Amazon CEO Jassy: AI inference costs need to fall to about one-tenth of current levels
In an a16z interview, Amazon CEO Andy Jassy said AI inference costs are still an order of magnitude too high, and that for AI to be widely used in enterprise software, costs need to fall to about one-tenth of current levels. He said Amazon plans to spend $220 billion in capital expenditure this year, that its main compute supply for 2026 is essentially fully booked, and that most of 2027 has also been reserved; the in-house Trainium chip offers about 30% to 40% better price-performance than comparable processors, and has already secured more than $225 billion in future revenue commitments.
Why read this
AI score 52
Jassy offered the judgment that inference costs need to drop by an order of magnitude, and explained his chip strategy by comparing it with the history of Amazon's in-house chips.
Topics
Body · AI translation
Amazon CEO Jassy: To Make AI Widely Used in Enterprise Software, Costs Need to Drop to One-Tenth of Current Levels
2026-10-24 09:51 IT Home - Qingyuan
Thanks to IT Home user Wu Ying Ge's Left Hand for the tip!
IT Home, October 24 news: Amazon CEO Andy Jassy believes that AI inference costs are still too high, and that to make AI widely used in enterprise software, costs need to drop to about one-tenth of current levels, and that Amazon's custom chips are expected to help reduce costs.
In an interview released by venture capital firm a16z on the 23rd local time, Jassy revealed: "In my view, for AI to be adopted as widely as possible, inference costs are still an order of magnitude too high."
Amazon plans to invest $220 billion (IT Home note: about 1.47 trillion yuan at current exchange rates) in capital expenditure this year, yet still cannot build enough infrastructure to meet AI demand. Jassy revealed that the main computing supply quota for 2026 is basically fully booked, most of 2027 has also been reserved by customers, and a considerable portion of 2028 has also sold out. In his view, with market demand so strong, even if Amazon invests a large amount more, it will not be wasted.
The main factor constraining enterprise adoption of AI is inference cost. Every time an AI model answers a question or completes a task, it consumes computing power. If the benefits of introducing AI are not enough to offset higher operating expenses, enterprises have little incentive to restructure their existing software.
Jassy compared this to Amazon's experience years ago. At the time, Intel dominated the cloud server processor market. Amazon wanted higher computing performance at the same cost, but found it difficult to persuade Intel to prioritize this demand, and ultimately decided to develop the Graviton processor itself. In his view, "the same story is playing out again in AI."
The AI chip market is currently dominated by Nvidia. Amazon has launched its custom Trainium chip, hoping to offer customers a lower-cost option. Jassy did not directly name Nvidia in the interview, only stressing that the market's dominant supplier remains an important partner for Amazon.
According to the introduction, Trainium offers about 30% to 40% higher price-performance than comparable processors, so customers can get more computing power for the same money. "Our chip business plays an important role in reducing inference costs."
Amazon previously disclosed that Trainium has secured more than $225 billion (about 1.5 trillion yuan at current exchange rates) in future revenue commitments, including agreements with Anthropic and OpenAI.