At a Wednesday investor briefing, Qualcomm said Microsoft and Meta will use its new artificial‑intelligence chips, and that it will manufacture custom processors for two unnamed cloud‑service giants. The San Diego‑based company is aiming to establish a foothold in the data‑center AI market, which is currently dominated by Nvidia.
Why the shift?
Qualcomm’s move is driven in part by mounting pressure in the smartphone market: global demand for memory chips is being pushed up by the rapid expansion of AI infrastructure, while major customers such as Apple and Samsung are increasingly adopting in‑house chip designs. Those trends have dampened growth prospects in the phone segment, prompting Qualcomm to pursue alternative markets.
Products and customer choices
Qualcomm refers to the new chip family as "High Bandwidth Compute" (HBC), highlighting its cost‑effectiveness. Microsoft will deploy solutions from the HBC family; according to Qualcomm these products rely on cost‑efficient memory chips, in contrast to Nvidia’s expensive, high‑bandwidth approach and Cerebras Systems’ SRAM‑based solutions. Meta has selected a data‑center CPU designed specifically for that environment: the Dragonfly C1000.
Qualcomm also said revenues from the two unnamed cloud customers under contract for custom chips are expected to begin before the end of the current calendar year.
Competitive landscape and outlook
The data‑center AI chip market is crowded: Nvidia and Cerebras are established players, while Amazon (Graviton) and Google (Axion) bring proprietary solutions, and Arm Holdings is also active. Bank of America analysts estimate Qualcomm’s data‑center unit could reach roughly $2–5 billion in annual revenue by the 2027–2028 fiscal year.
Qualcomm is developing three types of chips: central processors, inference accelerators, and application‑specific integrated circuits (ASICs). The ASIC segment has previously seen activity from rivals such as Broadcom and Marvell.
Why it matters
Qualcomm’s entry adds another competitor to a market that has been largely shaped by Nvidia and a handful of custom‑chip vendors. Its cost‑focused strategy and the announced customer wins — notably Microsoft and Meta — indicate cloud providers and big tech firms are seeking more alternatives for AI infrastructure. The financial impact will become clearer over the coming years, but Bank of America’s projection suggests a potential multi‑billion‑dollar revenue stream for Qualcomm.
