Amazon’s custom chips are pointed to as a major scale business, with one analysis citing a $25 billion run rate

Amazon’s push into custom silicon for cloud computing has become a recurring theme in industry coverage, and a fresh market commentary now puts a specific scale figure on that effort. In a recent write-up published by Yahoo Finance, the author argues that Amazon’s custom chip business has moved beyond niche status, crossing a cited $25 billion run rate.

The article frames the growth as tied to AWS demand for tailored performance and efficiency across workloads, suggesting that Amazon’s custom-built chips are increasingly being selected as cloud customers build and run data center applications. The commentary characterizes the segment as gaining in popularity, rather than remaining a mostly internal optimization.

A key feature of the report is the focus on a run-rate estimate, which is a forward-looking way of annualizing current performance trends. While the piece highlights the $25 billion level, it does not, in the information available here, provide a breakdown of the underlying revenue components or the methodology used to convert chip-related activity into an estimated annual run rate.

Because the underlying source is a market-news analysis rather than an Amazon disclosure, the company did not provide accompanying detail in the cited post about the custom-chip business’s direct financial line item. Amazon typically reports AWS results in aggregate, and it generally does not publish a standalone “custom chip” revenue figure in a way that would let outside readers verify a $25 billion number directly.

For Amazon, the strategic logic behind custom chips is straightforward, even if segment-level financials are harder to observe. By designing specialized processors and related data center components, AWS can pursue lower power consumption, improved price-performance for certain workloads, and tighter integration between compute hardware and AWS software services. Those efforts matter in a business where cloud customers often compete on both performance targets and operating costs.

Within the broader technology sector, the thrust toward custom silicon is widely viewed as both a cost-control lever and a differentiation tool. Hyperscalers including Amazon have increasingly moved beyond off-the-shelf processors, betting that tailored chips can support scale across AI training and inference as well as general compute and networking needs. If the cited run-rate estimate reflects real momentum, it would imply that hardware-level bets are becoming meaningful to AWS’s economics rather than merely incremental.

Still, important uncertainty remains. The cited article provides the headline run-rate framing but does not supply enough verifiable, company-specific detail in the material available here to confirm how much of AWS’s overall hardware-related spending (or service revenue) should be attributed to custom chips, or how quickly that allocation is changing quarter to quarter.

What to watch next is whether Amazon supplies clearer disclosure in future reporting, such as more granular commentary on custom silicon adoption, capacity expansion tied to specific chip generations, or updates that connect hardware progress to AWS performance and customer demand. Investors and customers will also be looking for evidence that AWS is using its chip strategy to win new workloads, not just optimize existing ones.