Weather prediction company Tomorrow.io, self-described as “Resilience Platform,” is looking to add to their capabilities with a new AI-focused satellite constellation called “DeepSky.”
Tomorrow.io is one of a number of companies that is aiming to provide focused weather and climate prediction services to corporate and governmental clients. With ongoing uncertainty about the future capabilities of the National Weather Service (NWS) and National Oceanic and Atmospheric Administration (NOAA), Congress’ likely reversal of the Trump Administration’s severe cuts to NOAA notwithstanding.
With that uncertainty, there’s increasing room in the potential market for private sector supplements to governmental weather prediction services, and a number of companies have rushed to attempt to address that market.
In Tomorrow’s case, their goal is to customize the weather predictions to the needs and requirements of individual customers, rather than to provide general predictions. In a recent webcast, Tomorrow’s Atmospheric Data Scientist Randy Chase said that the concept is to not include “every single weather event,” but to “tune these forecasts for specific needs.” That can allow for more accuracy and relevance to customers making business decisions.
Tomorrow’s Co-founder and CEO Shmon Elkabetz added in a recent CNN interview that he saw the private firms as complimentary to governmental programs. Elkabetz said that the public services “created the foundations that we all rely on,” and that he sees the relationship as more akin to the “complimentary” relationship between SpaceX and NASA. Like Chase, Elkabetz emphasized the importance of helping “translate [weather data] into business decisions and action,” giving the example of an airliner client using the information to make scheduling decisions and to avoid costly last-minute diversions.
And, in turn, the company just received what they called “operational-grade validation” from NOAA. Tomorrow’s said that NOAA found that their satellites’ microwave sounders produce “well-calibrated data,” and that the preliminary report “shows consistently strong performance across Tomorrow.io’s constellation, with specific examples including radiometric accuracy, low noise in the water vapor channels, and strong cross-satellite consistency.”
Rei Goffer, Tomorrow.io co-founder and chief strategy officer, hailed the results, saying they “validate what we’ve believed from day one: [that]…a hybrid government-commercial model is not only possible, but essential.”
Tomorrow.io and artificial intelligence
For Tomorrow.io, however, the central focus hasn’t just been on space, but on the hot topic of the year: artificial intelligence. Chase mentioned that a big part of the “tuning” is in how they apply machine learning and artificial intelligence towards their data, and Tomorrow introduced DeepSky as “the world’s first AI-native, space-based atmospheric and oceanic sensing network.”
Saying that “artificial intelligence is transforming weather forecasting, driving major gains in accuracy, speed, and efficiency,” Tomorrow.io noted that a major bottleneck is in “the global observing system itself.” Machine Learning models require vast amounts of training data in order to be functional—or ”dense, high-frequency, and diverse observations” in Tomorrow’s words—and Tomorrow believes that current space-based assets aren’t equipped to keep up with that demand.
“DeepSky is built to close this gap”, Tomorrow.io said, and will be “purpose-designed to deliver the temporal density and observational diversity required by modern forecasting systems.” The company already has an “agentic weather and climate AI,” called Gale, that was launched last summer. Tomorrow described Gale as a “tailored” solution that is “not just [a] generic copilot, but agentic AI built for the real world.”
A larger and more capable constellation
This will be the company’s second constellation, after having finished launching their tenth and eleventh satellites in their current constellation of microwave sounder weather satellites back in November. The company said earlier this month that “it has achieved a 60-minute global weather revisit rate with the completion of its satellite constellation.”
While the announcement didn’t state the size of the planned constellation, they said that it will be a “proliferated” constellation, which generally implies a large number of satellites to ensure redundancy and reliability. The goal, according to Tomorrow, is “dramatically higher revisit rates than traditional systems, enabling faster refresh cycles for global and regional models and improved prediction of rapidly evolving and extreme weather events.”
This all suggests that DeepSky will ultimately be a much larger constellation than the one they’re currently operating.
In comments to SpaceNews, Goffer also said that the new satellites will be significantly larger and equipped with “instruments of a completely different caliber.” Goffer would not share what these instruments would be, but said that the satellites will carry “multiple very-high-impact, co-located sensors.”
It sounds ambitious, but SpaceNews noted that Tomorrow has attracted attention from a wide variety of government and enterprise customers, including Ford, Uber, and the U.S. Air Force.
Two of these potential future DeepSky customers were quoted regarding the new constellation. Nikhil Ahuja, Senior Director, Planning and Supply Chain at Amazon, emphasized the importance of the task, saying that operational resilience “depends on treating atmospheric data with the same rigor as any other mission-critical infrastructure,” and that DeepSky could “create a new class of AI-driven decision systems that are more adaptive and localized.”
BNSF Railway’s CTO, Matt Garland, also weighed in, agreeing that resilience “comes from continuously sensing operating conditions and translating that intelligence into network-wide decisions,” and adding that DeepSky’s contribution to agentic AI “represents a meaningful step towards defining a new category of what’s possible.”