UK BUSINESS AI UNIVERSE: Places: East of England: Labs

Legal name

Cusp AI Limited (as shown on its website)

Founded

2024

Founders

Dr Chad Edwards — CEO and Co-Founder; Prof. Max Welling — CTO and Co-Founder (UKTN, June 2024: "chief AI officer")

Chief executive

Chad Edwards — from 2024

Chief scientist or research lead

Prof. Aron Walsh — Chief Scientific Officer

Other leaders

Debbie Toms — Chief of Staff; Dr Markus Hoffmann — Chief Strategy Officer; Dr Felix Hanke — VP Scientific Engineering; Alessandro De Maria — VP Platform Engineering; Will Kay — VP Product and Data; Jonathan Fildes — VP Creative and Communications; Dr Monica Hansen — VP Strategy. Advisers: Geoffrey Hinton, Yann LeCun, Verity Harding, Kristin Persson, Lord John Browne, John Giannandrea, Martin Van Den Brink, Abhi Talwalkar

Hq

Cambridge (England: East of England) — 20 Station Road, CB1 2JD; other_offices: London, Amsterdam, Berlin, Tokyo, Singapore, United States

What it does

Builds an AI system for discovering new materials — for semiconductors, energy and climate uses — and runs it with an "AI Materials Foundry", a network of over 45 industry, laboratory and data partners that make and test the candidate materials.

Research focus

AI for materials discovery; five stages stated as Define, Generate, Simulate, Synthesise, Scale

Models and products

AI Materials Foundry, built around the MIRA platform — launched 20 Jul 2026

Customers or partners named

NVIDIA (compute infrastructure) and Meta (FAIR, Universal Model for Atoms) — core partners — 20 Jul 2026 — https://www.uktech.news/ai/cuspai-raises-334m-and-launches-ai-materials-foundry-20260720; Hitachi High-Tech, Samsung, Hyundai Motor Group, Merck — founding members — same url; industry partners listed on site incl. 3M, AMD, Applied Materials, Fujifilm, Goodyear, Henkel, Johnson Matthey, Lam Research, Oxford PV, SoftBank, Tokyo Electron, Umicore — read 30 Oct 2026 — https://cusp.ai/; lab partners incl. A*STAR, imec, University of Cambridge, DTU — same url

Compute

NVIDIA named as compute infrastructure partner — Jul 2026

Funding total as published

over 650,000,000 USD — read 30 Oct 2026 — https://cusp.ai/

Valuation reported

2,600,000,000 USD (£1.9bn) — 20 Jul 2026 — (trade) UKTN (single_source)

Rounds found

2024-06-18 — first round (stage not stated) — 30,000,000 USD (£24m) — Hoxton Ventures — Basis Set Ventures, Lightspeed, LocalGlobe, Northzone, Touring Capital, Giant Ventures, FJ Labs, Tiferes Ventures, Zero Prime Ventures; 2026-07-20 — Series B — 450,000,000 USD (£334m) — Kleiner Perkins, NEA — Bezos Expeditions, Glade Brook Capital Partners, Lux Capital, AMD Ventures, Tru Arrow Partners, StepStone, Sovereign AI Venture Fund, Invest-NL, John Doerr. (Sep 2025 $100m round held as DEAL-LAB-023; not re-checked.)

Uk government links

Britain's Sovereign AI Venture Fund in the Series B (Jul 2026)

Status now

Raised a $450m Series B at $2.6bn in July 2026 and launched the AI Materials Foundry with over 45 partners.

Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: single_source (rounds trade-only; total from company site).

Sources

1. cusp.ai/
2. cusp.ai/

Legal name

null (footer "© 2026 Hiverge. All rights reserved.")

Founders

Alhussein Fawzi — Co-founder and CEO; Bernardino Romera-Paredes — Co-founder and CTO; Hamza Fawzi — Co-founder and Chief Scientist (Professor of Mathematics, University of Cambridge)

Chief executive

Alhussein Fawzi — from null

Chief scientist or research lead

Hamza Fawzi — Co-founder and Chief Scientist

Hq

20 Station Road, Cambridge CB1 2JD (England: East of England); other_offices: null

What it does

A Cambridge company whose platform, the Hive, runs thousands of AI research agents to find better algorithms for a customer's problem and returns them as readable code. Founders previously worked on DeepMind's AlphaTensor and FunSearch. Uses so far include AI training speed-ups, Airbus logistics planning and quantum algorithms.

