Amazon CEO Andy Jassy said artificial intelligence infrastructure remains too expensive for widespread enterprise use, but that’s exactly what Amazon Web Services aims to change by developing its own chips and slashing inference costs.
Speaking on CNBC’s Squawk Box, Jassy pushed back against speculation that recent advances in artificial intelligence model efficiency, such as those introduced by DeepSeek, would reduce the need for infrastructure buildout.
“We have very high demand,” he said. “I don’t see us attenuating our building centers right now.”
Jassy emphasized that AWS is experiencing no drop-off in AI infrastructure demand, despite macroeconomic uncertainty and the looming threat of tariffs. He explained that while more efficient models are welcome, the challenges of AI infrastructure are deeper and more persistent.
“If you’re building frontier models like we are, you’re working on a lot of the same problems,” Jassy said. “The lower that we can make the cost of AI, the more customers are going to use it.”
Lowering cost unlocks greater customer spend
Jassy drew parallels between today’s AI shift and the early days of AWS. Lower costs per unit of compute, he said, often lead to increased customer spending—not less—as companies find new ways to innovate.
“It allows them to save money in what they’re building, but they don’t spend less,” he said. “It unleashes them to do more innovation.”
According to Jassy, there are two primary levers for reducing AI costs: the chips themselves and the cost of inference, the process of making predictions using trained models. While training dominates spending today, inference will become the cost center at scale.
AWS has responded by building its own custom AI chips, which Jassy says deliver 30% to 40% better price performance than current GPU-based instances. He also noted that reducing inference costs involves advances in hardware and software techniques.
Jassy framed the mission clearly: “If you sat in the meetings with the AWS team right now, they feel like it is their responsibility and their mission to make the cost of AI meaningfully less than today.”
That push to lower AI costs is a potential catalyst for certain segments in the crypto sector. Notably, developers have long explored AI use cases but often hit a wall due to infrastructure and cost constraints. With more affordable technology to come within years, blockchain-native AI applications, from on-chain analytics to decentralized autonomous agents, could become more feasible at scale.
