Artificial intelligence development is becoming increasingly expensive and technically demanding. Startups building advanced AI systems need access to specialist engineers, computing infrastructure, intellectual property and investment, while enterprises face their own challenges around security, reliability, compliance and deployment.
CommonAI is building a model designed to address some of these challenges through collaboration.
Launched in the UK in 2025 by Anthemis and Cambridge AI Venture Partners, CommonAI brings together AI engineering, compute infrastructure, shared intellectual property and investment support. Its focus has since expanded considerably, with programmes covering AI inference infrastructure and high assurance AI, alongside partnerships with organisations including the UK’s Advanced Research and Invention Agency (ARIA) and Barclays.
One of the most significant developments came in February 2026, when ARIA backed the creation of the Scaling Inference Lab, a £50 million initiative intended to test and accelerate new AI hardware and systems at scale.
This makes CommonAI an increasingly relevant name within the UK’s AI infrastructure and deep tech ecosystem.
What is Common AI?
CommonAI is a UK based collaborative engineering and computing platform designed to help AI startups, enterprises, researchers and other organisations develop and scale AI technologies.
The organisation’s current model brings together three major areas:
- Collaborative engineering programmes
- AI compute infrastructure
- Investment and capital partnerships
Its objective is to reduce some of the duplication and infrastructure barriers that can prevent smaller companies from competing in advanced AI development.
Rather than every startup having to independently develop the same underlying infrastructure, recruit large engineering teams and secure expensive compute resources, CommonAI aims to provide shared capabilities that members can build upon.
The company describes its approach as connecting AI creators, compute and capital. Its platform provides access to shared intellectual property, engineering resources and AI accelerators, while its broader ecosystem connects technology companies with investors and strategic partners.
Is CommonAI a Non-Profit?
There is an important distinction between the different parts of the CommonAI ecosystem.
CommonAI CIC is the non-profit membership organisation responsible for collaborative engineering programmes and the Digital Commons.
Alongside it sits CommonAI Holdings Ltd, the UK company behind the wider commercial ecosystem, including the platform’s compute and investment relationships.
CommonAI CIC describes itself as a deep tech AI non-profit membership organisation focused on collaborative engineering and the safe and responsible development of foundational AI technologies.
This is different from describing the entire CommonAI ecosystem as simply a non-profit AI research organisation.
The distinction matters because CommonAI’s current activities span engineering, infrastructure, membership, compute and investment partnerships.
Who Founded CommonAI?
CommonAI was launched through a collaboration between Anthemis and Cambridge AI Venture Partners (CAIVP).
The Cambridge AI Venture Partners team involved in the launch included:
- Sir Andy Hopper
- Professor Robert Mullins
- Dr Gavin Ferris
- Mike Halsall
Anthemis brought its experience in venture investment, while CAIVP contributed technical and engineering expertise from Cambridge’s deep tech ecosystem.
The broader CommonAI leadership and board currently includes figures associated with both organisations.
The company’s website lists Amy Nauiokas, Sir Andy Hopper, Mei Lim and Dr Gavin Ferris among its board directors, alongside other members of the wider team.
Where is CommonAI Based?
CommonAI has strong links to both London and Cambridge.
CommonAI Holdings Ltd lists its registered address at 19-20 Berners Street, London, W1T 3NW, while CommonAI CIC describes itself as being based in Cambridge and draws heavily on the city’s Silicon Fen technology ecosystem.
Cambridge is particularly important to the organisation’s identity because several of its founders and technical leaders have been involved in major UK computing and deep tech initiatives.
The organisation points to connections with projects and organisations including lowRISC, OpenTitan, the CHERI Alliance and Raspberry Pi.
What Does CommonAI Do?
CommonAI’s current offering can be divided into several interconnected areas.
Collaborative AI Engineering
CommonAI provides engineering programmes through which members can work together on specific technical challenges.
Members can gain access to shared intellectual property, engineering resources and programme infrastructure rather than having to build every capability independently.
The organisation says its membership model allows startups, enterprises, universities and public sector organisations to participate in collaborative programmes.
Shared AI Compute
Compute has become one of the largest costs associated with developing and deploying advanced AI systems.
CommonAI Compute is designed to simplify access to AI accelerators by providing multicloud, on-demand infrastructure.
The company says its compute service is designed to provide flexibility without locking users into a single cloud provider.
This is particularly relevant for startups that may not have the capital or infrastructure required to operate their own large-scale AI computing environment.
