Vanellus

Market position

Positioned against legacy thermal simulation tools and CPU-based or hybrid CPU/GPU solvers, with the open-source CFD package OpenFOAM used as the company's stated performance benchmark. The company describes a pilot-led rollout aimed at beta users rather than general availability, indicating an early commercial stage.

The company frames its differentiator as a solver built around GPUs from the ground up rather than a CPU solver with added GPU support, claiming resulting throughput gains over CPU and hybrid CPU/GPU physics solvers, alongside automated meshing that removes the need for a specialist meshing workflow and cloud delivery that avoids on-premise HPC procurement or contention. It also cites rigour through benchmarking against NASA turbulence cases.

Technology

A finite-volume method (FVM) CFD and thermal solver written to run on high-end GPUs, described as designed for massively parallel numerical work with GPU-oriented choices in meshing, memory layout and numerical solving algorithms; the company reports benchmarking against NASA turbulence cases and a roughly 25x speedup versus OpenFOAM on a high-end CPU node for a representative conjugate heat transfer problem. Surrounding components include automated mesh generation with live remeshing on geometry change, unified PCB and CAD geometry import, a range of thermal and fluid boundary conditions, cloud GPU execution, and a browser-accessible user interface. The company's public GitHub organisation contains Rust code, including idf-parser (a library for parsing Intermediate Data Format files used in PCB design) and forks of Bevy game-engine crates covering file dialogs, pan-orbit camera, mesh outlining, icon sets and an HTTP client modified to support CBOR, indicating a Rust and Bevy-based client application.

Go-to-market

Direct, pilot-led enterprise motion: the website invites prospective users to request a demo or start a pilot conversation, with the company stating it is geared toward beta users on real design problems where simulation can improve design velocity and time to market. Founder contact is provided directly (laurence@vanellus.tech).

Thermal engineers, product designers and engineering team leads working on dense electronics and tight form factors. Named application areas are server blades (air and liquid cooling, heat pipes, IC thermal behaviour, dense PCB layouts, heat sinks, fans, enclosure layouts), consumer electronics (vapour chambers, battery performance, avoiding overspecified cooling) and data centers (facility PUE, hall layouts, higher cooling fluid operating temperatures). The company also cites relevance to automotive electronics and advanced manufacturing teams, and highlights AI server and data center cooling, automotive and constrained electronics, and cloud-scale parallel simulation campaigns.

Geography

United Kingdom. Registered office at Unit 6, The Courtyard, Sturton Street, Cambridge CB1 2SN, England. The Entrepreneurs First portfolio listing gives locations as England, UK; Oxford, UK; Cambridge, UK; and Streetly End, Cambridge CB21 4RP, UK.

History

Vanellus Technologies Ltd was incorporated in the UK on 15 July 2022 as a private limited company (number 14236573) with the SIC code 62012, business and domestic software development, and a registered office in Cambridge. The Entrepreneurs First portfolio listing also records a founding year of 2022. Its public GitHub organisation released idf-parser, a Rust IDF file parser for PCB design data, with activity recorded from 2025, and its earlier positioning on the EF portfolio page emphasised machine learning-accelerated fluid simulation, while the current website emphasises a GPU-native finite-volume solver for electronics thermal simulation. Companies House records show accounts made up to 31 July 2025 and a confirmation statement dated 14 July 2026, with the company status listed as active.

Risks & controversies

Sources disclose no funding, customers or revenue, and the product is described as at a pilot/beta stage, so commercial validation is unverified. The headline ~25x speedup versus OpenFOAM is a company-stated figure for a single representative conjugate heat transfer case and is not independently verified in the sources. Company descriptions are inconsistent across sources: the Entrepreneurs First portfolio page describes machine learning-accelerated fluid simulation with "a thousand speedups" and lists a 2022 founding year, whereas the company website describes a finite-volume GPU solver for electronics thermal simulation. The team is small (1-10 employees), creating key-person concentration risk.