Rocket launch providers, satellite operators, and even aviation companies rely on accurate predictions about “space weather,” like solar flares and geomagnetic storms, to inform their operations. But this vital information is still supplied mainly by governments using inaccurate, older modeling techniques.
Perceptive Space , a Canadian startup that emerged from stealth on Tuesday, is looking to change that. The company is betting that gains in machine learning and artificial intelligence can improve the accuracy of space weather forecasting while also providing near-real time updates.
The company was founded by Padmashri Suresh, an engineer who cut her teeth on sounding rocket and cubesat missions at Utah State University before embarking on a NASA-sponsored PhD on space weather and machine learning. There, she saw firsthand how inaccurate information on space weather can affect launch, satellite operations, and instrument performance. After leaving the university, she moved into the tech industry, though she says she was waiting for the right time to build a space weather company.
“Fast forward to 2022, SpaceX lost close to, I think, 38 or 40 satellites , where space weather was a driver,” she said. “I thought perhaps this was that critical mass that spurred me into action to build this company.”
Space weather was back in the news last year, when a far stronger than predicted solar maximum had catastrophic consequences for some satellite operators.
NASA and NOAA collect space weather data using satellites data and radar and magnetometers on the ground, generating its predictions using bulky, fundamental physics-based models, some of which must be run on supercomputers. But even given this legacy competition, Suresh admits that “it’s a hard problem” to solve.
“You’re looking at physics at so many different scales,” she said. “You first need to understand the sun. You then need to understand this environment from the Sun to the Earth. There are so many drivers that are really impacting the overall space weather that we experience.”
But she says that we can extract more signals from data using AI, thanks to increasing access to high performance computing, and advances in prediction algorithms. There’s also simply more data, due to a greater number of satellites in low Earth orbit.
Investors including Panache Ventures, Metaplanet, 7Percent Ventures, Mythos Ventures and AIN Ventures are contributing $2.8 million to the efforts. Suresh said the capital would be used to accelerate product development from a bench-scale product to a fully fledged service for any launch provider or satellite operator in any orbit by next year. Most of the funding will go toward expanding the team from today’s 5 employees to 10 over the next year.
Perceptive already has a pilot program and a few early signups, customers that are providing early feedback, and the company aims to onboard more people to that program. Once the commercial product is live, customers will be able to sign up for a subscription that provides short- and long-term forecasting at different pricing tiers, based on the number of assets, the orbit, and other factors.
Thinking even longer-term, Suresh said that accurate space weather data will be essential for scaling the number of satellites in low Earth orbit or having a sustained human presence in space. She pointed out that the International Space Station mission controllers use space weather as an input to determine when it is safe for astronauts to perform spacewalks or when it is safe for companies like SpaceX to launch them to and from Earth. Even aviation companies use space weather to understand how much radiation exposure pilots are receiving.
“There’s a human factor,” she said.