Somewhere, a grid operator is staring at a heat dome.
It is the middle of a July night and the temperature refuses to drop. Air conditioners across four states are running flat out. Demand is climbing toward a number the model said was a once-in-a-decade event - and here it is again, second summer running. The operator does not need to know what the weather will probably do. The average day never breaks anything. What keeps them awake is the tail: the 1-in-1,000 hour where supply and demand cross in the wrong direction. That hour is exactly the one most forecasts were never built to see.
Sunairio was built for that hour. The Baltimore company makes grid-forecasting software that sits at the intersection of three things that rarely speak the same language - weather, climate, and energy - and translates all of them into numbers a person on a trading desk or in a control room can actually use. Its pitch is refreshingly blunt: Never be surprised by extremes again. Outliers are in. Start planning like it.
Caption: Not the weather you expect. The weather you can't afford to miss.
A trader who got tired of being surprised.
Rob Cirincione spent a career on the energy markets, where weather is not small talk - it is the single biggest variable moving the price of power. He kept running into the same wall: the tools everyone used were good at describing the middle of the distribution and nearly blind to its edges. And the edges were where the money, and the risk, lived.
So in 2020 he and co-founder Mitchell Walk started Sunairio to build the tool he wished he'd had. The premise inverts the usual forecasting instinct. Most software tries to give you one best guess. Sunairio gives you a thousand plausible ones, then shows you which of them would hurt.
The name is a small piece of wordplay - sun, air, and "io," the flow of data. It is the whole company in a breath: turn the sky into information, and the information into a decision.
The bet underneath it all is that the grid is getting harder to predict on purpose. More renewables, more electrification, more weather-dependent everything. Which is either a problem or, if you have the right math, a market.
Fifty scenarios was never risk. It was a rounding error.
Traditional ensemble forecasts run about 50 possible weather paths and call it a day. Sunairio ONE runs 1,000. That is not showing off - it is the difference between sampling the ordinary and actually reaching the rare, grid-bending extremes that live in the tails.
Forecast paths per run - more paths, more of the tail
Feeding all of this is the SHED - Sunairio High-resolution Earth Data - a proprietary historical weather dataset covering the Continental US from 1950 to today, at roughly 100 times the resolution of standard reanalysis data like ECMWF's ERA5. History, it turns out, is a very good teacher of extremes. The system then uses advanced statistics and generative AI to simulate weather that hasn't happened yet, down to a single wind farm or solar array, one hour at a time.
Caption: The outlier isn't noise. In Sunairio's world, it's the signal.
One platform, from the next hour to the next 15 years.
Sunairio ONE
Award-winning ensemble grid forecasting. 1,000 hourly forecast paths spanning load, renewables, and grid stress - so planners see the extremes before the extremes see them.
The SHED
Sunairio High-resolution Earth Data: US weather from 1950 to now, at ~100x the resolution of ERA5. The historical bedrock under every simulation.
Asset-Level Modeling
Site-specific wind and solar generation, simulated with statistics and generative AI - useful for project selection and portfolio optimization.
Load & Grid-Stress
Probabilistic load forecasts and grid-stress analysis that fold in weather variability and long-term climate trends across power-market regions.
The people who can't afford to guess.
Sunairio sells to the sharp end of the energy system: utilities planning the grid, developers picking where to build, traders pricing the next cold snap, and investors sizing decade-long bets. Coverage spans four US power-market regions and select utility footprints - with named customers and partners including:
$8M+, an NSF grant, and a power giant on the cap table.
Recent chapters.
Six things worth knowing.
- The name blends sun, air, and io - the flow of data from sky to decision.
- It runs 1,000 possible weather futures per forecast - 20x the legacy standard of 50.
- Its SHED dataset reaches back to 1950 - older than much of the grid it forecasts for.
- CEO Rob Cirincione is a former energy trader who built the tool he wished he'd had.
- It forecasts as far out as 15 years - and as near as the next 15 days.
- HQ is Baltimore's Harbor East, not a coastal climate-tech hub.
Back to the heat dome.
The operator is still up. The temperature still won't drop. But the screen looks different now. Instead of one line marching toward a cliff, there are a thousand of them - a fan of plausible tomorrows, with the ugly ones already flagged and priced. The 1-in-1,000 hour isn't a surprise ambush anymore; it's a scenario that was run, ranked, and planned for days ago. That is the quiet thing Sunairio changes. It doesn't make the weather calmer. It makes the extreme legible - early enough to do something about it.
Averages are comfortable. They are also where portfolios and power systems go to get caught out. Sunairio's whole argument is that the interesting part of the forecast was always the part everyone else was rounding away. From a 19-person office in Baltimore, it's making a bet that in a more volatile climate, the company that takes the tail seriously is the one worth having on the desk at 3 a.m.
Caption: Same night. Same heat. A thousand fewer surprises.