On 15 June 2026, Reuters relayed a report from The Information that Qualcomm was in talks to acquire Tenstorrent, the AI-chip startup led by veteran silicon architect Jim Keller, at a valuation of between $8 billion and $10 billion. Reuters said it could not independently verify the report, and both Qualcomm and Tenstorrent declined to comment. Two weeks later, Keller reportedly denied that Tenstorrent was in acquisition talks with Qualcomm, saying the company was focused on building its own business. That leaves the story in an unusual place: a reported multibillion-dollar deal that was never confirmed and has since been denied in public reporting. It is worth writing about anyway, both because of what the denial does and does not rule out, and because the rumour itself is a clear read on how the chip industry is behaving right now.

What was reported, and what Keller said

The original report came with heavy caveats even before the denial. The talks were described as ongoing, the price as potentially subject to change, and the negotiations as something that could still fall through; both Qualcomm and Tenstorrent declined to comment to Reuters, which said it could not independently verify the report. Qualcomm's shares fell about 1% after it surfaced. This was, in other words, an acquisition report rather than an announced deal from the outset.

Keller's denial two weeks later, as reported through financial outlets, was direct but narrower than it first appears. He reportedly said Tenstorrent is not in acquisition talks with Qualcomm and is concentrating on its own path — its intellectual-property licensing operations and high-end scalable AI work in Japan and other markets — while leaving the door open to partnerships and noting that interest in large deals across the sector has increased. The denial, as reported, rules out current Qualcomm acquisition talks; it does not by itself rule out future partnerships or future transactions. That distinction matters, because it is easy to read a denial of one deal at one moment as a broader statement of intent, which this was not.

Why the rumour made sense

The reason the report was plausible enough to move Qualcomm's stock is that the strategic logic is real. Qualcomm has been trying to expand beyond the smartphone chips and modems that define it, and data-centre AI is one of its clearest targets. The company has announced its AI200 and AI250 data-centre accelerator platforms, expected to be commercially available in 2026 and 2027 respectively, and has since broadened that roadmap — adding a Dragonfly AI300 part alongside data-centre CPUs, inference accelerators, connectivity and custom-silicon ambitions. Tenstorrent, founded in 2016, builds AI systems around a RISC-V foundation and positions its architecture as an efficient alternative for certain AI workloads. Keller's own track record — he worked on chips at Apple, led the design of AMD's Zen architecture, and oversaw Tesla's self-driving chip effort — makes the team as much of the prize as the products.

That fits a broader pattern worth naming: for almost everyone in the industry who is not Nvidia, the choice is to buy or build a credible AI-accelerator capability, and the buying has been brisk. A Qualcomm move on Tenstorrent would have been an aggressive version of that instinct — acquiring architecture, intellectual property and one of the industry's most respected chip designers in a single stroke.

Why to be sceptical

Set the denial aside for a moment and the deal still has problems. The most obvious is price. Tenstorrent had previously been reported at a much lower, multibillion-dollar valuation — around $3.2 billion, per Tom's Hardware, when it was raising last year — so a jump to an $8–10 billion valuation would be a very large re-rating in a short time. The second is redundancy. As analysts have pointed out, Qualcomm already possesses AI-acceleration and CPU intellectual property of its own, and its broadened 2026 data-centre roadmap only underlines that, which makes paying $8–10 billion for more of it harder to justify than it would be for a company starting from nothing. Add Keller's reported denial on top of those two, and the base case shifts toward this deal not happening, at least not on the reported terms.

Key Takeaways

  • The Information reported on 15 June 2026 (in an account relayed by Reuters) that Qualcomm was in talks to acquire Jim Keller's AI-chip startup Tenstorrent at a valuation of $8–10 billion; the report was hedged (talks ongoing, price could change, deal could fall through), both companies declined to comment, and Reuters could not independently verify it.

  • Two weeks later, Keller reportedly denied — through financial outlets covering a Tokyo media event — that Tenstorrent is in acquisition talks with Qualcomm, saying it is focused on its own business while leaving partnerships open. The denial, as reported, rules out current Qualcomm acquisition talks specifically, not all future deal activity.

  • The rumour was plausible because the strategic logic is real: Qualcomm is pushing beyond smartphones into data-centre AI, with its AI200 and AI250 accelerator platforms (expected in 2026 and 2027) and a broadened roadmap adding Dragonfly AI300, CPUs and custom silicon; Tenstorrent offers RISC-V-based AI IP plus a highly regarded chip designer.

  • The reasons for scepticism are also real: Tenstorrent was previously reported at a much lower valuation (around $3.2 billion) versus the rumoured $8–10 billion, and Qualcomm already has AI-acceleration and CPU IP, making the price difficult to justify. The honest status is a reported deal, reportedly denied and unconfirmed — a signal of AI-accelerator M&A appetite more than a transaction to bank on.