Jim Keller Says Cerebras IPO Was Helpful As Tenstorrent Set To “Beat Them on Everything”, Confirms Meeting With Intel & Qualcomm CEOs “Hoping To Get A Big Deal”

Jun 27, 2026 at 05:45pm EDT
Samsung To Build Next-Gen Tenstorrent AI Chiplet Leveraging RISC-V Architecture 1

Jim Keller isn't bothered by Cerebras's recent IPO and says that he welcomes it, but Tenstorrent will still beat them on everything.

Tenstorrent CEO, Jim Keller, Signals Deal With Intel or Qualcomm While Promising To Beat Cerebras "on everything"

Tenstorrent recently introduced its latest BlackHole Galaxy server, a system with which it can disrupt the entire AI segment, with performance levels that crush the competition. We covered the announcement last month when the company demoed its Blackhole server undercutting a NVIDIA GB300 with up to five times better TCO.

Related Story Intel Nova Lake Dual-Tile CPUs Reportedly Feature Up To 474W PL2 Power Limit

Keller Accepts The Challenge To Beat NVIDIA, Cerebras & Others At AI TCO

During a talk with EETimes, Jim Keller offered a statement on Cerebras recent IPO and their recent performance numbers shared for the Kimi K2.6 (1T) model. Cerebras WSE-CS3 (Wafer Scale Engine) hardware is said to offer nearly 1000 tokens per second on this model, but Jim states that they can not only beat this performance easily with large-scale deployments of BlackHole servers, but they can do that at a fraction of the total system cost.

He went on to say that he is unfazed by the recent Cerebras IPO and considers it "helpful" as they are going to beat them on everything.

“The Cerebras [IPO and subsequent valuation] was helpful, especially as we’re going to beat them on everything,” he said. “Challenge accepted!”

Jim Keller - Tenstorrent CEO (via EETimes)

Talking about how Tenstorrent tackles KV Cache, which competitors such as NVIDIA have built dedicated solutions for, such as its Groq-LPX chips, Keller points out that KV Cache is stored in the DRAM of the same chip as the decode, and it's just as simple as that.

Tenstorrent is confident in its approach, and they are "really good at that". The KV Cache can be scaled using Tenstorrent's Tensor chips, which can store it locally in the SRAM, but if those aren't enough, the data can be streamed to the DRAM, which does incur a performance penalty but is still better than NVIDIA and Cerebras's solutions, which lack DRAM.

“I am often asked how we handle the KV cache,” he said. “It’s in the DRAM on the same chips as the decode, we don’t even think about it. We’re really good at that.”

Jim Keller - Tenstorrent CEO (via EETimes)

Jim also shed light on the recent rumors floating about of Tenstorrent being bought out by Qualcomm and Intel. While he didn't state anything on the first part, he did confirm that he met with CEOs of both camps and is hoping to get a big deal from one of the companies, mainly its RISC-V CPU IP. Tenstorrent has also been approached by a hyperscaler on its AI IP for a small chip, Jim Keller told EETimes.

“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.”

Jim Keller - Tenstorrent CEO (via EETimes)

IPO & Strong Orders For Galaxy Blackhole Servers

Currently, Tenstorrent is aiming for an IPO, which Jim commented on, stating that their investors are "very hot" on the proposition and are working towards establishing a wider presence in the global market and a supply chain. The company has already seen a surge in orders for its server and AI-specific hardware, such as Galaxy, such as an order for a 96-Galaxy pod comprising 3072 Blackhole chips by an offshore customer.

The rise in orders has to do with competitors' inability to meet the demands of all customers. A $100 million NVIDIA order can cost roughly $20 million when going with Tenstorrent due to its very disruptive price proposition, and the company can deliver them despite facing higher demands itself. Tenstorrent is currently developing 1000 Galaxy servers, of which half are already sold.

About the author: A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as Wccftech's Senior Editor for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking.

Follow Wccftech on Google to get more of our news coverage in your feeds.