On February 24, 2026, Eindhoven-based Axelera AI announced a new funding round of more than $250 million — which the company’s press release calls the largest investment ever in an EU AI semiconductor company. The round was led by Innovation Industries, with BlackRock and SiteGround Capital joining as new investors alongside existing backers including Bitfury, CDP Venture Capital, the European Innovation Council Fund, Belgium’s SFPIM, Invest-NL, and the Samsung Catalyst Fund. Founded in 2021, Axelera has now raised more than $450 million in equity, grants, and venture debt, and the fresh capital is earmarked for manufacturing scale, its customer success organization, and its Partner Accelerator Network.
In a market where NVIDIA owns data-center AI, Axelera is taking a different lane: it does not chase training, it specializes in AI inference on edge devices, and it competes on efficiency rather than raw throughput. Sifted’s coverage framed the raise flatly as money “to take on Nvidia.”
In-Memory Computing: Where the Efficiency Comes From
Axelera’s core technology is D-IMC, digital in-memory computing: SRAM arranged in crossbar arrays so that data is processed where it is stored. On conventional chips, the most expensive step is often not the arithmetic but moving data between memory and compute; D-IMC eliminates that movement outright, which is how the power budget stays low.
That is also how the company frames the use of funds. CEO Fabrizio Del Maffeo argues that data centers are hitting power and cooling limits while edge devices face even tighter energy and bandwidth constraints — and that “inference costs run 15 times higher than training over a model’s lifetime.” Efficiency is the business model.
The Metis and Europa Product Line
On the product side, the current flagship is the Metis AIPU: 214 trillion operations per second at roughly 10 watts, aimed at battery-powered deployments such as AI navigation on warehouse robots. It ships in two form factors — a PCIe accelerator card holding up to four chips, and a single-chip M.2 version for compact, low-power systems. The software stack is the Voyager SDK, built on Apache TVM, with a Model Zoo of prepackaged models to cut deployment friction.
The next-generation Europa platform is in development: 629 TOPS — more than double Metis — with 8 AI cores, 16 CPU cores, and 128 MB of memory. The company claims up to three times the performance per watt of competitors and cites an internal test exceeding 13,168 frames per second on computer-vision workloads. Volume manufacturing runs on partnerships with TSMC and Samsung.
Why Bet on the Edge
The commercial track record is worth noting: Axelera says it recently shipped to its 500th global customer, spanning defense and public safety, industrial manufacturing, retail, agritech, robotics, security, telecom, aerospace, and the enterprise market. The common thread is workloads where data cannot or should not make a round trip to the cloud: privacy law, sovereignty requirements, latency budgets, and bandwidth cost all push inference local.
For Europe, the round also carries industrial-policy weight. With the US and China dominating AI compute, the EU now has its first AI chip company financed at this scale, with public capital — the EIC Fund, SFPIM, Invest-NL — standing behind it. A European edge-inference supply chain is no longer just a talking point.
What It Means for Developers
Three practical effects. First, edge-inference hardware options are thickening: a 10-watt, 200-plus-TOPS part in PCIe and M.2 form factors lets vision, anomaly detection, and similar workloads run locally and indefinitely, without a cloud dependency or the recurring per-token bill that comes with one. Second, the software stack decides adoption speed — Voyager being built on Apache TVM means near-zero migration cost for teams already in the TVM ecosystem, which is the right bet for a challenger that lives or dies on switching costs. Third, every AI product builder should memorize the “inference costs 15x training” figure: shipping the model is the beginning, the long-run cost structure is set by inference efficiency, and that is precisely the reason dedicated edge silicon exists.
Sources
- Axelera AI Secures More Than $250 Million Funding on Global Commercial Growth — Axelera AI
- Edge AI chip startup Axelera AI raises $250M+ in funding round — SiliconANGLE
- AI chip startup raises $250m to take on Nvidia — Sifted
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.