Research focus

automated algorithm discovery; AI research agents; optimisation; quantum algorithms for combinatorial optimisation

Models and products

The Hive — algorithm-discovery platform — early access announced 17 September 2025; Hive integration with NVIDIA, Quantinuum and AWS — 16 March 2026

Customers or partners named

Airbus — Beluga long-horizon planning case study — 25 November 2025 — https://hiverge.ai/news; Quantinuum, NVIDIA, AWS — quantum algorithm work using CUDA-Q, NVIDIA GPUs on AWS — 16 March 2026 — https://hiverge.ai/blog/quantinuum-nvidia-collaboration

Papers and awards

first place in the Airbus Beluga scalability challenge (company blog) — September 2025; CIFAR-10 speedrun record, 1.99 seconds on one NVIDIA A100 (previous record 2.59 seconds) — 16 October 2025

Compute

says the Hive runs "on your own infrastructure or in the cloud, with your choice of AI models"; record run used "a single NVIDIA A100 GPU" — October 2025

Funding total as published

5,000,000 USD (seed) — as of 17 September 2025 — https://hiverge.ai/blog/introducing-hiverge-our-mission-and-early-results; NVIDIA planned investment October 2026, amount not disclosed

Rounds found

2025-09-17 — seed — 5,000,000 USD — Flying Fish Ventures — Ahren Innovation Capital, 10x Founders, Alpha Intelligence Capital, 1st Kind, StemAI, Vela Partners, Cur8 Capital, Jeff Dean (NEW: DEAL-LAB-9004); 2026-10-13 — strategic investment (planned) — undisclosed — NVIDIA — (already DEAL-LAB-028 / DEAL-LAB-2002)

Status now

Active Cambridge company; NVIDIA announced a planned minority investment on 13 October 2026.

Labs Round 7 deep profile (30 Oct 2026), as the researchers wrote it; confidence: verified.

Sources

1. hiverge.ai/
2. hiverge.ai/blog/introducing-hiverge-our-mission-
3. techfundingnews.com/nvidia-pledges-250m-of-compu
4. hiverge.ai/blog/accelerating-ai-training-made-si
5. hiverge.ai/blog/quantinuum-nvidia-collaboration
6. blogs.nvidia.com/blog/uk-ecosystem-deployment/

Founded

2025 (per The Jewish Chronicle; Cambridge Enterprise article gives both 2023 and 2025 — unresolved)

Founders

Jonathan Cohen-Gold — co-founder and Chief Executive; Lee Dunham — co-founder and Chief Business Officer

Chief executive

Jonathan Cohen-Gold

Hq

London (England: London) per The Jewish Chronicle (single source); other_offices: not stated

What it does

Houdini Bio makes software that redesigns the DNA in gene and cell therapies so the body's HUSH protein complex does not switch the therapy off. Machine learning predicts which DNA designs will stay active before a therapy reaches patients. It is a spin-out from the University of Cambridge Department of Medicine.

Research focus

Machine-learning DNA design to avoid HUSH-complex silencing of therapeutic genes, building on Professor Paul Lehner's research at Cambridge.

Models and products

AI DNA design engine — predicts durable therapeutic DNA sequences — not stated

Funding total as published

c. £1.5m (pre-seed plus non-dilutive grants) — as of 26 Jun 2026 — https://www.enterprise.cam.ac.uk/?p=38610

Rounds found

Jun 2026 — pre-seed (oversubscribed) plus grants — c. £1.5m — SCVC — Deep Science Ventures, Cambridge Enterprise Ventures (already DEAL-LAB-4025)

Status now

Trading; raised about £1.5m in June 2026 to develop its DNA design platform.

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: verified (funding: Cambridge Enterprise, an investor's parent, plus The Jewish Chronicle); founding year and HQ one source.

Sources

1. enterprise.cam.ac.uk/?p=38610
2. thejc.com/news/uk/houdini-british-jewish-start-a

Owner or parent

Microsoft Research

Founded or opened

Not stated (not stated on the lab page)

Lab director or lead

Aditya Nori — Vice President and Lab Director — date not stated

Other leaders

Austin Donnelly — Senior Principal Research Engineer Lead; Hitesh Ballani — Partner Research Manager; Katja Hofmann — Partner Research Manager; Siân Lindley — Senior Principal Research Manager; Ted Meeds — Senior Principal Research Manager; Rachel Howard — Director Business Management & Ops

Site and address as published

21 Station Road, Cambridge, CB1 2FB; and 198 Cambridge Science Park, Milton Road, Cambridge CB4 0AB

What it does

Microsoft's research lab in Cambridge. It designs hardware for Microsoft's cloud and AI data centres, including optical computing, storage in glass and datacentre networking and robotics. It also works on machine learning for science, industry and society, and on people-centred AI such as generative models of gameplay, tools that show where digital content comes from, and AI that helps people think.

