cd /news/ai-chips/amazon-in-talks-to-sell-custom-ai-ch… · home topics ai-chips article
[ARTICLE · art-32798] src=cryptobriefing.com ↗ pub= topic=ai-chips verified=true sentiment=↑ positive

Amazon in talks to sell custom AI chips to data centers

Amazon is negotiating to sell its custom Trainium AI processors to external companies for use in their own data centers, challenging Nvidia's dominance in AI hardware. The move follows Amazon's in-house chip business reaching a $20 billion annual revenue run rate, with Trainium3 offering up to four times the performance of its predecessor. Uber has already adopted Trainium3, signaling growing external demand for Amazon's silicon.

read2 min views2 publishedJun 18, 2026

The cloud giant's Trainium processors could soon land in rival data centers, challenging Nvidia's grip on AI hardware

Amazon isn’t just building AI chips for itself anymore. The company is actively negotiating to sell its custom Trainium processors to outside companies for use in their own data centers, a move that would transform Amazon from a cloud-services landlord into a direct competitor in the AI chip market.

The shift is significant because it means Amazon’s silicon, developed in-house by its Annapurna Labs division, would no longer live exclusively inside Amazon Web Services infrastructure. Other companies could buy the chips outright and run them on their own hardware.

The business behind the silicon #

Amazon’s chip ambitions aren’t exactly a side hustle. CEO Andy Jassy disclosed in April that the company’s in-house chip business had surpassed a $20 billion annual revenue run rate.

The latest generation processor, Trainium3, launched in late 2025 and represents a meaningful leap forward. Amazon claims it delivers up to four times the performance of its predecessor, Trainium2, while cutting training and inference costs by as much as 50% compared to conventional GPUs.

Uber is among the first external partners to adopt Trainium3, a deal confirmed in April.

Why Amazon is making this move now #

The roots of this strategy go back a decade. Amazon acquired Annapurna Labs in 2015, and the division began developing custom AI silicon around 2019. The early chips were modest, primarily designed to handle inference workloads within AWS. But with each generation, Amazon has pushed into training, the more computationally demanding side of AI, and scaled up significantly.

Anthropic’s commitment to deploying over one million Trainium chips in AWS data centers by the end of 2025, through a project internally called Rainier, demonstrated that Amazon’s silicon could handle frontier AI workloads at massive scale.

This broader push also reflects a trend among hyperscale cloud providers. Google has its TPUs, Microsoft has invested heavily in custom silicon through partnerships, and now Amazon is taking the additional step of selling its chips outside its own ecosystem.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our

Editorial Policy.

── more in #ai-chips 4 stories · sorted by recency
── more on @amazon 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/amazon-in-talks-to-s…] indexed:0 read:2min 2026-06-18 ·