KMU-innovativ: ICT and software-intensive systems

Up to 70% funding for leading-edge AI research, straight from the federal government to your SME.

KMU-innovativ IKT-SWS is the AI and software strand of the BMBF/BMFTR's KMU-innovativ initiative. It funds small and medium-sized companies carrying out high-risk leading-edge research in artificial intelligence, machine learning and software-intensive systems. The project management agency is VDI/VDE-IT (PT KI Berlin). The procedure has two stages: first a project outline (max. 12 pages), then, on invitation, the full application through easy-Online. Outline deadlines are 15 April and 15 October.

Eligibility

SME status under the EU definition (GBER Annex I): fewer than 250 employees and annual turnover of no more than €50m or a balance sheet total of no more than €43m, with linked and partner companies counted pro rata
The project has to fall within the technology field of ICT / software-intensive systems: AI, ML, hybrid AI, agentic systems, computer vision, NLP, data engineering or software-intensive systems
Leading-edge character with high scientific and technical risk and significant innovation beyond the international state of research; simply applying existing AI services isn't enough
The project qualifies as industrial research or experimental development within the meaning of Art. 25 GBER
An operating site in Germany
Not an undertaking in difficulty under Art. 2 no. 18 GBER; no outstanding EU recovery order
Evidenced R&D capability in AI and software (at least one qualified R&D specialist or cooperation with a relevant research institution)
For joint projects: the SME has to lead the consortium. Universities or research institutions may not be consortium leader (a requirement of PT KI Berlin)
The project may only start once the funding decision has been received (the ban on an early start under § 44 BHO and Art. 6 GBER)

Eligible costs

Staff costs: gross pay including the employer's social security contribution, apportioned to the R&D project (ML engineers, data scientists, research engineers, software architects)
Materials and consumables: sensors, test hardware, storage media, annotation material and the like
Equipment and plant (depreciation share): pro rata depreciation on GPU workstations, edge devices, servers, FPGAs, inference accelerators
Travel: project-related business travel under the BRKG (conference attendance, consortium meetings, pilot customer visits)
External services and subcontracts: cloud compute costs (AWS, Azure, GCP and others for model training), external datasets, annotation providers, specialist consultancy, security audits, external validation, all requiring project-specific attribution
Universities as joint partners receive up to 100% of eligible expenditure (AZA) plus a 20% project flat rate on the grant amount

Common pitfalls

Missing the deadline

The next deadline after 15 April 2026 is 15 October 2026. Miss it and you wait six months for the next chance. Finish the outline at least 2–3 weeks beforehand: the online tool can be overloaded on the deadline day.

Leading-edge character not evidenced

Merely applying available LLMs or hyperscaler AI services with no methodological research contribution leads to rejection. The outline has to evidence the leap beyond the international state of the art (NeurIPS, ICML, CVPR, ACL, ICLR) with benchmarks.

The wrong project management agency

AI- and software-centred projects belong with VDI/VDE-IT (PT KI Berlin), not with DLR-PT, PT Jülich or PTKA. Submitting to the wrong agency leads to delay or rejection.

A consortium without SME leadership

In joint projects an SME has to lead the consortium, even where a university is the substantive driver. Outlines with a different consortium leader are formally rejected by PT KI Berlin.

No data strategy or AI Act classification

If there is no statement on data provenance, GDPR compliance and, where relevant, AI Act classification (high-risk AI), the review panel judges the exploitation plan implausible. Address it early, already in the outline.

Starting the work early

Orders, contracts and hiring for the project before the funding decision are prohibited. A breach risks full clawback. Submitting the outline does not count as starting the project.

