# No chip yet, no benchmarks, $30M from Benchmark: Tendrils reportedly eyes a $1 billion valuation next

A Cambridge startup wants to replace 1970s-era CPU design with thousands of parallel cores. A UK minister cheered; Hacker News engineers were skeptical.

By Patrick Karsh · 567 Business
Published: 2026-11-11T12:12:45.584785Z
Updated: 2026-11-11T12:12:57.485955Z
Canonical: https://567business.com/news/tendrils-compute-30-million-benchmark-cpu

Cite citation_url. Preserve source links, dates and qualifications. date_modified is an edit date, not proof that every claim was rechecked. Third-party images and quoted material retain their own rights. Resolve relative links against https://567business.com/news/tendrils-compute-30-million-benchmark-cpu.

Tendrils Compute, a small chip startup in Cambridge, England, has raised $30 million in a round led by Benchmark to "redo the CPU." It has not yet published any hardware benchmarks, [Tech.eu noted](https://tech.eu/2026/11/11/tendrils-compute-raises-30m-to-redo-the-cpu/). And before the round was even public, the company was already discussing a fast follow-up raise that could value it at more than $1 billion, according to people who spoke to [Newcomer](https://www.newcomer.co/p/scoop-benchmark-backs-early-stage) in October.

That gap between ambition and proof is why the raise split opinion. Investors and a UK minister cheered it as a bold bet on new hardware. On Hacker News, many engineers called it a familiar idea that has failed before.

## What Tendrils says it is building

Tendrils announced the round on November 10 in a post titled ["$30M Raised to Redo the CPU"](https://tendrils.co/blog/2026-11-10/30m-raised). Its pitch is that today's processors are held back by "decisions dating back to the 1970s," because each new chip must stay compatible with older, single-core designs.

Instead, Tendrils plans a processor with thousands of cores working in parallel. It is built on a model of computing called interaction nets, which French computer scientist Yves Lafont described in 1990, [according to investor Cloudberry VC](https://cloudberry.vc/news/why-we-invested-in-tendrils/). In that model, a program is a graph of small nodes. The machine computes by rewriting small pieces of the graph, and pieces that are ready can run at the same time. The company is also building [Vine](https://vine.dev/), an open-source programming language meant to target this kind of chip.

The company announced the raise on X, with Benchmark general partner Jack Altman joining its board.

https://x.com/TendrilsCompute/status/2120200074349457524

## Who put money in

Benchmark led the round. Visionaries Club, the UK Sovereign AI Fund, FR8, Antimatter, Cloudberry VC, Conviction, Embassy Ventures and SV Angel also took part, Tendrils said. The team includes founder Nils Cremer, who worked on interaction nets at ETH Zürich, and Hugo McNally, who came from open-source chip group lowRISC, Cloudberry wrote. Cremer's LinkedIn profile lists the company as founded in May 2025 with one to ten employees.

The UK government's stake drew the most heat. Kanishka Narayan, a Labour MP whose X profile names him the UK's minister for AI, said the government was investing through the Sovereign AI fund. He framed it as "turning British ingenuity into new industries, good jobs and growth."

https://x.com/KanishkaNarayan/status/2120215626463547893

## Engineers on Hacker News: "build it and they will come"

The [Hacker News thread](https://news.ycombinator.com/item?id=50447036) on the announcement drew more than 110 comments in its first 13 hours. The most common critique was history. Commenters listed past parallel chips that struggled to find users, from Inmos's Transputer to the Cell processor in the PlayStation 3, GreenArrays and Adapteva's Parallella.

One top comment called it a "build it and they will come" project. Another asked who would want "an exotic architecture on an old process node running a custom programming language half a decade from now." One user wrote that it "feels like a bad use of the UK Sovereign AI Fund."

Others pushed back. Some argued that AI coding tools could make it far cheaper to port software to a new chip, which was the wall that sank many earlier designs. We did not find an active Reddit discussion of the raise, so Hacker News is where the technical debate is happening.

Some well-known programmers were openly hopeful. Victor Taelin, creator of the HVM runtime, which is built on the same interaction net ideas, said he has long seen it as "the optimal way to build hardware in this physical universe."

https://x.com/VictorTaelin/status/2120215419810414844

## Why investors are paying up for chips

Tendrils is riding a wave of chip money. Newcomer reported that Benchmark won a "hot competition" to lead the round, and listed recent chip deals that show how high prices have climbed.

| Company | Deal | Reported value |
| --- | --- | ---: |
| Etched | $700 million raise, August | $21 billion |
| Fractile | Talks on a new round | $6.5 billion |
| OLIX (UK) | $312 million raise, August | $3.3 billion |
| Tendrils | $30 million raise, November | Not disclosed |

Source: [Newcomer](https://www.newcomer.co/p/scoop-benchmark-backs-early-stage), reported October 1, 2026. Fractile's figure was for talks, not a closed round.

Benchmark also has a reason to like chips. Its partner Eric Vishria co-led Cerebras Systems' Series A in 2016, and Benchmark saw a 20.5 times return when Cerebras went public this spring, Newcomer reported. Cerebras listed in May at $185 a share. As of early October, it had traded as high as $386 but sat near its offering price.

## What the company has not shown

Tendrils has not said when a chip will be made, which process it will use, or how fast it will run compared with today's CPUs and GPUs. Its round size is also small for a new processor. One Hacker News commenter said $30 million is a plausible budget to make a first silicon prototype. Another called it "not a lot for a project of that size." The company did not disclose its valuation.

None of that means the idea is wrong. It means the proof has not arrived. Investors are paying for a team and a theory, with the hardware still to come.

For more on how venture money is chasing hardware even as public chip stocks swing, see our story on [Nominal's tripled valuation and Path Robotics' raise](/news/path-robotics-nominal-hardware-startups-raise). For the other side of the chip trade, read how [Nvidia lost almost $500 billion in two days](/news/nvidia-500-billion-jensen-huang-probe).