Funding

Sparxity reaches $90M in total funding with Lockheed Martin Ventures backing

The Dresden chip developer is building on a San Francisco presence and US research deployments as it funds next-generation hardware for sparse AI models.

Scaling in America Editorial11 Nov 2026
Germany→San Francisco, CA
Company: Sparxity
Illustration for: Dresden’s Sparxity hits $90M total funding as Lockheed Martin Ventures backs sparse AI chips
Illustration for: Dresden’s Sparxity hits $90M total funding as Lockheed Martin Ventures backs sparse AI chips

Dresden-based Sparxity has closed a Series A backed by Lockheed Martin Ventures and GlobalFoundries, bringing cumulative funding secured by its team to $90 million. The financing adds strategic investors as the German AI chip developer builds on an existing US presence, including a San Francisco branch and deployments at American research institutions.

ICM HPQC Fund led the round, announced on November 10. Other participants included GS Futures, Cloudberry Ventures, TGFS and the European Innovation Council Fund, alongside angel investors Richard Socher and Nancy Xu. Existing investors Atlantic and Silicon Catalyst also participated. The Series A’s size was not disclosed separately: the $90 million figure represents cumulative funding, not the latest round alone.

Formerly known as SpiNNcloud, Sparxity develops computing systems designed to skip unnecessary operations in AI and high-performance computing workloads. Chief executive Hector Gonzalez told Tech Funding News that the financing would support a second product, a new chip tapeout for sparse AI workloads and production to fulfil current orders.

An American foothold, not a first landing

Sparxity already has customers in the US and Europe. Its named American deployments include Sandia National Laboratories and the University of Texas at San Antonio, where its technology is used on a US National Science Foundation project. Gonzalez splits his time between the Dresden headquarters and San Francisco office, according to Tech Funding News.

The company has 72 employees across Germany, France, the UK, New Zealand and the US, and is hiring. The financing announcement does not specify a new American office, a US hiring target or a local manufacturing investment.

Lockheed Martin Ventures’ participation strengthens Sparxity’s American investor connections, but the disclosed backing should not be confused with a customer order or defence deployment. GlobalFoundries brings an established technical relationship: it worked with the Dresden team on the first tapeout of the SpiNNaker2 chip in 2021.

Competing alongside GPUs

Sparxity’s proposition is that increasingly selective AI models need hardware designed for that selectivity. Sparse models activate only parts of their parameters for a given input. That can reduce computation, but the hardware must also schedule the remaining work and move data efficiently for those theoretical savings to translate into faster, cheaper operation.

The company is targeting up to 44 times faster inference on ultra-sparse models with its next-generation chip. That is a development target, not a demonstrated result from the forthcoming product. Its existing systems have delivered company-reported improvements on selected workloads, including faster sparse solvers and lower energy use for event-based machine learning.

Gonzalez positions Sparxity as complementary to Nvidia rather than a wholesale GPU replacement. “Nvidia systems are absolutely excellent and they’re excellent for dense workloads,” he told Tech Funding News. He identifies Moffett AI, a Shenzhen company founded in Silicon Valley, as a more direct competitor in sparse computing processors.

For its US strategy, that distinction matters: Sparxity is seeking a role alongside established processors in data centres, rather than asking operators to replace their entire computing stack. Its opportunity depends both on delivering the new silicon and on model developers adopting workloads that benefit from its architecture.

European foundations for commercial scale

Founded in 2021 as a TU Dresden spin-out, Sparxity developed with substantial European public support. Earlier funding included a €2.5 million EIC Transition grant in 2023 and €10 million in combined grant and equity funding through the EIC Accelerator, announced in 2025.

The company now reports three chip tapeouts, more than 70,000 chips produced, 12 racks deployed and over 30 customers worldwide. Revenue quadrupled between 2024 and 2025, although no absolute revenue figure was disclosed. Its European deployments include Leipzig University and the Jülich Supercomputing Centre.

A letter of intent with Deutsche Telekom points towards deployment in T-Systems’ data centre infrastructure, with the first deployment due in 2028, according to Tech Funding News. That prospective industrial rollout, alongside existing US research installations, frames the next challenge: turning a research-led commercial base into broader infrastructure adoption while the next chip remains in development.

SparxitySpiNNcloudLockheed Martin VenturesGlobalFoundriesSparse computingAI inferenceDresdenSan Francisco

Related Stories

Keep reading
Illustration for: Walking-assist robot WIM S: Ottobock to sell in US as wellness device, no price announced
Illustration for: Tech Giants Invest in German Robotics Start-Up at $8.5 Billion Valuation
Illustration for: Agile Robots Raises $500M at $8.5B, SoftBank Leads [2026]