Research groups or focus

Future AI Infrastructure; Machine Intelligence; People-Centric AI; Research Engineering

Notable outputs

WHAM (World and Human Action Models) — generative models of game worlds and player actions from the Muse programme, built with Xbox Game Studios' Ninja Theory; WHAM-RT is playable in Copilot Labs — 2025; the team moved into the Microsoft.AI product group (editor's note, 23 Sep 2026); AOC — analog optical computer for AI and optimisation workloads — date not stated; Project Silica — cloud storage written into quartz glass with femtosecond lasers — date not stated; Project Provenance — tools showing where digital content comes from — date not stated; Tools for Thought — date not stated; Teachable AI Experiences (Tai X) — date not stated; Project Volano — date not stated; CRAFT (datacentre robotics) — date not stated; Project Egypt (greenhouse-gas data) — date not stated; Memora — memory representation for AI agents — 29 Jun 2026; MicroLED datacentre networking — 17 Mar 2026

Selected papers

"WHAM: World and Human Action Models" (title as listed by Microsoft) — Nature — 2025 — https://www.nature.com/articles/s41586-025-08600-3

Partners named

Xbox Game Studios' Ninja Theory — WHAM — 2025; Ottobock — prosthesis image dataset for AI — Jul 2026

Uk government links

Not stated on pages opened

Status now

Active: the lab's news page lists items up to 30 Jul 2026.

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: one source (the lab's own pages).

Sources

1. microsoft.com/en-us/research/lab/microsoft-resea
2. microsoft.com/en-us/research/people/adityan/
3. microsoft.com/en-us/research/lab/microsoft-resea
4. microsoft.com/en-us/research/project/wham/
5. microsoft.com/en-us/research/lab/microsoft-resea
6. blogs.microsoft.com/on-the-issues/2025/09/16/mic

Owner or parent

Nokia (Nokia Bell Labs)

Lab director or lead

Bo Olofsson — "Software Innovations for Industry Verticals & Site Leader, Nokia Bell Labs Cambridge" — date not stated

Other leaders

Fahim Kawsar — Founding Director of Pervasive Systems research, Cambridge, UK; Michael Eggleston — Data & Devices Group Leader (lab listed at New Providence, New Jersey and Cambridge)

Site and address as published

Broers Building, 21 J.J. Thomson Avenue, Cambridge, CB3 0FA (West Cambridge campus)

What it does

Nokia Bell Labs' site on the University of Cambridge's West Cambridge campus. It works on multimodal AI and machine learning for connected devices and wearables, on social computing and on materials science. It is one of two sites of Bell Labs' Data and Devices lab, which builds sensors and AI for health and industrial monitoring.

Research groups or focus

"multimodal artificial intelligence/ machine learning" for IoT and wearables; social computing; materials science; Pervasive Systems research; Data and Devices Lab (AI Research Lab); founding member of the Research Centre for Mobile, Wearable Systems and Augmented Intelligence

Notable outputs (Data and Devices lab, two sites; the site of each project is not stated)

The Patchkeeper — wearable monitor with ECG, PPG, IMU and electronic stethoscope; The Gecko — compact solar-powered environment sensor; OCT-based biochemical sensing; battery-free backscatter RFID asset tracking — dates not stated

Selected papers (Data and Devices lab page, which covers both sites)

"Quantified Canine: Inferring Dog Personality From Wearables" — CHI '23 — 2023 — https://dl.acm.org/doi/10.1145/3544548.3581088 ; "Heart Rate Extraction from Abdominal Audio Signals" — ICASSP 2023 — 2023 — https://ieeexplore.ieee.org/abstract/document/10096600 ; "Stress Inference from Abdominal Sounds using Machine Learning" — EMBC 2022 — 2022 — https://ieeexplore.ieee.org/abstract/document/9871165 ; "The Gecko Sensor: An Ultra-Compact, Low-Cost, Solar-Powered Environment Monitoring Device" — IEEE Sensors 2022 — 2022 — https://www.nokia.com/sites/default/files/2024-11/the_gecko_sensor_an_ultra-compact_low-cost_solar-powered_environment_monitoring_device.pdf ; "ComFeel" — Proc. ACM IMWUT vol. 4 no. 4 — 2020 — https://www.nokia.com/sites/default/files/2024-11/comfeel.pdf

Partners named

University of Cambridge — the research centre above; University of Oxford; UCL; Lancaster; Newcastle; King's College London — academic partners as listed on the site page

Status now

Active, per the Nokia location page read on 31 Oct 2026.

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: one source.

