5 Billion People Have No Weather Radar Overhead — Tomorrow.io Just Revealed How It Plans to Fix That
On September 10, Tomorrow.io unveiled the full instrument architecture for DeepSky — its next-generation satellite constellation built specifically to feed real-time data into AI weather forecasting models.
The problem it's solving is bigger than most people realize: roughly 71 percent of Earth is covered by ocean with virtually no ground-based weather radar, leaving more than five billion people worldwide outside radar coverage entirely. Storms routinely intensify over open water with almost no one watching, until they hit a coastline.
Here's what each DeepSky satellite will carry, and why it matters:
→ Precipitation radar — resolves the vertical structure of rain inside a storm
→ Hyperspectral microwave sounders — reads atmospheric conditions across multiple wavelengths
→ Infrared and visible imaging — tracks surface thermal activity
→ GNSS radio occultation — anchors calibration against atomic time
→ GNSS reflectometry — reads ocean winds through heavy precipitation, where other instruments go blind
The company's own framing: these instruments aren't just stacked together, their effects compound — co-locating active and passive sensors creates a continuous cross-verification loop that delivers the foundational training data required for next-generation AI forecasting.
Some concrete numbers behind the plan:
→ Tomorrow.io already operates 13 satellites in orbit today, achieving a 60-minute global revisit rate
→ DeepSky's initial phase is being built by York Space Systems under a $187 million contract
→ The company raised $175M in February specifically to fund this build-out
→ No firm launch date yet — Tomorrow.io says only "low single-digit years," with its current Gen-1 constellation continuing to operate until DeepSky comes online
What stands out to me here is the framing shift. For years, the AI weather conversation has been about better models and more compute. Tomorrow.io's CEO put it plainly: the real bottleneck isn't the algorithm anymore, it's what data even exists to feed it. As he said in a briefing: "If you don't know what's happening in real time, it's very hard to know what's going to happen in an hour, in two hours, in three hours."
That's a quiet but important admission for the entire AI industry, not just weather. The next leap in accuracy for a lot of AI systems isn't going to come from a bigger model. It's going to come from someone finally building the sensor that was missing in the first place.
If this constellation ships on schedule, the biggest winners may not be meteorologists — it could be coastal communities, shipping routes, and disaster-response teams in parts of the world that have never had real-time storm data before.
#WeatherTech #AI #SatelliteTechnology #ClimateResilience #SpaceTech #DeepSky
— 𝔖𝔞𝔫𝔡𝔢𝔢𝔭 ℜ𝔞𝔦𝔷𝔞