computing over the past four years. Everything from ChatGPT to corporate AI tools runs on its chips. That position is now facing its most serious challenge, and today's Microsoft announcement is the clearest evidence of it.
Microsoft has officially committed to deploying AMD's Helios system inside its Azure cloud. Helios is AMD's first complete, rack-scale AI platform: it bundles processors, networking chips, and software into one integrated system designed to run the large AI models that power services like Copilot and Azure AI. Shipments begin in the second half of 2026. Meta, OpenAI, and Oracle are also signed as early customers.
Nvidia still holds roughly 80 to 86% of the AI chip market by revenue, and its fiscal 2026 revenue hit $215.9 billion, up 65% year over year. AMD's data center revenue, by comparison, was $5.8 billion in Q1 2026 alone, but that is still about one thirteenth of Nvidia's pace. The gap is real. What is changing is the direction of travel.
The reason large companies are adding AMD is not primarily performance. It is risk management. When one supplier controls your most critical input, that supplier sets the price, controls the timeline, and decides who gets capacity first. Microsoft, Meta, and OpenAI are all spending tens of billions on AI infrastructure each year. They need leverage.
AMD has been aggressive in building that leverage. It signed a deal with OpenAI last year to deploy six gigawatts of computing capacity and offered significant equity incentives to lock in the relationship. A separate six-gigawatt deal with Meta followed. Now Microsoft. The customer list is becoming hard to ignore.
The Anthropic angle is still unconfirmed but worth watching. A code file from a senior AMD software executive listed Anthropic as a customer, and AMD gave Anthropic its highest internal priority rating, on par with Meta. Anthropic already uses Nvidia GPUs, Google's custom chips, and Amazon's custom processors. If AMD is added, Anthropic would have the most diversified chip supply of any major AI company. Anthropic's annualized revenue has reportedly surged past $30 billion, and at that scale, supply concentration is a real operational risk.
The software gap is the honest caveat here. AMD's hardware has become cost-competitive for running mid-sized AI models. For the largest models and the most demanding workloads, Nvidia still leads in software maturity. That is precisely what Anthropic's engineers are evaluating. AMD's software stack has improved significantly but has not fully closed the gap.
For business operators who use AI tools through cloud platforms, the near-term impact is indirect. Your Azure or AWS interface does not change. But over the next two to three years, more chip competition means cloud providers gain pricing power over their suppliers, and that eventually flows to customers as lower costs or more available capacity. Companies that have been rationed on AI computing resources because of supply constraints stand to benefit most.
The bigger structural point: AI infrastructure used to be about picking the best model. Increasingly, it is about who controls the hardware the models run on. The companies that built dependency on a single chip supplier are now methodically reducing it. That shift is now confirmed, not speculative.