Sources

1. nokia.com/bell-labs/about/locations/cambridge-uk
2. nokia.com/people/fahim-kawsar/
3. nokia.com/bell-labs/research/air-lab/data-and-de

Founders

Alex Ward — co-founder and CEO; Caelan Anderson — co-founder and CTO

Chief executive

Alex Ward

Chief scientist or research lead

Caelan Anderson — CTO

Hq

Cambridge (England: East of England); other_offices: Barcelona, Spain (Outside the UK)

What it does

Tolemy Bio builds an AI "virtual cell" platform that helps teams understand, simulate and improve the living cells used in cell and gene therapies and in drug manufacture, for example by designing the growth media cells are cultured in.

Research focus

"The AI-native virtual cell platform for therapeutic development and manufacture"; cell biology, advanced therapeutics, bioprocessing.

Models and products

Orbit — AI-native environment for cell biology workflows — named 12 May 2026; AI media-formulation prediction — custom cell-culture media design — 5 Dec 2024

Customers or partners named

MFX and Aminoacids.com — joint R&D on AI-derived media for T-cell culture — 5 Dec 2024 — https://mfx.bio/news-press-releases/press-release-mfx-tolemy-aa-2024/

Rounds found

12 May 2026 — pre-seed — €1.4m — Norrsken Evolve — Big Sur Ventures, JME Ventures, Masia, an unnamed UK stealth fund (already DEAL-LAB-4005)

Status now

Trading; raised a €1.4m pre-seed round in May 2026 for early customer and partner deployments.

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: one source.

Sources

1. tolemy.bio/team
2. tolemy.bio/platform
3. tech.eu/2026/05/12/tolemy-bio-lands-eur14m-to-po
4. mfx.bio/news-press-releases/press-release-mfx-to

Owner or parent

Toshiba Europe Limited (part of it since April 2020)

Founded or opened

1991, as Toshiba Cambridge Research Centre; "Toshiba's first overseas corporate-level R&D laboratory"

Lab director or lead

Roberto Cipolla — has led the lab since July 2007 (current title not given)

Other leaders

Sir Michael Pepper — led the lab from 1991 to July 2007; Chao Zhang — multimodal foundation models, Vision & Learning Group; Rudra Poudel — lead scientist, World Models for Reinforcement Learning, Vision & Learning Group

Site and address as published

Cambridge (no address on pages opened)

What it does

Toshiba's research lab in Cambridge, founded in 1991. It works on quantum communications and devices, speech and language, and computer vision and learning. Since 2024 it has expanded "embodied AI": systems that adapt quickly to new places, keep learning, and take instructions from people in plain language. Toshiba is putting £15m into this over five years and plans a first industrial prototype in 2027.

Research groups or focus

Quantum Information Group (Quantum Communications Research; Quantum Devices Research); Embodied AI: Language & Interaction Group and Vision & Learning Group

Notable outputs

Whisma — speech-LLM for zero-shot spoken language understanding — 2024; ReCoRe — training framework for world models — CVPR 2024; dialogue-based localisation (pose estimation through natural language) — CVPR 2024; LUCES and a Toshiba CAD dataset (linked, no details) — dates not stated

Selected papers

"Whisma: A speech-LLM to perform zero-shot spoken language understanding" — IEEE SLT — 2024 — https://www.toshiba.eu/cambridge-research-laboratory/embodied-ai/ ; "Recurrent Reinforcement Learning with Memoroids" — NeurIPS — 2024 — same url ; "NPLMV-PS: Neural Point-Light Multi-View Photometric Stereo" — WACV — 2025 — same url ; "An LLM Feature-based Framework for Dialogue Constructiveness Assessment" — EMNLP — 2024 — same url ; "ToolReAGt: Tool Retrieval for LLM-based Complex Task Solution via RAG" — ACL 2025 KnowFM workshop — 2025 — same url ; "Conditional Multi-Stage Failure Recovery for Embodied Agents" — ACL 2025 REALM workshop — 2025 — same url

Funding or budget as published

£15m from Toshiba over five years for embodied AI — 17 Jun 2024 — https://www.toshiba.eu/cambridge-research-laboratory/news/toshibas-leading-cambridge-research-laboratory-crl-doubles-down-on-cutting-edge-embodied-ai-to-bring-ai-into-the-real-world/ (single source)

Partners named

University of Cambridge — Cavendish Laboratory and Engineering Department — ongoing

Status now

Active; the first industrial embodied-AI prototype is planned for 2027.

Labs Round 8 deep profile (31 Oct 2026), as the researchers wrote it; confidence: one source.

Sources

1. toshiba.eu/cambridge-research-laboratory/about-u
2. toshiba.eu/cambridge-research-laboratory/news/to
3. toshiba.eu/cambridge-research-laboratory/embodie