Qualcomm targets $15bn data centre revenue

Qualcomm has launched the Dragonfly data centre portfolio, agreed to acquire Modular and named Meta as an early customer. The move positions Qualcomm around AI inference efficiency and a $15bn data centre revenue target by 2029.

• Dragonfly centres on C1000 CPU and AI300 accelerator targeting inference workloads

• Meta deployment gives early hyperscaler credibility before 2028 production ramp

The fact

Qualcomm has launched Dragonfly, a data centre portfolio built around the C1000 CPU and the AI300 accelerator for AI inference workloads. The company says AI300 delivers four to eight times the performance per watt of current GPU-based architectures. Meta will deploy C1000 CPUs in its next-generation server fleet. Qualcomm also agreed a $3.8bn all-stock acquisition of Modular and targets $15bn in data centre revenue by 2029.

The Assessment

Dragonfly is built around a clear market pressure: AI inference is becoming more expensive and power-hungry. Qualcomm's bet is that efficiency will rival raw compute as AI workloads scale. The Modular acquisition — at $3.8bn — shows it does not want to compete on silicon alone. Acquiring the compiler and software stack means Dragonfly can run efficiently across mixed hardware environments. Meta gives the plan early credibility, but Qualcomm needs more hyperscaler wins to convince the market.

What to Watch

Watch whether Qualcomm adds more hyperscaler customers before C1000 production begins in 2028, and whether AI300's power-efficiency claims are independently validated.