Tendrils is betting on the CPU bottleneck faced by AI agents. If its parallel architecture takes off, the computing-power narrative may shift from GPUs to heterogeneous computing; on the crypto side, focus on DePIN and avoid chasing concepts without tokens. $BTC $ETH

Flash: Tendrils raises $30 million in seed funding to build a general-purpose processor capable of running thousands of cores in parallel
BlockBeats News, November 11: Cambridge semiconductor startup Tendrils has raised $30 million in seed funding, approximately €26 million. The round was led by Benchmark, with participation from Visionaries Club, the UK Sovereign AI Fund, FR8, Antimatter, Cloudberry VC, Conviction, Embassy Ventures, SV Angels, and several angel investors. Founded in 2025 by CEO Nils Cremer and co-founders T6 and Hugo McNally, the company is developing a general-purpose processor architecture that enables thousands of cores to work in parallel. The new architecture is based on the interaction-net computation model, representing programs as interconnected nodes whose independent operations can run simultaneously. It aims to move beyond traditional shared mutable memory and linear instruction sequences. The company is also developing Vine, an open-source programming language with Rust-like syntax to make it easier to adapt software to this parallel-first architecture. Tendrils says that while GPUs already handle large-scale numerical computing, AI agents still rely heavily on CPUs for irregular tasks such as executing code and retrieving documents. Jack Altman, general partner at Benchmark and co-founder of Lattice, will join the board.