Back to Headlines
Tech
Jun 18, 2026
Analyzed by GPT OSS 120B

Amazon Aims to Challenge Nvidia by Selling Its Trainium AI Chips

AI Summary
Amazon Web Services is exploring the sale of its in‑house Trainium AI chips to third‑party data centers, a move that could generate a $50 billion run‑rate and directly contest Nvidia’s dominance. The strategy hinges on excess capacity, partnerships with TSMC, and a broader push to monetize Amazon’s AI hardware.

Amazon’s Push to Turn Trainium Into a Commercial AI Chip

Amazon Web Services Peter DeSantis told Bloomberg that the cloud unit is in early‑stage talks to sell its proprietary AI accelerator, Trainium, to other companies for data‑center deployments. The initiative stems from CEO Andy Jassy's April shareholder letter, where he hinted that the chips were “so coveted” they might be sold externally.

AWS Announces Talks to Sell Trainium to Third‑Party Data Centers

The discussions are still nascent, according to TechCrunch, and no specific buyers have been disclosed. DeSantis declined to name potential customers, while AWS spokesperson Doron Aronson confirmed that “it’s quite possible we’ll sell racks of them to third parties in the future.”

  • Current chip: Trainium – sold out instantly upon release.
  • Next generation: Trainium4 – not expected for more than a year.
  • Manufacturing partner: TSMC, now the foundry’s largest customer.

Potential $50 Billion Revenue Run‑Rate vs Nvidia’s $326 Billion

Jassy projected that a standalone chip business could reach an annual run‑rate of $50 billion. By comparison, Nvidia is operating on a $326 billion revenue run‑rate, making Amazon’s ambition comparable to Intel’s annual sales.

  • Amazon’s projected chip revenue: $50 billion
  • Nvidia’s current AI hardware revenue: $326 billion
  • Intel’s annual revenue (for context): roughly $70 billion

What Amazon’s Chip Sales Could Mean for the AI Hardware Landscape

Selling Trainium would shift AWS from a pure‑play cloud services model to a hybrid hardware‑software provider. The revenue would still be largely indirect, as AWS monetizes AI token processing, storage, security, networking, and monitoring services. However, a broader chip distribution could accelerate adoption of Amazon’s software stack, potentially eroding Nvidia’s market share in data‑center accelerators.

Future Outlook: Amazon’s Path to a Viable Nvidia Challenger

If AWS can generate surplus capacity through TSMC and manage waiting‑list pressures, it could establish a $50 billion niche that forces Nvidia to defend its lead more aggressively. The next 12‑18 months will reveal whether Amazon can scale production fast enough to compete with Nvidia’s entrenched ecosystem and whether third‑party customers will adopt Trainium over established GPU solutions.