AI Funding Austria 2026: Open Calls, FFG and Tax Credit

Do not budget around a funding programme just because its information page is still online. The official KMU.DIGITAL notice says KMU.DIGITAL 4.0’s budget is fully exhausted, applications for consulting and implementation are no longer possible, and the programme will not continue from 2027. The earlier suggestion to wait for its budget to reopen is no longer a sound planning assumption.

This is an editorial overview, not funding approval, tax advice or a complete inventory of every Austrian programme. The status snapshot is 5 October 2026. Application windows and eligibility are separate questions. A listed maximum is not an entitlement, and a grant, loan and tax credit are not the same kind of support.

Programme status and amounts: checked 5 October 2026

Deadline boundary: AI Ökosysteme is open at this review date, not indefinitely. After 6 October 2026 at 12:00 Vienna time, do not describe this call as open without an official extension or replacement. Do not rush a new consortium application merely because a deadline is near.

aws: an implementation route, but the named calls are closed

AI-Start: Green concerns first-time AI implementation with a sustainability focus and an implementation partner. AI-Adoption concerns innovative trustworthy AI with a regulatory or standards focus. Neither is a generic reimbursement for buying an assistant subscription. Both official pages require application before project start. Do not start work expecting a future call to reimburse it retrospectively; a future programme must be assessed on its own published conditions.

FFG: distinguish ordinary innovation from frontier AI research

The Basisprogramm supports experimental development, not only training a new foundation model. Explain the novelty, technical uncertainty, methods, resources and commercial use. Scientific cooperation is possible but not compulsory there. The Innovationsscheck instead requires eligible services from a research institution; an ordinary agency invoice is not automatically such a service.

AI Ökosysteme is narrower still: its 2026 call targets cooperative industrial research advancing core Hybrid AI or Green AI technology. The FFG page explicitly says that merely adapting existing models to a use case, vertical market or heterogeneous data is insufficient. Applications must be in English. An ordinary RAG rollout should not be presented as frontier research simply to fit the call.

Forschungsprämie: 14% for qualifying R&D

The FFG describes a 14% research premium for eligible research and experimental development, including businesses without a profit. For in-house R&D, request the free annual FFG opinion through FinanzOnline after the financial year and separately submit the tax-credit application to the tax office. The tax office decides the entitlement; an FFG opinion is not a funding decision.

The BMF’s 2026 research-premium reform guidance addresses the tax-accounting basis, direct investments and the retention/correction rules for those investments. It does not turn all near-market software work into qualifying research. Document the genuine knowledge objective and uncertainty, and obtain project-specific advice on the expenditure base.

Combine support without inventing an effective rate

Do not add “80% grant plus 14% tax credit” into a headline rate. Determine the eligible base for each instrument, distinguish repayable finance, and confirm treatment of subsidized expenditure and applicable aid ceilings. Keep each cost allocated to the relevant project or phase and obtain the funding body’s and tax adviser’s assessment before claiming a combination. The detailed discussion belongs in our guide to combining aws, FFG and the research premium.

Which route should you investigate first?

1. Using existing AI tools: justify the rollout on its own business case. Do not treat closed aws or KMU.DIGITAL budgets as available cash.
2. Genuine experimental development: discuss the technological uncertainty and financing structure with FFG Basisprogramm.
3. A bounded research-partner project: consider the Innovationsscheck and obtain the eligible institution’s offer.
4. Documented in-house R&D: maintain records for the research premium rather than reconstructing them at year-end.
5. Preparing a 2027 budget: do not carry KMU.DIGITAL forward. Treat unannounced successor programmes as unknown, not committed funding.

What to have ready before you commit money

Prepare a one-page scope with the smallest useful deliverable, a cost plan separating staff and third parties, the technical uncertainty where R&D is claimed, the intended project-start date, evidence of your own financing and an owner for application records. Get the programme-specific interpretation of “project start” before placing binding orders. Approval, eligible timing and cash disbursement are separate milestones.

Funding should support delivery, not dictate unnecessary scope. Our AI consulting guide for Austria separates workshops and rollouts from custom builds; the software-agency selection guide explains how to assess the implementation partner. These are different decisions from choosing a grant.