Qualcomm Acquiring RISC-V Chip Start-Up Tenstorrent?

[Daniel Nenni]

In the last few days, rumours have been circulating that Qualcomm is in talks to buy Tenstorrent for around $8-$10 billion. If true, it would be Qualcomm’s third major chip acquisition in the space of just a few months.

Building a Data Center Stack, Piece by Piece​

It has only been a few months since Qualcomm bought Ventana Microsystems, a developer of RISC-V-based CPU IP. A few months before that, Qualcomm acquired data center interconnect start-up Alphawave Semiconductors for $2.4 billion. Acquiring Tenstorrent adds two things to Qualcomm’s data center arsenal that it does not already have: Tenstrorrent’s BUDA software stack that allows AI models to run on non-NVIDIA chips and Tensix cores developed by Tenstortrent specifically for AI training:

· The BUDA Software Stack​

BUDA allows standard AI models written in PyTorch or TensorFlow to compile and run on non-NVIDIA chips without any code rewrite. This is important because one of NVIDIA’s biggest “moat” is its proprietary CUDA software stack, which is universally used by developers. BUDA gives cloud customers a path to switch away from NVIDIA hardware and software but without breaking their existing software ecosystems.

· Tensix Cores for AI Training​

Qualcomm’s Hexagon AI chips used in the AI100 and AI250 rackscale systems are essentially scaled-up mobile chips. They are designed for inference, not for training Large Language Models (LLMs) or for handling the type of continuous, high-throughput workloads used in multi-rack data centres. From the outset, Tenstorrent’s proprietary Tensix cores are designed for high-performance AI training. Unlike standard mobile chips, they can handle both high-end AI training and dense, continuous workloads.

Source: Tenstorrent

By acquiring Tenstorrent, Qualcomm gains a ready-made, enterprise-grade AI accelerator architecture that can be scaled within its data centre server racks and used as part of its newly announced Dragonfly data center brand. This will allow Qualcomm to immediately compete more effectively in the high-margin server rack space, rather than wasting 5 years building a comparable processor from scratch.

DEAL AT A GLANCE​

• Reported valuation: $8–10 billion
• Follows Ventana Microsystems (RISC-V CPUs) and Alphawave ($2.4B, interconnects)
• BUDA stack gives cloud customers a path from NVIDIA hardware without rewriting code
• Tensix cores fill the AI training gap in Qualcomm’s Hexagon-based portfolio
• Fits under Qualcomm’s newly announced Dragonfly data center brand

Analyst Viewpoint​

Qualcomm’s multi-billion-dollar acquisition of Tenstorrent is a calculated play to tap into the fast-growing AI data centre market and further diversify its business model away from its mainstay smartphone market. Acquiring Tenstorrent would enable it to overcome two of its biggest challenges: Nvidia’s near-monopoly of the AI training market and its reliance on ARM by leveraging the RISC-V IP acquired from Tenstorrent and Ventana.

However, there are risks. Qualcomm’s engineers will need to manage two different types of AI accelerators (Hexagon NPU/Tenstorrent Tensix cores) and multiple CPU architectures (ARM-based Oryon, Ventana and Tenstorrent RISC-V IP) simultaneously. Combining these into a single, cohesive product line under the new Dragonfly brand could lead to chaotic roadmaps and delayed chip rollouts.

If the deal closes, it would be one of the largest semiconductor acquisitions of 2026 to date and a clear sign that Qualcomm is serious about data centers. However, it remains to be seen whether it can integrate multiple chip architectures into a product line that competes at the top end with NVIDIA, AMD as well as the hyperscalers’ own custom silicon solutions.

Qualcomm Acquiring RISC-V Chip Start-Up Tenstorrent?

In the last few days, rumours have been circulating that Qualcomm is in talks to buy Tenstorrent for around $8-$10 billion. If true, it would be Qualcomm’s third major chip acquisition in the space…

counterpointresearch.com

 

[Jozo035]

Aiming for IPO​

Keller declined to comment on reported takeover bids from companies including Intel and Qualcomm, confirming only that he has indeed met with the CEOs of both companies, as well as all the major hyperscalers, in order to pitch them Tenstorrent’s hardware IP.

