Frontier AI: SME Champions, Phase 1

Overview

Innovate UK will invest up to £3.1 million in UK SMEs developing ambitious frontier artificial intelligence and machine learning technologies.

The competition is designed to help SMEs test whether technically ambitious AI and ML innovations can work in practice before progressing towards larger-scale development and commercialisation.

For this competition, frontier AI means AI or ML systems that achieve state-of-the-art performance against established benchmarks or introduce genuinely new-to-the-world capabilities. The technical advance must come from innovation in model or system architecture, training methodology, core learning algorithms or core control algorithms.

Phase 1 focuses on early technical validation. Projects are expected to de-risk the core technology, generate robust evidence and establish a credible route to defensibility and scale.

This is the first stage of a planned three-phase funding pipeline. Only successful Phase 1 applicants will be invited to apply for Phase 2, subject to business case approval and the launch of a separate competition.

At the end of Phase 1, successful projects must produce a technical white paper covering technical progress, validation evidence, remaining risks, an innovation defensibility plan, Phase 2 readiness and business scaling assumptions.

Scope

Projects must focus on novel AI and ML technology or methodology development and validate their Phase 1 objectives against clear metrics.

Projects must have a credible route to defensibility, such as protectable intellectual property, proprietary data advantages, specialist know-how or another credible barrier to entry, alongside a route to scale.

Projects must start at Technology Readiness Level 1 to 3 at the point of application.

The technology must have the potential to scale at pace, although applicants are not required to prove or validate scalability during Phase 1.

Projects must demonstrate that the technology can work in principle. They can focus on demonstrating a key critical component and may use synthetic or simulated data for validation.

Where an existing capability already exists, projects must demonstrate how they will achieve state-of-the-art performance. Where the capability is genuinely novel, projects must demonstrate how it will deliver a new-to-the-world capability.

AI must be the core technical contribution to the project and the main source of operational advantage.

Phase 1 projects must deliver:

• a proof-of-concept prototype of the whole stack, core architecture or critical subsystem component
• a clear validation methodology against predefined metrics
• evidence of technical novelty
• a scaling rationale covering the technology and business model
• a technical white paper

Applications must also sufficiently describe:

• the serviceable market and customers
• background IP and the route to defensibility
• technical novelty and benchmarked validation
• one key uncertainty that Phase 1 will resolve

Innovate UK intends to support a portfolio of projects across different themes, technologies, markets, technological maturities and research categories.

Key themes and topics

Projects must fall within one or more of the following themes. Applicants must focus on one specific priority within each selected theme.

Theme 1: AI-enabled health and life sciences

The focus is on developing and deploying AI for drug discovery, development and manufacturing of medicines, clinical trials and healthcare delivery.

Projects must focus on one of:

• predictive healthcare
• medicines discovery
• medicines manufacturing optimisation
• genomics and multi-omics

Theme 2: Advanced materials with AI

The focus is on building an AI-first capability across the materials research, development and innovation pipeline, including applications in aerospace, net zero technologies, defence and semiconductors.

Projects must focus on one of:

• material prediction, generative models and multi-objective optimisation
• physics ML models for discovery and simulation acceleration
• multimodal knowledge discovery platforms
• generalisable world models or data-efficient learning for materials process optimisation and manufacturing scale-up
• open-set or uncertainty-aware models for materials characterisation, defect detection and diagnosis

Theme 3: Secure AI for national security and defence

The focus is on enabling the UK defence sector to integrate secure AI-enhanced command and control and AI-enabled sensors.

Projects must focus on one of:

• advanced AI and autonomous decision systems, including AI for decision advantage and mission delivery, multi-source data fusion and analytics, and hybrid AI architectures for causal and multi-step planning with adaptive reasoning
• resilient edge intelligence, including AI systems designed for degraded, denied and contested environments, and AI-native hybrid quantum-AI and semiconductor-AI approaches for sensing, low-SWaP inference and neuromorphic technologies

Theme 4: AI safety and assurance technologies

The focus is on technical methods, evaluation infrastructure and control mechanisms that allow AI systems to be tested, constrained and verified against safety-relevant failures.

Projects must focus on one of:

• agentic runtime control and action assurance, including technologies for constraining, verifying, authorising or interrupting actions taken by autonomous and tool-using AI systems
• dynamic AI evaluation, simulation and safety test environments for evaluating dynamic system-level behaviour
• formal and runtime verification for AI-enabled autonomous or safety-critical systems
• automated adaptive adversarial safety testing, including technologies that automatically generate, discover and reproduce previously unknown AI safety failures

The competition will fund feasibility studies and industrial research projects.

Projects will not be funded if they:

• do not sufficiently quantify the serviceable addressable market
• do not sufficiently demonstrate clear background IP ownership or rights
• do not provide a specific route to defensibility
• do not align with the selected competition theme and priority
• do not provide a named baseline, metrics and numeric validation targets
• do not sufficiently address the competition scope
• primarily consist of literature reviews or requirements gathering without substantive experimental R&D
• focus on routine integration, deployment, orchestration or productisation of existing AI tools or third-party models without novel AI or ML development
• do not deliver measurable and specific objectives
• lack clear technical novelty or a genuine feasibility challenge
• do not result in defensible foreground IP
• primarily focus on non-AI or non-ML R&D
• develop fully autonomous targeting
• primarily develop hardware, sensor hardware, quantum hardware, platforms or other enabling infrastructure where AI is only a secondary feature or analytics layer
• are dependent on export performance
• are dependent on the use of domestic inputs

Project duration

1 to 3 months

Must start by 1 April 2027

End by 30 June 2027

Award value

Total eligible project costs must be between £150,000 and £250,000. The amount of grant funding available depends on the applicant’s organisation size and the applicable funding rate.

Funding rates

For feasibility studies, which evaluate a project’s potential by uncovering key factors and supporting well-founded decisions about further development, funding is available for eligible project costs of:

• up to 70% for a micro or small organisation
• up to 60% for a medium-sized organisation

For industrial research projects, which build new knowledge and skills to improve or develop products, processes or services, funding is available for eligible project costs of:

• up to 70% for a micro or small organisation
• up to 60% for a medium-sized organisation

Only UK-registered SMEs are eligible to lead and apply to this Phase 1 competition, so the large organisation funding rate is not relevant to eligible applicants.

Eligibility criteria

• Projects must be led by a UK registered micro, small or medium-sized enterprise (SME)
• Projects must be single applicant
• Subcontractors are not allowed
• All funded project work must be carried out within the UK
• Applicants must intend to exploit the results from or in the UK
• An SME can only submit one application
• Subsidy control and State aid rules apply

‍