Latest with AI
Wednesday, April 15, 2026
AI Business

Amazon CEO: Our Custom Chip Business Is ‘On Fire’ and Ready to Challenge Nvidia

Amazon CEO Andy Jassy revealed the company's custom chip business has surpassed a $20 billion annual revenue run rate with triple-digit growth, declaring it ready to challenge Nvidia's dominance in AI computing.

Picsum ID: 507

Amazon CEO Andy Jassy declared his company’s custom chip division “on fire” in his annual shareholder letter, claiming it has reached a scale that positions it as a direct competitor to Nvidia, Intel, and AMD in the semiconductor industry. The disclosure marks one of the most striking moments in Amazon’s evolving identity as a hardware company — not just a cloud provider.

A $20 Billion Chip Business With Triple-Digit Growth

Jassy’s letter revealed that Amazon’s combined chip business — spanning its Graviton CPUs, Trainium AI accelerators, and Nitro networking architecture — has surpassed a $20 billion annual revenue run rate, with triple-digit percentage growth year over year. He went further, arguing that if the chip division operated as a standalone company selling to both AWS and external customers, its annual run rate would reach approximately $50 billion.

The letter pointed to Trainium3, which began shipping in early 2026, as delivering 30 to 40 percent better price-performance than its predecessor. Capacity for Trainium4 — still roughly 18 months from availability — is already nearly sold out. Amazon has secured Trainium deals with OpenAI, Anthropic, and Apple, demonstrating that even the most powerful AI labs are willing to diversify away from Nvidia.

A Challenge to Nvidia’s Dominance

“Virtually all AI thus far has been done on NVIDIA chips, but a new shift has started,” Jassy wrote, drawing a direct comparison to how Amazon’s Graviton processor eroded Intel’s dominance in server CPUs after launching in 2018. He noted that Graviton is now “used expansively by 98 percent of the top 1,000 EC2 customers,” with the same trajectory beginning for Trainium in AI workloads.

AWS AI Revenue at $15 Billion Run Rate

Alongside the chip disclosures, Jassy revealed that AWS’s AI revenue has reached a $15 billion annual run rate. He framed the company’s planned $200 billion in capital expenditures for 2026 not as spending without conviction but as a calculated bet — noting that Trainium could save Amazon “tens of billions of capex dollars per year” and deliver several hundred basis points of operating margin advantage compared to third-party alternatives. “It’s hard to overstate my optimism for what’s ahead,” he wrote.