Amazon Custom Chips Enter Global Top Three
According to Amazon CEO Andy Jassy, the company’s in-house custom chip business has grown into a major force in the semiconductor industry, ranking among the world’s top three data center chip businesses. If operated as an independent external business, the segment could potentially generate annual revenue of up to $50 billion.
Amazon’s custom chip portfolio includes the Graviton processors, Trainium AI accelerators, and Nitro networking chips, which are designed to enhance the performance and efficiency of its AWS cloud infrastructure. The business has already surpassed $20 billion in annual revenue and has maintained triple-digit year-over-year growth. In the first quarter alone, revenue increased nearly 40% sequentially.
Among these products, Trainium AI chips have gained significant market traction, securing more than $225 billion in long-term customer commitments. Major technology companies have signed large-scale agreements for Trainium capacity, with OpenAI and Anthropic committing to 2GW and 5GW of computing capacity, respectively. Uber has adopted Graviton processors to improve its ride-matching services, while Meta has signed agreements to deploy tens of millions of Graviton-based cores.
Demand for Amazon’s AI chips continues to exceed supply. Jassy said that Trainium2 delivers approximately 30% better price-performance compared with comparable GPUs and has nearly sold out. Meanwhile, Trainium3, scheduled for shipment in early 2026, is already almost fully reserved, while most capacity for the upcoming Trainium4 has also been secured ahead of its official launch.
The rapid expansion of Amazon’s chip business is closely tied to the strong recovery of AWS. Data shows that AWS revenue increased 20% year over year in 2025, with growth accelerating to 24% in the fourth quarter. In the first quarter of 2026, AWS revenue rose another 28% to $37.6 billion, marking its fastest growth rate in 15 quarters. Operating income for the quarter reached $14.2 billion, up 23% year over year.
However, Amazon’s aggressive AI infrastructure expansion also brings rising costs and operational risks. The company expects total capital expenditures to reach $200 billion in 2026, as it continues to invest heavily in artificial intelligence infrastructure. These investments have significantly reduced free cash flow, which fell from $25.9 billion in the previous year to just $1.2 billion over the past 12 months. If AI computing demand slows before these investments generate sufficient returns, Amazon’s profitability and stock performance could face pressure.
