Ekkono and the EIC Accelerator: Machine Learning on the Edge

Ekkono Solutions, founded in 2016 in Varberg on the Swedish west coast and led by CEO Patrik Durvik, makes software that trains machine-learning models directly on small connected devices. Its EIC project set out to build Synthesis, a federated learning system in which "individual device models are collected, compared, correlated & combined into supermodels", for manufacturers of industrial and automotive equipment, according to CORDIS.

The European Innovation Council granted €2,499,875 towards a €3.57 million project from May 2023 to December 2025. The final report says "all core components of the system were fully developed", including a model-agnostic federated learning method, on-device data summaries, synthetic datasets and model-similarity metrics for spotting faulty devices, and a dashboard that handles fleets of more than 10,000 models. It was validated in pilots in manufacturing and in energy and thermal systems. In November 2025 Ekkono signed a multi-year licence with Siemens Switzerland. Its 2024 revenue was about SEK 15.5 million with a pre-tax loss of about SEK 8.5 million, Hallands Nyheter reported.

Where Ekkono Stands Now

Ekkono appointed Patrik Durvik, a former Autodesk and Hexagon executive, as CEO in March 2025 to build its commercial side. Siemens Smart Infrastructure said Ekkono "expedited the integration of AI in our products", according to the licence announcement. The company targets industrial, building and product automation and the automotive sector. Its 2024 operating margin was minus 56.2%, better than minus 130.8% in 2023.

The Technology: Learning Where the Data Is

The report says operational data from connected equipment is often confidential and OEM access to it is restricted. Ekkono's software learns on the device itself, and Synthesis compares and combines the resulting models across a fleet without seeing the raw data, flagging devices that behave abnormally. The technology comes from more than seven years of research at the University of Borås, the company says.

What the EIC Project Promised, and What CORDIS Shows

The project promised federated learning for the industrial internet of things. The report documents a complete, scalable system validated in two sectors and calls it "technically mature, scalable and ready for further deployment and commercialization". It does not report paying Synthesis customers. CORDIS lists the project as closed.

1. Federated learning systemDeliveredAll components built.
2. Fleet dashboardDelivered10,000+ models.
3. Pilot validationDeliveredTwo sectors.
4. Commercial Synthesis salesUnverifiedNot reported.

The Money

Ekkono's revenue grew strongly in 2024 but it remained loss-making for the fifth consecutive year, with 20 employees. Whether the Siemens licence covers Synthesis or the core edge library, its value, and whether the EIC Fund invested under the blended award are not public.

What Cannot Be Checked

Why Europe Should Care

Operational data from industrial equipment is often confidential, and streaming it all to the cloud is "not practical or desirable", the report says. Ekkono is a small Swedish company building it. The EIC funded its fleet-learning system, and a large European industrial customer has signed on.

The Verdict

Ekkono and the EIC have not been asked for comment for this article. It is based entirely on public records and published statements, linked throughout.