Qualcomm weighs a $8B-$10B Tenstorrent buy to accelerate RISC-V AI chips

Qualcomm is reportedly circling Tenstorrent in a deal that could land between $8 billion and $10 billion, according to The Information citing an anonymous source. The talks are described as ongoing, with no certainty that a deal will be reached, but the size and target are the kind of signals executives pay attention to. This is not “let's partner” territory. It is “let's own the stack and the roadmap” territory.

Tenstorrent is a Canadian AI chip startup, and its entire pitch is anchored to RISC-V, an open instruction set architecture that is permissively licensed. Qualcomm's interest in that framing lines up with what it has been publicly doing on RISC-V already, including datacenter ambitions and earlier moves. Cristiano Amon, Qualcomm's chief, has made bullish statements about AI opportunities, and a purchase of Tenstorrent would be a direct way to translate that AI enthusiasm into silicon and software muscle built around the open standard.

So why does this matter beyond Qualcomm and Tenstorrent's balance sheets? Because the datacenter AI chip ecosystem is increasingly about control and ecosystem pull. Whoever supplies a credible compute platform tends to shape the tools developers build, the compilers and stacks vendors support, and the expectations customers develop. RISC-V, in that context, is not just a technical detail. It is a bet on whether an open architecture can become a serious default in high-performance computing and AI accelerators.

Tenstorrent is already showing what that bet could look like. Its Galaxy Blackhole AI compute platform went on sale earlier this year. The platform packs 32 Blackhole accelerators, and each accelerator includes 768 RISC-V cores, all placed into a 6U enclosure that runs its own software stack. In other words, this is not a concept pitch. It is a packaged system with a defined approach to compute, presumably designed to be used as a coherent product rather than scattered components.

Qualcomm, meanwhile, is also tied to RISC-V for another, more tactical reason: its licensing fight with Arm. Arm previously wanted to nix Qualcomm's license to create its own Arm-based processor silicon, and Qualcomm has been in a court battle over that licensing issue. That context matters because it changes what “strategic interest” means. RISC-V becomes not only an ecosystem opportunity, but also a hedge against being boxed in by Arm's licensing terms. At the same time, Qualcomm's datacenter products use home-brew Hexagon neural processing units, but the company continues to rely on Arm processors in its Snapdragon range. That mix is important: a Tenstorrent acquisition could be a way to increase RISC-V depth without pretending Qualcomm can flip its entire architecture overnight.

This would also not be Qualcomm's first RISC-V related move. In December, Qualcomm picked up Ventana Micro Systems, another company designing RISC-V CPUs targeting datacenter and enterprise applications. Financial details were not disclosed, but estimates put that acquisition between $200 million and $600 million. Put those numbers next to a potential $8 billion to $10 billion Tenstorrent play, and you can see the underlying shift. Qualcomm could be moving from “buy pieces” to “buy a platform” if the talks progress.

There is also a governance and competitive layer here. Qualcomm and Tenstorrent were asked to comment, but no confirmation is in the reporting you have. Still, the mere fact that a deal window of up to $10 billion is being discussed suggests internal conviction. Large AI chip acquisitions can also be a way to win attention from customers who are tired of vendor roadmaps that change with every quarter. A larger commitment to Tenstorrent's RISC-V platform could make Qualcomm's datacenter story more coherent, especially if it supports developers who want one architecture across product lines.

Outside the RISC-V camp, Arm does not appear panicked. Arm has said it expects datacenter chips to soon be its main source of revenue. That posture implies Arm views the market as something it can still monetize even if Qualcomm spends more heavily on RISC-V and pushes harder to distance itself from Arm and its owner SoftBank. If Qualcomm does pull the trigger, the question for other executives is simple: will this be a one-time bet, or the start of a broader reallocation of capital toward open architectures in AI compute? For anyone making platform decisions in datacenter infrastructure, that difference is worth tracking.