Amazon CEO Andy Jassy recently unveiled the formidable scale of the company’s in-house chip business, revealing commitments exceeding $225 billion in future revenue for its Trainium artificial intelligence (AI) accelerators. This substantial figure, disclosed during Amazon’s first-quarter earnings call in April, underscores the strategic importance and rapid expansion of custom silicon within the tech giant’s ecosystem, particularly for its cloud computing arm, Amazon Web Services (AWS).
Massive Commitments Drive Future Revenue
The more than $225 billion in revenue commitments specifically for Trainium chips highlights robust external demand for Amazon’s custom AI silicon. These commitments are not merely speculative; they are backed by major industry players. Amazon’s first-quarter report detailed a commitment from OpenAI to consume approximately two gigawatts of Trainium capacity starting in 2027. Similarly, Anthropic has agreed to secure up to five gigawatts of current and future generations of Trainium chips. Beyond AI accelerators, other custom chips are also seeing significant adoption: Uber is utilizing Graviton processors for its ride-matching services, and Meta Platforms has signed on to deploy tens of millions of Graviton cores.
Jassy emphasized the competitive edge of these offerings, stating that the company’s Trainium2 chip delivers “about 30% better price-performance than comparable GPUs, and has largely sold out.” Demand is currently outpacing supply, with Trainium3, which began shipping at the start of 2026, already “nearly fully subscribed.” Even Trainium4, still over a year away from broad availability, has seen “much” of its capacity reserved, indicating strong forward-looking confidence from customers.
Internal Scale and Market Position
While the $225 billion figure represents future commitments, Amazon’s custom chip unit is already a substantial operation. Comprising Graviton processors, Trainium AI accelerators, and Nitro networking chips, all deployed within AWS, the business currently boasts an annual revenue run rate exceeding $20 billion. This unit is experiencing “triple-digit percentage rates year over year” growth, according to Jassy. He further asserted that, as best the company can tell, its custom silicon operation is now “one of the top three data center chip businesses in the world.”
Jassy also offered a hypothetical scenario: if Amazon’s in-house chip operation were a standalone entity selling its chips to outside buyers, its annual revenue run rate would be approximately $50 billion. This hypothetical figure illustrates the immense potential and inherent value of Amazon’s silicon development capabilities, even if the company primarily deploys these chips internally to enhance AWS services.
AWS Acceleration and Strategic Investments
The momentum in Amazon’s chip business is intrinsically linked to the accelerating performance of AWS, the company’s primary profit engine. AWS revenue growth accelerated to 28% in the first quarter of 2026, reaching $37.6 billion. Jassy noted this as the segment’s “fastest pace in 15 quarters,” following growth rates of 24% in the fourth quarter and 20% for all of 2025. This robust growth translated into significant profitability, with AWS generating $14.2 billion of operating income in the first quarter, a 23% increase from $11.5 billion a year earlier.
However, this expansion comes with substantial capital expenditure. Amazon anticipates approximately $200 billion in capital expenditures across the company in 2026. These aggressive AI investments have impacted the company’s free cash flow, which fell to $1.2 billion for the trailing 12 months, down from $25.9 billion a year earlier. The primary risk associated with this spending is the potential for AI computing demand to cool before these massive investments yield their full returns, which could affect Amazon’s profits and stock performance.
Competitive Landscape and Valuation
Despite its rapid growth and significant commitments, Amazon’s chip business operates in a market dominated by established players like Nvidia. Amazon itself remains a major Nvidia customer, with plans to deploy more than 1 million Nvidia GPUs starting in 2026, as announced in the same first-quarter report that highlighted Trainium’s success. Trainium’s core selling proposition is its cost per unit of computing, and Amazon offers its custom chips exclusively through AWS, rather than as standalone products on the open market.
From an investment perspective, Amazon’s stock, trading at approximately $255 per share as of this writing, reflects a valuation of about 30 times earnings. This multiple would be somewhat higher if a $16.8 billion pre-tax gain from the company’s Anthropic investment, booked in the first quarter, were excluded. Shares have seen a modest increase of around 10% year-to-date, even as AWS accelerates. Analysts suggest that the burgeoning chip business, with its $20 billion run rate, triple-digit growth, and $225 billion in commitments, strengthens the investment case for Amazon, particularly given the accelerating performance of its most profitable segment.