The application process

1 — Check SME status and the thematic field. Verify SME status under GBER Annex I (including VC holdings and linked companies). Check the project core: does the research focus lie in AI, ML or software-intensive systems? If in doubt, contact the PT KI hotline: kmui-ikt@vdivde-it.de, tel. 030-310078-498 (Tue 9–11, Thu 13–15).
2 — Decide between a joint project and a single project. In a consortium with a university or research institution (+15% funding rate, stronger assessment) the SME has to lead. Involve cooperation partners early.
3 — Write the project outline (max. 12 pages). The outline should be exactly 10–12 pages (A4, 11pt, 1.5 line spacing). Structure: topic and objective, state of the science with a comparison to the state of the art, the innovative R&D approach, a work plan with measurable milestones, an exploitation plan, team qualifications. Sketch the data strategy and AI Act classification already at this stage.
4 — Submit the outline by the deadline. Submit through PT KI Berlin's online tool (vdivde-it.de/de/kmu-innovativ-ikt). Next deadline: 15 October 2026. The outline has to be submitted with a short-form SME self-declaration and a company profile. For a consortium, attach the partner list and division of tasks.
5 — Wait for the assessment. PT KI Berlin's review panel assesses the outlines within around 2 months of the deadline. Positively assessed outlines receive an invitation to submit a full application.
6 — Submit the full application through easy-Online. The deadline is typically 3 months after the invitation. Submit through foerderportal.bund.de/easyonline/: AZK (companies) or AZA (universities), a detailed project description, the formal SME declaration (GBER Annex I), an exploitation plan, a cooperation agreement for consortia, the last 2 sets of annual accounts, a commercial register extract and CVs of the project leadership.
7 — Funding decision and project start. Around 2–3 months after the full application, the funding decision is issued with ancillary conditions (NKBF 2017 for companies, NABF for universities). The project may only begin once the decision has been served.

How upsmart helps

We match your project against more than 2,000 programmes and establish whether PT KI Berlin or another agency is responsible. We determine the R&D category — industrial research or experimental development — calculate the maximum funding rate including the cooperation bonus under Art. 25 GBER, validate SME status including linked companies and VC holdings, and keep the next deadline in view (15 October 2026). We check cumulation with the Forschungszulage and the GBER intensity limits for overlaps. You decide, your team submits.

Combining it with other programmes

Combining with the Forschungszulage (FZulG) is possible, but exclusively for different cost blocks. Costs funded by KMU-innovativ may not simultaneously feed into the FZulG assessment base. Combining with ZIM or Horizon Europe for the same costs is ruled out. Cumulation with other programmes is permitted as long as the GBER aid intensity limits (Art. 25 (5)–(6)) are observed.

Frequently asked questions

Only SMEs as defined by the EU: fewer than 250 employees and annual turnover of no more than €50m or a balance sheet total of no more than €43m. The registered office and operating site have to be in Germany. Unlike ZIM, larger companies cannot apply.

Small SMEs receive up to 70% of eligible costs for industrial research, medium-sized SMEs up to 60%. For experimental development it is 45% (small) and 35% (medium-sized SMEs). If you cooperate effectively with a research institution, up to 15 percentage points are added. The guide figure for eligible costs per SME share is €500,000.

The procedure has two stages. First you submit a project outline (max. 12 pages) through PT KI Berlin's online tool; the deadlines are 15 April and 15 October. After a positive assessment (around 2 months' wait) you are invited to submit the full application through easy-Online. The agency is VDI/VDE-IT, PT KI Berlin (kmui-ikt@vdivde-it.de).

The three most common mistakes: (1) no leading-edge character — anyone merely applying existing LLMs or AI services without a methodological contribution is rejected; (2) an outline longer than 12 pages — pages beyond that aren't read; (3) the wrong agency — AI and software projects belong with VDI/VDE-IT, not DLR-PT or PT Jülich.

Yes, a combination is possible, but only for different cost blocks. Costs funded by KMU-innovativ may not feed into the FZulG assessment base. You can claim your own contribution and staff costs not covered by KMU-innovativ for the Forschungszulage.

Read on in the blog

Funding strategy across the innovation process

Primary sources

Förderdatenbank Bund — KMU-innovativ ICT— Federal Ministry for Economic Affairs and Climate Action
VDI/VDE-IT — KMU-innovativ ICT (project management agency)— VDI/VDE Innovation + Technik GmbH
Thematic portal on AI and software-intensive systems— VDI/VDE Innovation + Technik GmbH / PT KI Berlin
BMFTR — KMU-innovativ programme page— Federal Ministry of Research, Technology and Space
easy-Online — federal application portal— federal authorities / BMBF

A 30-minute demo on your actual case. We'll show you how the platform structures funding programmes and makes overlaps with other schemes visible.