Shared Intellectual Property
Another central component is the Digital Commons.
The concept is to create a protected environment in which participating organisations can access shared technical resources and intellectual property developed through CommonAI programmes.
This creates a middle ground between completely closed corporate research and completely open public development.
Members can collaborate while operating within defined governance and IP frameworks.
CommonAI’s Scaling Inference Lab
The Scaling Inference Lab is currently one of the most important parts of CommonAI’s work.
The programme received major backing from ARIA, the UK’s Advanced Research and Invention Agency.
ARIA’s 2025 to 2026 annual report states that it committed £50 million to build the Scaling Inference Lab, a national testbed designed to help startups demonstrate and develop next-generation AI hardware.
CommonAI says the lab will create five or more rack-level AI systems, with each system acting as a structured pilot for emerging technologies.
The objective is to test new hardware and systems in realistic, full-system environments rather than evaluating individual components in isolation.
Why AI Inference Matters
AI inference refers to the process of running a trained AI model to generate an output.
As AI adoption expands, inference is becoming an increasingly important part of the computing infrastructure required to operate AI systems.
Training large models can require enormous amounts of compute, but inference can also become a major cost once those models are being used by large numbers of people and businesses.
CommonAI’s Scaling Inference Lab is therefore focused on a critical question for the next phase of AI development:
How can AI systems deliver greater performance while reducing the cost and infrastructure required to run them?
The programme is intended to provide startups with an environment where new hardware and system technologies can be tested at scale.
What is the High Assurance AI Programme?
CommonAI is also working on High Assurance AI, a programme focused on making AI systems safer and more dependable in industries where mistakes can have significant consequences.
These include:
- Financial services
- Healthcare
- Critical infrastructure
- Other highly regulated sectors
CommonAI says deployment in these areas can be restricted by concerns around compliance, security, auditability and unpredictable AI behaviour.
The programme is working on technical approaches including:
Verified Outputs
The programme aims to develop methods for establishing greater confidence in AI outputs, including approaches based on mathematical verification.
Secure Execution
AI-generated code can create security risks.
CommonAI’s High Assurance programme therefore includes work on sandboxed environments intended to isolate AI-generated code.
Multi-Model Validation
Using multiple models to cross-check outputs can provide another layer of validation.
Intelligent Filtering
The programme also includes work on preventing data leakage and policy violations.
These technologies are particularly relevant as businesses move from experimental AI deployments towards systems that interact with sensitive data and important operational processes.
CommonAI and ARIA
The relationship between CommonAI and ARIA represents one of the most significant developments since the platform launched.
ARIA is a UK government-backed research agency established to fund high-risk, high-reward scientific and technological research.
In February 2026, ARIA announced its support for the Scaling Inference Lab.
ARIA’s annual report subsequently confirmed the £50 million commitment for the national testbed.
The partnership places CommonAI within a much broader UK effort to strengthen domestic AI infrastructure.
It also reflects a shift in the AI market.
The conversation is no longer only about developing increasingly capable models. Increasing attention is now being placed on the hardware, inference systems and infrastructure needed to run those models efficiently.
CommonAI and Barclays
CommonAI has also expanded its corporate partnerships.
In June 2026, Barclays invested in CommonAI and joined its High Assurance programme. CommonAI said the partnership would focus on challenges associated with deploying AI in financial services, where compliance, correctness and security are particularly important.
The partnership is significant because financial services represent one of the industries where AI deployment needs to balance innovation with strict requirements around risk, data and regulation.
Barclays’ participation gives CommonAI access to a major financial institution while giving the bank a role in shaping technical approaches to high assurance AI.
CommonAI’s Growing Member Ecosystem
CommonAI’s ecosystem has continued to grow during 2026.
One example is Oriole Networks, which joined CommonAI in June 2026 through the Scaling Inference Lab.
Oriole is expected to use the programme to pilot an AI cluster based on its PRISM optical networking technology and validate its performance and integration at scale.
This illustrates how CommonAI’s model can bring infrastructure companies and AI developers together within the same technical environment.
Another notable member is Callosum, which announced a $100 million seed funding round in August 2026.
CommonAI says Callosum is using the Scaling Inference Lab to validate its technology for orchestrating heterogeneous AI compute across different models and chips.
These developments demonstrate that the CommonAI ecosystem is moving beyond its initial launch concept and into practical infrastructure testing and collaboration.
How CommonAI Supports AI Startups
AI startups face several barriers that are particularly difficult during the early stages of development.