“I’m hoping to get a big deal out of one of those guys, because our RISC-V CPU IP is great,” he said. “One of the hyperscalers is also looking at our AI IP for a small chip.”

While hyperscalers have developed their own big chips for AI, smaller AI chips like those used in edge devices cannot just use a cut-down version of the same IP, Keller said. Tenstorrent’s AI IP is designed to be scalable, and it has been fully productized (it comes with everything needed to scale from, say, one to 1,000 cores, Keller said).

The two big exits for Tenstorrent’s startup competitors in the last six months have been an (effective) acquisition and an IPO. Tenstorrent is aiming to IPO, Keller confirmed, and is building out its supply chain and international presence with that in mind.

“Right now our investors are very hot on IPO,” he said.

Does Tenstorrent’s potential as a decode accelerator necessarily make it an attractive acquisition target for a GPU company? Keller said some kind of strategic deal or joint go-to-market is more likely.

Both sovereign infrastructure and the big frontier labs want to control their own destiny when it comes to hardware and software, he said. “Lots of things could happen,” he added.

Following TT-Deploy, Tenstorrent has received orders for its hardware, Keller said, with the biggest purchase order being for a 96-Galaxy cluster to be shipped outside the U.S. (96 Galaxies is 3,072 Blackhole chips). Tenstorrent’s biggest customer to date remains AI& in Japan, whose CEO is former Tenstorrent executive David Bennett.

“Some of what happened is a bunch of people had $100-million orders with Nvidia, but Nvidia won’t ship for a year, so they’ve taken a $20 million Tenstorrent machine because it’s a lot cheaper,” Keller said.

Tenstorrent is in the process of building 1,000 Galaxy servers, at least half of which have already been sold, he said.

“Our stuff is working pretty good, we have 10 customers with Galaxies on site, we’re past the proof-of-concept stage,” Keller said. “We’re starting to get follow-on orders… I want to get 10 happy customers, and then 20, and then 30.”

SANTA CLARA, Calif. — When EE Times visited Tenstorrent CEO Jim Keller’s office a year ago, the whiteboard outside his office door read: “We’re going to WIN!” On returning a year later, it reads: “Holy Shit, That’s Fast!”

https://www.eetimes.com/jim-keller-on-tenstorrents-blackhole-scaling-and-ipo-ambitions/

 

[swka]

Great for the many rounds of investors of Tenstorrent to get OK return after years of pivoting; QualComm on the other hand must has seen some technology, beyond talents, to pay that price

Qualcomm Acquiring RISC-V Chip Start-Up Tenstorrent?

In the last few days, rumours have been circulating that Qualcomm is in talks to buy Tenstorrent for around $8-$10 billion. If true, it would be Qualcomm’s third major chip acquisition in the space…

counterpointresearch.com

 

[blueone]

I'm not impressed with Tenstorrent. Nuvia I understood, this one, not so much. AheadComputing looks like a potentially better RISC-V alternative as a basis for future datacenter development.

 

[Xebec]

I'm not impressed with Tenstorrent. Nuvia I understood, this one, not so much. AheadComputing looks like a potentially better RISC-V alternative as a basis for future datacenter development.

Even if the products are weak - the IP and Engineering talent could be valuable? Tenstorrent I believe has made a real effort to address CUDA via some open sourced software efforts for AI, too.

(or this may be more of a defensive move - acquire them before someone else does).

 

[blueone]

Even if the products are weak - the IP and Engineering talent could be valuable?

Maybe. I don't know any of the people (I'm aware of).

Tenstorrent I believe has made a real effort to address CUDA via some open sourced software efforts for AI, too.

I thought that too, but their website doesn't mention CUDA. Or perhaps I just missed it.

(or this may be more of a defensive move - acquire them before someone else does).

I think this was really a team acquisition, but I suspect it has a lot to do with Keller's celebrity value.