Nullspace, Inc., developer of next-generation electromagnetic (EM) simulation software, announced a partnership with Eclipse Space to speed up satellite antenna design. Eclipse is using Nullspace as the simulation engine inside Eclipse’s own AI-driven design workflow, where software agents set up, run and refine antenna simulations day and night while Eclipse’s RF engineers make the decisions.
The work comes as more governments rethink who controls their space infrastructure. As space becomes critical for national security and economic competitiveness, nations need to own and control these constellations. For those efforts, cost, speed and innovation matter: each satellite needs antennas designed for its mission, and that design work is one of the slowest steps between concept and orbit.
A Decades-Old Bottleneck
Before an antenna is built, engineers simulate how it will perform. Much of the software used for that work was designed decades ago, long before cloud computing or AI. A single simulation of a complex design can take hours to weeks, and engineers typically need many rounds of adjusting, re-running and reviewing results before a design is ready for prototyping.
Nullspace was built to remove that limit. Validated on U.S. defense hardware for more than a decade, its solver simulates complete RF systems, such as a full phased array on a satellite, at a scale and complexity that older tools weren’t built to handle. Because Nullspace is written entirely in Python and can run fully headless, AI agents can drive it directly, on premises or in the cloud.
How Eclipse is Using It
In Eclipse’s workflow, AI agents handle the repetitive loop: generating a design, preparing it for simulation, running it in Nullspace, reading the results and deciding what to change next. Eclipse can run many of these simulations in parallel, significantly shortening the overall design process. The result is a force multiplier for Eclipse’s engineers: a single RF expert can direct dozens of design explorations at once, spending their time on judgment calls rather than setup and waiting.
The results come from physics, not prediction. Nullspace solves the governing equations of electromagnetics directly, so the AI decides what to try, but the answer to whether a design works is a full-fidelity simulation that engineers can trust.
“Nullspace solves the governing equations of RF physics” said Dr. Masha Petrova, CEO of Nullspace. “Yet, it has been built for the age of AI. For decades, engineering simulation software was reserved for a few experts. Tools took months to learn and required cumbersome licensing and deployment. Nullspace is designed to be run by AI agents, deployable in the cloud and on prem, and with the new consumption model it empowers creation of incredibly fast design cycles.”
Licensing that Matches how Agents Work
Eclipse is also using Nullspace’s new consumption-based licensing model, which launches publicly early next year. Instead of buying fixed seats, Eclipse’s agents check out a Nullspace license when a simulation needs one and release it when they finish, so Eclipse pays only for the simulation time it actually uses.
“Our customers are building networks they’ll depend on for everything from next-generation communications and orbital data centers to national security. This is a competitive race around the globe, and they can’t afford to be delayed,” said Derek Huerta, CEO of Eclipse Space. “Nullspace lets our engineers explore far more antenna designs in far less time while staying in charge of every decision that matters. That means faster, more cost-effective and reliable constellations for the nations and companies we serve.”