The EIC Accelerator is designed for breakthrough deep-tech companies, but “deep tech” is not a simple synonym for hardware, patents or university spin-outs. The programme funds companies across MedTech, biotechnology, engineering, energy, advanced materials and software, which means applicants need to understand both the official definition and how to translate it into a convincing company-specific case.
A product that can be reproduced in a weekend with standard tools will struggle to demonstrate the long development cycle, technical risk, financing gap and defensible innovation expected by the EIC. A deep-tech proposal must identify the body of work that makes the solution difficult to build, validate and copy.
Official reference: European Innovation Council, EIC Work Programme 2026.
The EIC Definition of Deep Tech
The 2026 EIC Work Programme defines deep tech as technology based on cutting-edge scientific advances and discoveries that requires continuous interaction with new ideas and results from the laboratory. Deep-tech innovations can combine advances across the physical, biological and digital domains and should have the potential to deliver transformative solutions.
The Work Programme distinguishes deep tech from high tech. High tech can refer primarily to research-and-development intensity, while deep tech concerns the underlying scientific or engineering frontier, the uncertainty of the development process and the difficulty of converting the result into a scalable product.
The accompanying characteristics refer to long and uncertain research and innovation cycles, tangible products and industrialisation processes, links to higher-education or research ecosystems, mission-driven development and repeated technical de-risking.
The Definition Is Guidance, Not a Hardware-Only Rule
The laboratory, tangible-product and university language creates a natural fit for hardware and science-based ventures, but funded-company portfolios also contain software businesses and companies that did not originate as university spin-offs.
An applicant should therefore avoid two opposite mistakes. A software company should not assume that digital technology is automatically deep tech, and a hardware company should not assume that physical complexity alone proves a breakthrough. Each must identify the scientific or technological advance, the evidence of performance and the barrier that prevents straightforward replication.
Industry Context Changes How Technology Is Understood
The same underlying digital capability can appear very different when applied to different markets. Generic software for selling consumer products can look like an ordinary commercial application, while a technically similar system used for chemical synthesis, biotechnology manufacturing, advanced agriculture or industrial engineering can sit inside a much more demanding scientific and operational environment.
This does not mean that adding scientific vocabulary turns conventional software into deep tech. The industry context matters because it determines the validation requirements, specialist knowledge, performance constraints, integration challenges and cost of failure. A project-management platform for a highly regulated or technically complex industrial process can face a defensibility challenge that generic project-management software does not.
Define the Company's Body of Work
A useful test is to ask: What has this company built that a well-funded competitor could not reproduce today? The answer is the company's body of work.
Depending on the business, that body of work can include:
- Patented materials, devices, processes or system architectures.
- Trade secrets, manufacturing parameters and hard-won process knowledge.
- Years of prototype development and performance data.
- Unique datasets, trained model weights and specialised evaluation systems.
- Regulatory, clinical or industrial validation that competitors have not completed.
- Scientific expertise concentrated in a team with unusual domain experience.
- Integrated hardware and software whose performance depends on the complete system.
A user interface is rarely a sufficient body of work because it can be observed and copied. The defensible value normally lies underneath it: the model, data, architecture, experimental evidence, process integration or technical knowledge that makes the visible product possible.
The Frontier-AI Example
Frontier artificial-intelligence companies provide a useful software example. Their products are delivered through software, but the defensibility is not the chat interface alone. It lies in model weights, training methods, data pipelines, evaluation capabilities, computing infrastructure and the research expertise required to create and improve the model.
A company such as Anthropic protects its commercial position partly by keeping its frontier models closed rather than publishing the complete weights. If the interface were the only asset, a competitor could reproduce the product quickly; the accumulated model and research work creates the meaningful barrier.
An EIC software applicant should be able to perform the same analysis on its own product. It must explain what exists beyond standard application code, why that asset is technically difficult to recreate and how the company will preserve its lead.
Deep Tech Still Needs a Market
Scientific complexity is not sufficient on its own. The EIC Accelerator evaluates whether the proposed innovation represents a substantial cost or performance improvement, can create or disrupt a market and has a credible route to international scale.
The proposal should connect the technical body of work to a customer outcome. A patent matters because it protects a performance advantage; a model matters because it enables a result competitors cannot offer; a manufacturing process matters because it changes cost, quality, yield or scalability.
Questions for Testing Deep-Tech Fit
- Which cutting-edge scientific or technological advance makes the product possible?
- What measurable improvement does it provide over the state of the art?
- Why has the problem not already been solved by an established company?
- What experiments, prototypes or operational evidence validate the core claims?
- What part of the solution took years rather than weeks to create?
- Which patents, know-how, models, data or processes prevent rapid copying?
- Why does the development require substantial patient capital?
- How does the technology create a valuable and scalable customer outcome?
- What must still be de-risked between the current TRL and commercial deployment?
How to Present Deep Tech in an EIC Proposal
The strongest narrative begins with the technical frontier and then moves through evidence, defensibility and market impact. It should explain what is new, prove what has already been achieved, identify the remaining risk and show why the resulting advantage matters commercially.
For software companies in particular, the proposal must move beyond feature lists. Evaluators and jurors need to see the body of work beneath the product and understand why a large technology company could not simply copy it after reading the application.
Deep tech is ultimately not a label an applicant chooses. It is a case built from scientific or technological novelty, accumulated evidence, difficult development, defensible assets and the potential to transform a significant market.
