Data-center AI accelerators (AI200 and AI250)
Rack-scale inference systems announced in October 2025; AI200 deployments planned for 2026 with HUMAIN as first named customer.
What it is
Rack-scale AI inference accelerators built on Qualcomm NPU technology, with large LPDDR memory per card. AI200 is planned for 2026 and AI250 for 2027, with an annual product cadence after that.
Why it matters to the stock. This is the first attempt to sell into the data-center accelerator market against Nvidia and AMD. Management expects non-handset growth to accelerate once data-center revenue starts.
News
Amazon gets warrant for up to 25M Qualcomm shares tied to $60B of purchases
Qualcomm filed an 8-K on September 8, 2026 describing a multi-generation collaboration to supply custom AI inference silicon and related hardware for AWS data centers.
Qualcomm closes Modular acquisition, paying mainly in stock
Qualcomm completed its acquisition of Modular, the AI software company behind the Mojo language, with a closing date of July 28, 2026 stated in its quarterly report.
Qualcomm sets fiscal 2029 targets of $40B non-handset and over $15B data-center revenue
At its 2026 Investor Day, Qualcomm set fiscal 2029 targets of $40 billion in non-handset revenue and more than $15 billion in data-center revenue.
Qualcomm agrees to acquire AI software company Modular
Qualcomm agreed to acquire Modular Inc., an AI software infrastructure company whose platform runs AI models across different chip types without rewriting code for each one.
Qualcomm posts $10.6B fiscal Q2 revenue and guides Q3 to $9.2B to $10.0B
Qualcomm reported fiscal second-quarter revenue of $10.6 billion, down 3% from a year earlier, and non-GAAP EPS of $2.65 versus $2.85.