Qualcomm Pursues $8-10B Acquisition of Tenstorrent to Enter Data Center AI Chip Market
The AMW Read
Qualcomm entering data center AI chip market via acquisition updates the player map for AI infrastructure segment; adds credible third inference architecture contender alongside NVIDIA/AMD.
Qualcomm Pursues $8-10B Acquisition of Tenstorrent to Enter Data Center AI Chip Market
Qualcomm is in advanced acquisition negotiations with AI chip startup Tenstorrent at a valuation of $8-10 billion (approximately 12-15 trillion won), according to The Information. The deal would far exceed Tenstorrent's $2.6 billion valuation from its December 2024 funding round. Tenstorrent was founded in 2016 by legendary chip architect Jim Keller, who previously led processor development at Apple and Tesla's autonomous driving hardware. The startup differentiates itself from NVIDIA's GPU-dominated approach by focusing on specialized AI accelerators built on the open-source RISC-V architecture, targeting higher efficiency for specific inference workloads.
This acquisition represents a strategic pivot for Qualcomm, whose revenue remains heavily concentrated in mobile and PC processors. Qualcomm has been actively diversifying into data center AI silicon, having acquired UK-based interconnect specialist Alphawave Semi for $2.4 billion last year. The Tenstorrent acquisition would accelerate Qualcomm's ability to compete in the AI accelerator market currently dominated by NVIDIA, AMD, and emerging Chinese players. At Qualcomm's current ~$200 billion market cap, a $8-10 billion deal is substantial but digestible, signaling serious commitment to data center expansion.
The acquisition attempt exemplifies the capital-intensive acqui-licensing pattern where incumbent semiconductor firms acquire AI chip startups to gain architectural talent and product lines rather than building from scratch. Jim Keller's involvement adds rare talent-acquisition value — his track record at Apple, Tesla, and AMD makes Tenstorrent's RISC-V inference architecture one of the most credible non-GPU alternatives. If completed, this deal would update the competitive map of the AI silicon landscape (segment 04), potentially creating a third major inference chip architecture alongside NVIDIA's CUDA ecosystem and AMD's ROCm platform. The $8-10 billion price tag also tests the upper boundary of AI chip startup valuations outside the hyperscaler acquisition ecosystem.
