Forget AMD: Amazon Declares War on Nvidia by Selling Its Own AI Chips

The artificial intelligence boom has triggered one of the largest infrastructure buildouts in technology history. The largest cloud providers are on pace to spend more than $750 billion on AI infrastructure this year alone. For years, much of that spending flowed directly to Nvidia (NASDAQ:NVDA | NVDA Price Prediction), whose GPUs became the gold standard for training and running AI models. 

However, the next phase of the AI race may look very different. The biggest buyers of AI chips are increasingly designing their own, and Amazon (NASDAQ:AMZN) appears ready to take that strategy one step further by potentially selling its custom processors to outside customers.

Amazon Wants More Than Cost Savings

Bloomberg reports Amazon is exploring the possibility of selling its custom-designed AI chips to third parties, expanding beyond their current role inside Amazon Web Services (AWS).

This isn’t a surprise. During Amazon’s earnings conference call earlier this year, CEO Andy Jassy openly discussed the growing demand for AWS’s custom silicon. The company has spent years developing its Trainium and Inferentia processors to reduce reliance on third-party suppliers while lowering AI computing costs for customers.

The strategy has already reached substantial scale. Amazon’s internal chip business now generates revenue approaching $50 billion on an annualized basis. That would make it larger than the entire revenue base of companies like Advanced Micro Devices (NASDAQ:AMD), which generated $34.6 billion in 2025 revenue.

Here’s where Amazon would fall in the chip hierarchy:

Amazon’s chip operation is no longer an experimental side project. It has become a business large enough to compete with established semiconductor leaders.

Why Nvidia Should Pay Attention

Nvidia still dominates AI acceleration. The company generated $193.7 billion in fiscal 2026 data center revenue and maintains a commanding lead in AI software through its CUDA ecosystem. That software advantage remains one of Nvidia’s strongest competitive moats.

Yet Amazon has something Nvidia doesn’t: direct relationships with millions of cloud customers. AWS remains the world’s largest cloud platform, holding roughly 28% of the global cloud infrastructure market. Every enterprise building AI applications on AWS already sits inside Amazon’s ecosystem. That creates a powerful distribution advantage.

Instead of convincing customers to buy standalone hardware, Amazon can bundle AI chips, cloud services, storage, networking, and software into a single package. For customers focused on lowering AI costs, Trainium-powered services could become an attractive alternative to Nvidia-powered offerings.

Surprisingly, Amazon doesn’t need to beat Nvidia technologically to win market share. It only needs to offer enough performance at a lower total cost.

The Opportunity Is Massive — but So Is the Challenge

Granted, building chips is one thing. Building an ecosystem is another. Nvidia’s success stems from more than hardware. Developers have spent over a decade building tools and applications around CUDA. That creates switching costs that competitors continue to face.

That said, Amazon has advantages few challengers can match:

• AWS generated $128.7 billion in revenue last year.
• Amazon produced more than $139 billion in operating cash flow in 2025.
• The company can deploy custom chips across its own cloud network before expanding externally.

Those resources allow Amazon to refine products at enormous scale while funding future development. If Amazon begins selling chips directly to enterprises or other cloud providers, the competitive landscape could shift rapidly.

Key Takeaway

In short, Amazon’s custom silicon ambitions represent one of the most credible threats Nvidia has faced during the AI era.

Nvidia remains the industry leader and continues to benefit from unmatched software support and developer adoption. But Amazon’s combination of cloud dominance, deep financial resources, and growing chip expertise gives it a path few competitors possess.

For investors, the key point isn’t that Nvidia is about to lose its crown. It’s that the largest customer in the AI ecosystem is increasingly becoming a competitor. Regardless of whether Amazon ultimately sells chips directly, every AI workload that runs on Trainium instead of Nvidia hardware shifts a portion of AI economics back into Amazon’s hands.

In the end, that may be one of the most important developments in the AI market over the next decade.