Skip to content
AI.info

The Pulse

Sparxity completes Series A as team funding reaches $90 million

Sparxity's Series A brings cumulative funding secured by its team to $90 million. The financing will develop chips for dynamic sparsity in AI inference, with Sparxity targeting up to 44× faster inference in ultra-sparse models.

Sparxity completes Series A as team funding reaches $90 million

Sparxity closed a Series A, bringing cumulative funding secured by its team to $90 million. The financing will develop specialized chips for fine-grained dynamic sparsity, with Sparxity targeting up to 44× faster AI inference in ultra-sparse models.

Source via.ritzau.dk

The round was led by ICM HPQC Fund. Lockheed Martin Ventures, GlobalFoundries, GS Futures, Cloudberry Ventures, TGFS and the EIC Fund also participated. Angel investments came from Richard Socher of Recursive Superintelligence and Nancy Xu of Moonhub, which the release describes as Salesforce-acquired; Atlantic and Silicon Catalyst continued their support.

Targets for sparse inference

Sparxity is developing chips for models that use dynamic sparsity in inference. Sparxity’s stated principle is to “compute only what matters,” and its systems are intended to reduce energy use by avoiding unnecessary operations and data movement.

The release reported results on selected workloads including sparse solvers that ran 10× faster, event-based machine learning with 18× lower energy use, and optimization problems with runtimes 20× shorter. Those figures describe selected workloads, according to the release; the figures were presented by Sparxity as results from its current systems for high-performance computing and AI.

Systems and customers

Revenue quadrupled between 2024 and 2025, and Sparxity says momentum continued in 2026. There were also three tapeouts, more than 70,000 chips produced and 12 racks deployed, alongside a customer base of more than 30 worldwide, including what Sparxity describes as some of the globe’s largest supercomputing facilities.

The financing follows a multi-million-dollar deployment of Sparxity systems at the Jülich Supercomputing Centre, part of Forschungszentrum Jülich. The release describes Jülich as home to Europe's first exascale supercomputer. Sparxity also announced a partnership with Deutsche Telekom to bring its technology into T-Systems' data centre infrastructure.

Company’s stated plans

T-Systems CEO Ferri Abolhassan said Sparxity was expected to join the company’s AI factory soon, and praised its energy-efficient use of AI computing power.

The new financing will help develop systems for ultra-sparse AI workloads. Sparxity recently adopted the Sparxity name, having previously been called SpiNNcloud.

Sources

Related stories

Explore

More articles