Skip to content
gentic.news — AI News Intelligence Platform
Connecting to the Living Graph…

Listen to today's AI briefing

Daily podcast — 5 min, AI-narrated summary of top stories

Amazon executive Peter DeSantis gestures while discussing Trainium AI chips, with data center servers and AWS…
Big TechScore: 87

Amazon Weighs Selling Trainium Chips, Targeting $50B Nvidia Challenge

Amazon is in early talks to sell Trainium AI chips to third parties, aiming at a $50B opportunity that would challenge Nvidia's $326B dominance.

·22h ago·3 min read··6 views·AI-Generated·Report error
Share:
Source: techcrunch.comvia techcrunch_ai, dcd_newsCorroborated
Is Amazon selling its Trainium AI chips to compete with Nvidia?

Amazon is in early talks to sell its Trainium AI chips to third-party data centers, CEO Andy Jassy says the business could be worth $50 billion annually, challenging Nvidia's $326 billion revenue run rate.

TL;DR

AWS in talks to sell Trainium to other data centers. · Andy Jassy pegs chip opportunity at $50B annually. · Nvidia's revenue run rate is $326B, dwarfing Amazon's ambition.

Amazon is in early talks to sell its Trainium AI chips to third-party data centers, AWS AI chief Peter DeSantis told Bloomberg. CEO Andy Jassy has pegged the chip business opportunity at roughly $50 billion annually — a direct challenge to Nvidia's $326 billion revenue run rate.

Key facts

  • AWS AI chief Peter DeSantis confirmed Trainium sale talks to Bloomberg.
  • Andy Jassy estimated chip business at $50B annual run rate.
  • Nvidia's current revenue run rate is $326B.
  • Current Trainium and next-gen Trainium4 capacity are sold out.
  • Nvidia recently became TSMC's largest customer, surpassing Apple.

AWS is exploring selling its custom Trainium AI chips to other companies for use in data centers, according to a Bloomberg interview with Amazon AI chief Peter DeSantis. According to TechCrunch, the talks are in early stages and AWS declined to name potential buyers. The move stems from CEO Andy Jassy's April shareholder letter, where he wrote: "If our chips business was a standalone business, and sold chips produced this year to AWS and other third parties (as other leading chips companies do), our annual run rate would be ~$50 billion."

The $50B vs. $326B math

A $50 billion chip business would make Amazon a significant player — roughly the size of Intel's annual revenue — but it would not dethrone Nvidia, which is currently on a $326 billion revenue run rate. The challenge is less about raw revenue and more about breaking Nvidia's near-monopoly on AI training silicon, where the H100 and Blackwell families dominate.

The capacity dilemma

AWS has historically resisted selling chips directly because it profits more from the full cloud stack — compute, storage, networking, monitoring — that runs on its silicon. Jassy noted in the same letter that current Trainium capacity sold out "almost instantly," as did capacity for the upcoming Trainium4, which is over a year from availability. Selling chips externally would force AWS to either prioritize outside buyers over its own cloud customers or secure additional wafer allocation from TSMC — a fab where Nvidia recently became the largest customer, surpassing Apple.

The competitive context

The announcement comes as Nvidia CEO Jensen Huang declared a new $200 billion market selling CPUs for AI, directly entering Intel and AMD territory. Amazon's move is symmetrical: it is pushing into Nvidia's core silicon business while Nvidia pushes into Amazon's core infrastructure business. AWS spokesperson Doron Aronson confirmed the company may sell chips, saying, "While we've historically declined requests to sell chips directly, Andy noted it's quite possible we'll sell racks of them to third parties in the future."

What to watch

Watch for any formal Trainium sale announcement at AWS re:Invent in December 2026, and whether Amazon can negotiate additional TSMC wafer capacity without disrupting its own cloud supply. Also monitor Nvidia's response — potential pricing pressure on Blackwell or accelerated Vera Rubin timelines.

Amazon President and CEO Andy Jassy


Source: techcrunch.com


Sources cited in this article

  1. Peter DeSantis
  2. Jassy
  3. Doron Aronson
  4. Andy
Source: gentic.news · · author= · citation.json

AI-assisted reporting. Generated by gentic.news from 5 verified sources, fact-checked against the Living Graph of 4,300+ entities. Edited by Ala SMITH.

Following this story?

Get a weekly digest with AI predictions, trends, and analysis — free.

AI Analysis

This is not just a product expansion — it is a structural pivot in AWS's business model. AWS has historically monetized AI through cloud services, not silicon sales. Selling Trainium externally means Amazon is willing to sacrifice some vertical integration profit for market share in the broader AI hardware market. The capacity constraint is the binding factor: AWS cannot currently meet its own demand, let alone external demand, without TSMC allocation. The real story is whether Amazon can secure additional wafer starts without displacing Nvidia, which is now TSMC's largest customer. If Amazon succeeds, it would create a credible second source for AI training silicon — something the industry desperately needs given Nvidia's pricing power and allocation control. The $50B figure is aspirational, not operational; it assumes AWS can scale production to match demand, which is far from guaranteed.
This story is part of
The AI Infrastructure War Shifts from Chips to Developer Tools
Nvidia's enterprise pivot and AWS's OpenAI bet collide with Cursor's quiet ascent
Compare side-by-side
Amazon vs Nvidia
Enjoyed this article?
Share:

AI Toolslive

Five one-click lenses on this article. Cached for 24h.

Pick a tool above to generate an instant lens on this article.

Related Articles

From the lab

The framework underneath this story

Every article on this site sits on top of one engine and one framework — both built by the lab.

More in Big Tech

View all