Qualcomm Predicts $150 Billion Data Center Chip Revenue by 2029, Secures Microsoft and Meta Orders

Qualcomm has announced that its data center business is on track to generate $150 billion in revenue by 2029, as the chipmaker aggressively expands beyond its core smartphone chip business. The news sent Qualcomm shares up more than 12% in after-hours trading.

Chief Financial Officer Akash Palkhiwala told investors that Qualcomm’s data center business would reach $50 billion in revenue by fiscal 2027, with $10 billion coming from new custom chip customers. The company also raised its long-term revenue outlook, now projecting $400 billion in total chip revenue from non-handset segments by 2029 — up sharply from a prior forecast of $220 billion. By that point, smartphone chips are expected to account for just one-third of Qualcomm’s total chip revenue.

“We will truly achieve business diversification,” Palkhiwala said.

Arm, which provides the underlying technology for multiple Qualcomm chip families, saw its stock rise 5% following the announcement.

Earlier in the day, Qualcomm revealed that Microsoft and Meta will adopt its new AI chips. Additionally, the company is developing custom chips for two other unnamed hyperscale cloud providers. The chipmaker introduced a new category it calls High Bandwidth Compute (HBC) chips, which leverage commodity memory chips — the same kind used in phones and laptops — rather than the expensive HBM high-bandwidth memory used by Nvidia or the SRAM employed by Cerebras Systems.

Microsoft will use Qualcomm’s new chip category, while Meta plans to deploy the Dragonfly C1000, Qualcomm’s new central processor designed specifically for AI data centers, marking the company’s entry into the cloud CPU market currently contested by Arm and Nvidia.

Tony Pialis, head of Qualcomm’s data center business, said: “With our performance advantage on a per-unit-cost basis, we deliver enormous value to the entire industry.”

Qualcomm has also secured custom chip orders from two major hyperscale cloud providers, with revenue expected to materialize within the current calendar year. Pialis declined to name the customers, noting: “We don’t need to actively reach out to hyperscale cloud providers at all — they come to us looking for cooperation.”

Qualcomm’s push into AI chips comes as the smartphone market faces mounting pressure. Surging demand for AI infrastructure has strained memory chip supply, squeezing margins across the mobile supply chain. Meanwhile, key customers like Apple and Samsung are increasingly developing their own chips, eroding Qualcomm’s traditional revenue base.

The company announced in April that it plans to deliver data center processors and various AI chips to market by the end of this year. Qualcomm is simultaneously developing three types of chips for customers: central processors, AI inference accelerators, and application-specific integrated circuits (ASICs) — a segment where competitors Broadcom and Marvell have already achieved rapid growth.

Bank of America analysts cautioned in a note to clients that while Qualcomm is re-entering the high-growth AI chip market, it faces intensely fierce competition from established giants including Nvidia, upstart Cerebras, and custom chip lines from Amazon’s Graviton and Google’s Axion families.