UK AI startups doing genuine R&D actually have decent public-funding options — better than most sectors. Here’s the real map for 2026, cited to official sources. (dgm implements osFoundry as an independent partner.)

Your best routes: R&D-based funding

If your startup is genuinely developing AI — not just applying off-the-shelf models — the strongest routes are:

  • Innovate UK competitions — non-dilutive grant funding for novel, commercialisable R&D (note flagship Smart Grants are paused; other competitions run).
  • R&D tax relief (merged scheme) — and especially ERIS for loss-making, R&D-intensive SMEs (R&D ≥30% of total spend), worth ~£27 from HMRC per £100. For a pre-profit AI startup, that’s meaningful.

Finance and devolved support

Investment, not grants

The Sovereign AI Unit (launching April 2026, backed by up to £500m) invests in and supports UK AI companies — that’s investment, not a grant, so expect it to take a stake or be structured as investment.

Keep early burn low

Here’s the founder-friendly reality: you need less capital to start than the old model assumed. Usage-priced platforms, bring-your-own-key access, and local inference (running models on your own hardware) keep early costs down. Use a Start Up Loan for runway, build a lean MVP, prove traction, then raise or borrow more against evidence.

Where osFoundry and dgm fit

dgm scopes a lean MVP and implements it on osFoundry — usage-priced, no per-seat fees, bring-your-own-key, local/self-hosted inference — all of which suits a capital-efficient startup. As you scale, it self-hosts in your own cloud or uses an EU region for UK data-sensitive work (osFoundry publishes US/EU/JP regions, not a UK one).

dgm is an independent integration partner with zero integrations so far, and not a grant/investment-application service. To scope a capital-efficient AI MVP, book a consultation with dgm. General information, not financial advice.