AI running without server, data center, or empty energy costs? Meet embedl, Building a Swedish Deep Tech Startup AI Layer Cloud Provider absolutely I don't want you to know.
Embedl, a spin-out from the Chalmers University of Technology, has raised 5.5 million euros in the pre-series to promote what could be the next big change in AI.
The round, led by Chalmers Ventures, also includes Fairpoint Capital, Seb Greentech, Spintop Ventures and Stoaf, bringing Embedl's total funding to 12 million euros.
Until now, only AI professionals have been able to optimize Edge. With the hub, Embedl brings its power to all developers and product teams. The company is building core AI infrastructure for sectors such as defense, robotics and automotive. This is an industry that is putting pressure on deploying smarter systems without balloon costs or energy footprints.
At the heart of its mission is “Optimizing Edge AI Inference.” This is a once-niche field that has appeared in the spotlight. As global demand for artificial intelligence continues to surge, so will the need for models that can run quickly and efficiently on local hardware without relying on cloud connectivity or high-power data centers.
“The world needs to make AI more energy efficient. “AI applications are exploding, but we can't explode energy consumption. Our technology makes AI feasible for the mass deployment of physical products.”
From university labs to industrial craftsmen
Founded in 2022 and based on research by Professor Devdatt Dubhashi of Data Science at Chalmers, Embedl's Tools will help businesses optimize and deploy AI models for devices such as autonomous vehicles, drones, industrial robots, and surveillance systems.
Its core product, the Model Optimization SDK and its upcoming embedded hub, are designed to reduce models, reduce power usage, and dramatically reduce inference times. In real use cases, Embedl claims energy savings of up to 83%, memory savings of 95%, and faster inferences of 18 times.
This is especially important in an industry where devices must function independently. From cloud access, in real time and often under severe energy constraints.
Edge AI: From hype to competitiveness
“Edge AI” refers to the execution of machine learning models on hardware devices, not centralized data centers. In 2024, inference costs exceeded the global training costs of AI models. This is a turning point that forces businesses to rethink how they run AI workloads.
Embedl currently places itself as the efficiency layer for AI in Edge, the space currently considered one of the most commercially urgent in AI deployments.
From Bosch to Kodiak Robotics, existing clients from Embedl have already used the SDK to optimize complex deep learning models used in real devices, from tracks to wear sensors. Shubham Shrivastava, head of machine learning at Kodiak, called technology a “game-changing” for its ability to “examine cognitive blocks,” adjust performance to hardware and deploy across the platform.
Funding for wider recruitment
The new Capital will support the commercial rollout of Embedl Hub, Embedl's SaaS platform, aimed at making Edge Eye deployments accessible to teams without deep AI expertise. While SDKs are targeted at advanced users, Hubs serve a wider developer base, including those that embed AI into everyday products.
It bets on a wider trend. Edge AI is rapidly moving from R&D to essential items for cost-oriented OEMs.
“This funding is a sign that Chalmers have the technical expertise to build great AI solutions,” said Jonas Bergman, investment director at Chalmers Ventures. “We expect great things from Embedl. This is just the beginning.”
Going globally, leaving no waste
With still less than 20 teams and 23% of women, Embedl competes not only with large players but with prevalent industry mindset. Internal AI optimization is always better than outsourcing tools.
“Our biggest competitor is the 'not invented here' syndrome,” Salomonsson said. “But when the team sees our benchmarks and results, we quickly gain their trust.”
The startup has not disclosed its current assessment, but its goals are clear. Over the next 3-5 years, Embedl aims to lead edge eye space and expand its product portfolio while targeting industries that need to run AI without over-calculation.
Additionally, global AI energy usage may skyrocket and edge devices may be on time to grow across sectors.