Compute Costs
Training and running advanced AI systems can require significant amounts of computing power.
Access to flexible infrastructure can reduce the need for startups to make large upfront investments.
Engineering Resources
Specialist AI engineers are in high demand.
A shared engineering environment can give smaller companies access to technical capabilities that would otherwise be difficult to build internally.
Intellectual Property
Developing foundational technology from scratch can be expensive and time-consuming.
CommonAI’s Digital Commons model is intended to allow members to access shared IP while maintaining protected governance arrangements.
Access to Capital
CommonAI also aims to connect its technology ecosystem with investment.
The organisation works with asset managers and investment firms developing thematic funds aligned with its engineering programmes.
This creates a model where technology, infrastructure and capital can be connected rather than treated as separate parts of the startup journey.
CommonAI’s Investment Strategy
Investment is an important part of CommonAI’s wider ecosystem, but it is worth distinguishing the platform from a conventional venture capital fund.
At launch, Anthemis announced plans to raise a High Assurance AI fund focused on UK and European companies applying AI to complex and regulated industries.
The broader strategy was intended to connect portfolio companies with not only investment capital but also the technical resources available through CommonAI.
CommonAI’s current website describes partnerships with asset managers launching thematic investment funds aligned with its engineering programmes.
This creates a different model from a traditional accelerator or venture fund.
The idea is to connect:
Technology + Compute + Engineering + Capital
within the same ecosystem.
Is CommonAI an AI Startup?
CommonAI is better understood as an AI ecosystem and collaborative infrastructure platform rather than a conventional AI application startup.
It does not primarily compete by selling a consumer chatbot or a single AI software product.
Instead, its focus is on the infrastructure and technical foundations required by organisations developing AI technologies.
Its activities now span:
| Area | CommonAI’s role |
|---|---|
| AI engineering | Collaborative development programmes |
| Compute | Access to AI accelerators and infrastructure |
| AI infrastructure | Scaling inference and hardware testing |
| Shared IP | Digital Commons and protected collaboration |
| High assurance AI | Technical foundations for safer AI deployment |
| Investment | Connections with thematic investment funds |
| Membership | Startups, enterprises, academia and public sector |
This positioning makes CommonAI particularly relevant to the UK AI infrastructure and deep tech ecosystem.
Why CommonAI Matters to the UK AI Sector
The UK’s AI sector has developed a strong reputation for research, startups and technical talent.
However, AI infrastructure is expensive and increasingly concentrated among a small number of large technology companies.
CommonAI is attempting to address part of that problem through shared infrastructure and collaboration.
Its approach could be particularly relevant to smaller AI companies that need access to compute and engineering resources but cannot justify building everything independently.
The Scaling Inference Lab adds another dimension because it provides a physical environment for emerging AI infrastructure technologies to be tested at scale.
This could help bridge the gap between laboratory research and commercial deployment.
CommonAI and Sovereign AI
The organisation’s work also connects with the wider debate around sovereign AI.
Sovereign AI refers broadly to a country’s ability to develop and operate AI systems using domestic capabilities, infrastructure, data, talent and governance.
For the UK and Europe, this has become increasingly important as AI infrastructure becomes strategically significant.
CommonAI’s emphasis on UK and European AI startups, shared compute, domestic engineering capability and high assurance technologies fits into this wider conversation.
The organisation itself has published perspectives on sovereign AI and the future of AI infrastructure.
However, CommonAI should not be described as a sovereign AI model provider. Its role is more accurately characterised as supporting the infrastructure, engineering and collaborative ecosystem around AI development.
What Makes CommonAI Different?
The central difference is its attempt to combine several resources that are normally fragmented.
A startup might traditionally need to find:
- Engineers from one network
- Cloud infrastructure from another provider
- Intellectual property through licensing or internal research
- Investors through a separate network
- Industry partners independently
CommonAI is trying to bring these elements closer together.
Its members can participate in collaborative programmes, access shared resources and connect with other companies, researchers, institutions and investors within the ecosystem.
The approach is particularly relevant for deep tech AI companies where infrastructure costs and technical complexity can create significant barriers to growth.
CommonAI’s Current Programmes
As of 2026, two major CommonAI programmes stand out.
Scaling Inference
The Scaling Inference programme focuses on improving the economics and performance of AI inference.
Its £50 million lab is designed to provide a practical testbed for emerging hardware and AI systems.
High Assurance
The High Assurance programme focuses on making AI more suitable for sectors where reliability, security, compliance and auditability are essential.
