Qualcomm targets $15B in data center chip revenue by 2029; announces Meta, hyperscaler deals
Qualcomm announced a major strategic pivot at its June 24 investor day, forecasting more than $15 billion in annual data center chip revenue by fiscal 2029—a dramatic expansion from its traditional smartphone focus. The company revealed that it has already secured two unnamed 'hyperscaler' customers (industry parlance for cloud giants like Amazon, Microsoft, or Google) who will begin using Qualcomm's custom ASICs by the end of calendar 2026. Meta has also been named as a data center customer, utilizing Qualcomm's new AI accelerators and custom silicon solutions.
Qualcomm's data center strategy spans four pillars through 2028: connectivity (electrical and optical interconnects up to 800G today, 1.6T by late 2026), custom silicon for leading partners beginning in 2027, new Dragonfly C1000 server CPUs in 2028 with Qualcomm's proprietary High-Bandwidth Compute (HBC) architecture featuring up to 250 Oryon cores at 5GHz, and AI accelerators optimized for inference decoding. HBC stacks place LPDDR memory in-package atop full XPU accelerators, targeting power efficiency and bandwidth superior even to forthcoming HBM4e. Qualcomm is not pursuing the full data center stack but instead focusing initially on AI decoding—a narrower and more achievable beach.
The company also doubled its broader non-smartphone revenue forecast for 2029 from $22 billion to $40 billion, with automotive contributing $10 billion (up from $5+ billion annualized in Q2 FY2026). Qualcomm's CFO indicated $5 billion in data center revenue should arrive by fiscal 2027, with $1 billion coming from new custom chip customers. Stock surged 12% in after-hours trading on the announcement, with BofA raising its price target to $195 and Cantor Fitzgerald to $200.
For infrastructure teams focused on cost, Qualcomm's entry signals real design-win momentum in the market—hyperscalers are 'pulling' the company in, per Qualcomm VP. Unlike Nvidia's general-purpose GPU stack, Qualcomm is betting power efficiency (not breadth) wins the inference game, competing directly with Broadcom/OpenAI's Jalapeño, Amazon's custom Trainium, and Microsoft's Maia. The $15B target by 2029 implies aggressive capture: monitor Qualcomm CPU C1000 and custom ASIC announcements in 2027–2028 to assess real traction versus forecast.
Sources
- Primary source
- Qualcomm rolls out AI data center CPU, signs Meta as major customer
“Qualcomm Inc., trying to expand its dominance from mobile phone chips, forecast sales of more than $15 billion a year by fiscal 2029 in the booming market for AI components in data centers.”
- Qualcomm Forecasts Billions in New Data Center Revenue
“Qualcomm has a four-part strategy through 2028 with connectivity, custom silicon, new CPUs, and AI accelerators for inference optimization.”
- Qualcomm forecasts $15 billion data center chip sales by 2029
“Qualcomm also said it expects $40 billion in revenue from chips outside its smartphone stronghold by 2029, up from previous estimates of $22 billion.”