EDP (Energias de Portugal) is a Portuguese energy utility group with electricity generation, distribution and renewables operations.
How EDP works with startups
Energy Starter is EDP's open innovation program, now in its 9th edition. The 2025 cohort went through an online pitching round, follow-up meetings with EDP business units, and a two-day bootcamp before an Open Day presentation.
Selected startups run fully funded pilots with access to EDP's data and assets, backed by EDP Ventures. Across its editions the program has run more than 150 pilots, produced 40 commercial implementations and 30 venture capital investments, and generated an estimated 140 million euros in business.
EDP uses PVRADAR's SDK and soiling database to develop proprietary site-specific soiling models. EDP aimed to develop models based on its experience, reducing uncertainty early in the project lifecycle. The partnership started in 2025; 50 additional soiling datasets from locations worldwide were added to the private measurement database. EDP engineers can assess soiling risk for any location worldwide in just minutes, using shared site-specific models.
EDP worked with Beebop to model the value of residential EV charging and PV systems across smart charging scenarios. EDP needed a unified view of the value of EV flexibility across behind-the-meter and front-of-the-meter layers, and alignment across trading, retail and executives. Simulation of EV-only and EV+PV scenarios using two years of household data, market prices and charging-session data from EDP's connectivity partner. 5x value multiple increase due to targeted marketing; modelled end to end across the value chain for EV + PV. Building on these results, EDP is preparing the design and launch of commercial flexibility products for EV users.
EDP, the largest Portugal electrical utilities company, worked with Defined.ai on AI-enabled detection models for inspecting electricity poles, as a proof-of-concept. Pole inspection meant flying specialist teams by helicopter to photograph poles and fill in damage reports by hand, which was slow, labor-intensive and expensive. A proof-of-concept pole detection model was trained on 3,000 of 12,500 gathered images; damage detection models used 900 high-resolution images. The results are described as promising, and the models put EDP on the path to faster, cheaper and more accurate asset management.
These are selected publicly described collaborations, not a complete procurement record or a measure of spending.
Help keep this profile accurate
Represent this team, or spotted something that changed? Open the map profile to suggest a correction, request verification or report a removal. Submissions are reviewed before publication.