How to Raise for an AI Startup in the UK (2026)
What UK investors ask AI founders, which tax reliefs and public compute programmes apply, and how to prepare the evidence.
The market you are raising in
AI is where much of the money went last year. The British Business Bank's 2026 equity tracker reports that investment into AI companies reached a record £5.4 billion in 2025, accounting for 44% of total equity investment into UK smaller businesses and 26% of deals.
Read those two numbers together. AI took a much larger share of the money than of the deals, which points to large rounds at the top. For a first-time founder, investor appetite is real and so is the competition for attention.
The same report says seed deal numbers fell 27% in 2025 and seed stage companies took longer to raise. So I'd plan an AI seed round with the same care as any other: a tight list, clear evidence and enough runway. This is general information, not financial, tax or legal advice.
What investors will ask an AI founder
I'd expect the first question to be about defensibility: what do you have that a competitor using the same foundation model lacks? Good answers are specific. Proprietary data, a workflow you understand better than anyone, distribution into a customer group, or measured performance on a task that matters.
Expect questions on unit costs. Know what you pay for inference and training per customer or per task, how that changes with volume, and what happens to your margin if a model provider changes its prices.
Bring evidence of quality. A short evaluation document that shows how you test the system, on what data and with what results does more than a demo, because investors see a great many demos.
Data rights and data protection
Investors will check that you have the right to use the data your product depends on. Keep the licences, customer contract clauses and terms of use that give you those rights in one folder.
If you process personal data, the Information Commissioner's Office has guidance on AI and data protection covering accountability and data protection impact assessments, lawfulness, transparency, fairness and accuracy. The ICO notes that the guidance is under review because of changes made by the Data (Use and Access) Act, so check the current version.
I'd complete a data protection impact assessment for the main product before the raise and put it in the data room. It answers a diligence question in advance and it forces you to write down where your data comes from.
SEIS, EIS and knowledge-intensive status
Most UK angels will ask about SEIS or EIS. EIS has a higher tier for knowledge-intensive companies, and an AI startup that spends heavily on research may meet it. From 6 April 2026 a knowledge-intensive company can raise up to £20 million a year and £40 million in its lifetime under the venture capital schemes, against £10 million and £24 million for other companies.
Section 252A of the Income Tax Act 2007 sets the tests. On costs, at least 15% of operating costs must have gone on research and development or innovation in one of the three preceding years, or at least 10% in each of them. The company must also meet either an innovation condition, based on creating intellectual property that is expected to form the greater part of the business within ten years, or a skilled employee condition, where skilled employees make up at least 20% of full-time equivalent staff. Ask your adviser how the tests apply to a young company.
Investors benefit as well. HMRC's guidance gives an annual EIS investment limit of £2 million where at least £1 million of that goes into knowledge-intensive companies. Mention your status in the pitch if you have it, and get advance assurance to back it up.
R&D tax relief for compute-heavy companies
R&D tax relief matters to AI companies because of what counts as a cost. HMRC's guidance says that for accounting periods beginning on or after 1 April 2023, qualifying expenditure includes data licence costs and cloud computing costs. Cloud computing covers data storage, hardware facilities, operating systems and software platforms.
Under the merged scheme the R&D expenditure credit rate is 20%. Loss-making, R&D-intensive SMEs can use enhanced R&D intensive support, which gives an extra 86% deduction on qualifying costs and a payable credit worth up to 14.5% of the surrenderable loss. The intensity test is relevant R&D expenditure of at least 30% of total expenditure. These rules apply to accounting periods beginning on or after 1 April 2024.
Build the expected claim into your forecast and label it clearly. Investors will count it as non-dilutive funding only if the claim is prepared properly. Our R&D tax credits guide covers the notification and filing steps.
Public compute and government programmes
The AI Research Resource is the UK's suite of advanced supercomputers for AI. GOV.UK describes a Rapid Access route open to UK-registered micro, small and medium-sized organisations, a Gateway route for researchers from academia, industry and other UK organisations, and an Innovator route that opened on 7 November 2025. Calls open and close through the year, so check the current page.
Sovereign AI is the government's £500 million programme to back UK AI founders. The announcement of 16 April 2026 describes fully funded access to the UK's largest AI supercomputers, with up to 1 million GPU hours available per startup, fast-track visa decisions and equity investment. Callosum was named as the first equity investment, and the unit is chaired by James Wise.
Innovate UK's BridgeAI programme supports AI adoption in agriculture and food processing, construction, creative industries, and transport, logistics and warehousing. It offers funding and support to help organisations assess and implement AI, which is worth knowing if you sell into those sectors.
Regulated sectors
Check which regulator covers the sector you sell into, because its rules apply to your product whether or not it uses AI.
If your product supports diagnosis or treatment, the Medicines and Healthcare products Regulatory Agency says many software and AI products are regulated as medical devices. If it carries out a regulated financial activity, the Financial Conduct Authority's authorisation process applies. Our healthtech and fintech guides cover the timings.
Investors in these sectors will ask for your regulatory plan in the first meeting. A one-page summary of what applies, what you have done and what remains is enough to start.
Building the investor list
With so much AI capital in the market, the work is in finding the investors who back your kind of AI company at your stage. A fund that leads large growth rounds in foundation models is a poor match for a small seed round in vertical software.
I'd sort investors by four things: stage, sector focus, cheque size and a relevant deal in the last 12 months. Then look for the partner who has backed a company with a similar buyer, because they will understand your sales cycle.
Lead your outreach with evidence. One line on customers or measured results, one line on what is proprietary, and the amount you are raising.
See which UK investors fit you
Give us your website and raise. We rank UK investors against your company and explain each fit. Your top 10 are free.
Find my investorsSources
- British Business Bank: Small Business Equity Tracker 2026
- ICO: Guidance on AI and data protection
- GOV.UK: EIS and VCT changes, investment limit increase and restructure
- Income Tax Act 2007, section 252A
- HMRC: Tax relief for investors using venture capital schemes
- HMRC: Check what R&D costs you can claim
- HMRC: R&D tax relief, the merged scheme and ERIS
- GOV.UK: AI Research Resource
- GOV.UK: First backing through the UK's Sovereign AI
- Innovate UK Business Connect: BridgeAI
- MHRA: Software and artificial intelligence as a medical device