Qualcomm shares more details on new data center AI chips

In short: Qualcomm detailed its planned lineup of data center chips, aiming to compete with Nvidia in the fast growing market for AI computing.

What happened

Qualcomm, best known for smartphone chips, used its investor day in New York to lay out more of its data center chip roadmap. It said the lineup will include an “AI accelerator” (a chip built to run AI tasks faster), plus CPUs (general purpose computer brains), custom chips, and connectivity chips.

A key part of the presentation was a design Qualcomm calls high-bandwidth compute, or HBC. The company says it puts memory closer to the computing part of the chip, an approach known as near-memory compute. Think of it like moving a pantry next to the stove so a cook spends less time walking back and forth.

Qualcomm said this can reduce the time and energy spent moving data between the processor and memory. That matters because memory chips, including high-bandwidth memory or HBM (a type of very fast memory often used in AI servers), have become a bottleneck for building AI systems.

Qualcomm chief executive Cristiano Amon also said the company is looking at selling these data center products in China. He told Nikkei Asia that Qualcomm may design chips for Chinese customers that follow US export controls. China made up 46 percent of Qualcomm’s revenue in 2025, mostly from smartphone chips.

Why it matters

AI services, from chatbots to image tools, run on large data centers filled with expensive chips. If Qualcomm can make data center chips that use less power and rely less on scarce memory parts, it could lower costs and add competition to a market where Nvidia is the leading supplier.

Source: Financial Times