R&D Tax Credit for Software
& IT
Software companies are among the biggest beneficiaries of the FZulG. Companies developing proprietary algorithms, AI systems or custom software can have up to 25% of staff costs refunded through the tax authority.
- •Software companies can recover up to 25% of R&D costs through the R&D Tax Credit.
- •Eligible: new algorithms, AI/ML development, novel architectures, innovative database solutions.
- •Founder self-contributions: €100/hour (max. 40h/week) – up to €52,000 R&D Tax Credit per person annually.
- •Even open-source projects are eligible, provided technical uncertainty and systematic research are present.
Why Software Companies Benefit Most
The software industry has one of the highest approval rates at the BSFZ: around 85% of submitted IT projects receive a positive certificate. The reason: software development frequently meets the criteria for experimental development under §2(1) FZulG, as technical uncertainties in algorithm design, system architecture or performance optimisation are ubiquitous.
Particularly relevant for the sector: since the Growth Opportunities Act 2024, material costs and investment depreciation are also eligible – an advantage for companies using cloud infrastructure or specialised hardware for R&D. The maximum assessment base of €12 million allows refunds of up to €4.2 million (SME) or €3 million per year.
NOVARIS has already guided more than 15 software companies through the application process – from AI startups to established ERP providers. Typical project volumes range between €80,000 and €450,000 in refunds per year.
What is funded in
Software & IT?
These project types are typically eligible under FZulG §2 – provided technical uncertainty existed and the objective could not be achieved with standard means.
AI & ML Systems
Development of proprietary algorithms, neural networks, NLP models or computer vision systems that go beyond the current state of the art.
Custom Software & Architectures
Development of industry-specific software solutions, novel system architectures or proprietary databases for which no off-the-shelf solutions exist.
Embedded Systems & IoT
Firmware development, real-time systems and proprietary communication protocols for embedded hardware – from microcontrollers to FPGAs.
The 4 FZulG Criteria for
Software Projects
The outcome was not predictable with certainty at the start of the project.
Structured development following scientific methods, documented and reproducible.
The project generates new knowledge about relationships, methods or techniques.
The result is a new or substantially improved product, process or service.
Pure software development without a physical product is fully eligible. No prototype or patent is required.
Eligible costs include: gross wages of in-house developers and up to 70% for contract research.
Typical refund per developer/year: €12,500 – €18,750
Internal estimate: €150,000. After NOVARIS analysis, additional R&D components identified in infrastructure and algorithm engineering.
15+ projects supported, 100% approval rate. Not a single application rejected.
Without vs. with NOVARIS — typical difference
Illustrative example based on average client results. Actual results may vary.
Industry-Specific Requirements for Software & IT
The R&D Tax Credit for software development places special demands on project documentation. Unlike hardware-oriented sectors, there is no physical prototype – instead, the source code itself serves as the central evidence of R&D activity. The BSFZ assesses whether a technical uncertainty was overcome that goes beyond the current state of the art. Pure adaptation work, configurations or the integration of existing frameworks are not considered eligible research. The key distinction is between routine development and genuine experimental development under § 2(1) FZulG.
The choice of development methodology significantly influences the funding documentation. With agile methods (Scrum, Kanban), sprint-by-sprint allocation of R&D components is recommended: each sprint is analysed to identify which user stories genuinely break new technical ground. In waterfall projects, phases such as conception, prototyping and validation can be more clearly delineated. In both cases, commit histories, code reviews, technical specifications and test protocols are essential evidence documents. We support teams in preparing their existing development processes – whether GitLab CI pipelines, Jira boards or Confluence documentation – for the BSFZ application.
The BSFZ criteria for software novelty are based on the OECD's Frascati Manual. A project is eligible if it systematically seeks new knowledge or applies existing knowledge in a novel way. Typical eligible projects include: development of new algorithms for machine learning, architecture of novel real-time systems, research into database optimisations beyond known methods, or the design of innovative security architectures. The challenge often lies in clearly demonstrating the experimental nature of the development – even to reviewers without deep IT expertise.
Particularly relevant for IT companies: Even open-source contributions and the development of internal frameworks can be eligible, provided a demonstrable R&D component exists. Cloud-native development, microservice architectures and DevOps innovations offer numerous starting points. Our consultants with a computer science background typically identify 30–40% more eligible project components in standard software portfolios than generalists.
Typical Funding Amounts in the Software Sector
Calculation Example: SaaS Startup with 8 Developers
- • Gross wage costs R&D staff: €640,000 / year
- • Thereof eligible R&D share (approx. 60%): €384,000
- • R&D Tax Credit (25%): €96,000 / year
- • Cumulated over 3 years: up to €288,000
Calculation Example: Mid-Size IT Service Provider with Own Product
- • 3 developers on R&D project (gross wages: €240,000)
- • Plus contract research at university: €80,000
- • Eligible assessment base: €296,000 (wages + 70% contract research)
- • Annual R&D Tax Credit: €74,000
Typical Eligible Software Projects in Detail
The range of eligible R&D projects in the software sector is considerably broader than many companies initially assume.
Our tip: Many software companies underestimate the eligible portion of their daily development work. Even internal tool development, compiler optimisations or researching new testing methods can be eligible. In a non-binding initial consultation, our computer science experts typically identify 3–5 additional eligible project areas that have remained untapped.
Frequently Asked Questions
Further Resources on the R&D Tax Credit
Deepen your knowledge with our specialized resources:
Typical R&D Projects in IT
Over 70% of Forschungszulage applications in the IT sector involve software development projects. But which initiatives does the BSFZ actually recognize as eligible?
AI & Machine Learning Development
Developing novel algorithms for natural language processing, computer vision, or predictive analytics – including data preparation research and model architecture exploration.
Cloud Migration & Microservice Architectures
Researching novel migration patterns, developing automated containerization solutions, or designing self-healing microservice topologies with innovative orchestration approaches.
Cybersecurity Tools & Cryptography
Developing novel intrusion detection systems, post-quantum encryption methods, or zero-trust frameworks – often involving technical uncertainty about effectiveness against unknown attack vectors.
Custom ERP & Platform Development
In-house development of industry-specific ERP modules with innovative data models, real-time synchronization methods, or AI-driven process automation – clearly distinguished from mere configuration.
IoT Platforms & Edge Computing
Research on energy-efficient communication protocols, real-time edge architectures, or over-the-air update mechanisms for safety-critical embedded systems.
Did you know? According to the Stifterverband, approximately 38% of all approved Forschungszulage projects fall within the IT and software sector – making it the largest single beneficiary of the FZulG. Yet many IT companies fail to fully leverage their funding potential because they don't distinguish routine development from experimental development.