Together, the programmes address two major challenges facing AI adoption:
How do we run AI systems more efficiently?
and
How do we make AI systems more trustworthy in high-risk environments?
The Future of CommonAI
CommonAI has moved considerably beyond its initial 2025 launch.
Its early positioning centred on creating a collaborative platform that could give startups access to shared compute, intellectual property and investment.
By 2026, the organisation had developed this model into specific programmes backed by major institutions.
The ARIA backed Scaling Inference Lab is now a central part of its infrastructure strategy, while the High Assurance programme addresses the practical barriers to AI adoption in regulated industries.
Corporate participation is also growing.
Barclays has joined the High Assurance programme, while Oriole Networks has joined the Scaling Inference Lab. CommonAI member Callosum has also gone on to raise significant funding for its heterogeneous AI compute platform.
The next stage will depend on whether the organisation can translate this ecosystem model into measurable improvements in AI infrastructure, development costs, deployment reliability and commercial outcomes for its members.
Frequently Asked Questions About CommonAI
What is CommonAI?
CommonAI is a UK AI ecosystem that combines collaborative engineering, shared intellectual property, AI compute infrastructure and investment partnerships to support AI startups, enterprises, academia and public sector organisations.
Is CommonAI based in the UK?
Yes. CommonAI has strong connections to both London and Cambridge. CommonAI Holdings Ltd is registered in London, while CommonAI CIC is based in Cambridge and draws on the city’s Silicon Fen technology ecosystem.
Who founded CommonAI?
CommonAI was launched by Anthemis and Cambridge AI Venture Partners, with technical leadership involving Sir Andy Hopper, Professor Robert Mullins, Dr Gavin Ferris and Mike Halsall.
What is the CommonAI Scaling Inference Lab?
The Scaling Inference Lab is a £50 million AI infrastructure initiative backed by ARIA and operated through the CommonAI ecosystem. It provides a testbed for emerging AI hardware and systems to be evaluated at scale.
What is High Assurance AI?
CommonAI’s High Assurance programme focuses on the technical foundations needed to make AI safer and more dependable in sectors such as financial services and healthcare. Its work includes verified outputs, secure execution, multi-model validation and intelligent filtering.
Who is involved with CommonAI?
CommonAI’s ecosystem includes AI startups, enterprises, public sector organisations, academia and investors. Its current team and board include people associated with Anthemis and Cambridge AI Venture Partners.
Has CommonAI received government support?
Yes. ARIA committed £50 million to the Scaling Inference Lab, making the initiative an important part of the UK’s efforts to strengthen AI infrastructure and test emerging AI hardware.
Is CommonAI an AI model company?
No. CommonAI is better described as a collaborative AI engineering, infrastructure and computing ecosystem. Its work focuses on helping other organisations build, test and deploy AI technologies rather than operating primarily as a standalone AI model developer.
CommonAI: Key Facts
| Category | Details |
|---|---|
| Company | CommonAI |
| UK presence | London and Cambridge |
| Launched | 2025 |
| Founding organisations | Anthemis and Cambridge AI Venture Partners |
| Focus | AI infrastructure, engineering, compute and shared IP |
| Major programme | Scaling Inference Lab |
| Scaling Inference backing | £50 million |
| Government partner | ARIA |
| Other major partner | Barclays |
| Key areas | AI inference and High Assurance AI |
| Membership | Startups, enterprises, academia and public sector |
| Website | commonai.com |
Conclusion
CommonAI has evolved rapidly since its launch in 2025.
What began as a collaborative platform designed to give AI companies access to shared compute, engineering and intellectual property has developed into a broader UK AI ecosystem with dedicated programmes for AI inference and high assurance systems.
The £50 million Scaling Inference Lab backed by ARIA is now one of the most significant elements of the initiative, providing a national testbed for emerging AI hardware and infrastructure.
At the same time, CommonAI’s High Assurance programme is tackling another important challenge: how AI can be deployed in industries where security, compliance, reliability and auditability cannot be treated as secondary considerations.
With Barclays joining the High Assurance programme, Oriole Networks participating in Scaling Inference, and other members such as Callosum using the ecosystem to develop AI infrastructure technologies, CommonAI is becoming more deeply connected to the practical development of the UK’s AI stack.
For the UK and wider European AI ecosystem, the organisation represents an increasingly important experiment in collaborative infrastructure: bringing engineering, compute, shared technology and capital closer together so that AI companies do not have to build every part of the stack independently.