Dresden-based Sparxity closes a Series A taking total funding to $90 million, backed by Lockheed Martin Ventures, GlobalFoundries, and the EIC Fund. Europe's first sparse-native AI chip company is now deployed at JUPITER, the continent's fastest supercomputer.
SPARSE AI CHIPS
Sparxity closes Series A, bringing total funding to $90M with US defence and chip industry backing
On 10 November, Dresden-based Sparxity (formerly SpiNNcloud, a TU Dresden spin-off) closed its Series A, taking total funding to $90 million. The round was led by ICM HPQC Fund, with participation from Lockheed Martin Ventures, GlobalFoundries, GS Futures, Cloudberry Ventures, TGFS, and the EIC Fund.
Sparxity builds chips for dynamic sparsity: at inference, most of a neural network's nodes are inactive for a given input. The chip skips unused parts, claiming up to 44x faster AI inference on ultra-sparse models and 18x less energy on certain machine learning workloads. The company was selected for the EIC Accelerator in 2025, receiving €10 million in grants and equity.
Revenue quadrupled between 2024 and 2025. Sparxity has shipped over 70,000 chips across three designs and counts more than 30 customers, including major supercomputing centres. Its most significant deployment: a multi-million-dollar system at Jülich Supercomputing Centre (home of JUPITER, Europe's fastest supercomputer), installed in October 2026. T-Systems (Deutsche Telekom's B2B division) has signed a letter of intent to bring the technology into its data centres.
WHY IT MATTERS
European sovereign AI compute gets its first chip company at venture scale
Most European AI infrastructure runs on Nvidia GPUs: hardware designed, manufactured, and controlled in the US. Sparxity's thesis is that sparsity-native chips represent a structurally different approach. Rather than competing on dense compute throughput, it targets the energy and latency cost of sparse inference, where most of a network is irrelevant to a given task.
GlobalFoundries joining as an investor matters: GF operates fab facilities in Dresden. A European AI chip company, backed by a local fab operator, deployed at Europe's fastest supercomputer, and EIC-selected is a coherent story about sovereign compute infrastructure.
For founders: Sparxity's path to Series A ran through the EIC Accelerator, which provided €10 million in grants and equity before this round. The 2027 EIC Work Programme is funded at €1.8 billion and includes the EIC Accelerator. If you are building in AI hardware, edge compute, or energy-efficient inference, that route is real and is not crowded.
The weekly briefing
Funding calls, VC news and key deadlines. Free to read; unsubscribe anytime.
