Key Points
Google DeepMind has released WeatherNext 3, an AI weather model for energy infrastructure operators The model updates every hour and forecasts wind speed at turbine heights up to 100 meters It also projects solar radiation levels to estimate energy generation from photovoltaic panels WeatherNext 3 targets grid operators balancing power supply as renewable capacity grows DeepMind frames the tool as a direct contributor to faster adoption of clean energy
Google DeepMind has launched WeatherNext 3, a weather forecasting model designed for wind farms, solar producers, and electricity grid operators.
The company says the model updates every hour and forecasts conditions that affect renewable power generation.
DeepMind described the tool in a post on Sep. 14 and published additional technical detail through its blog.
What WeatherNext 3 Measures
WeatherNext 3 generates two primary forecast streams. The first covers wind speed and direction at heights up to 100 meters. That altitude matches where turbine blades operate on most modern onshore wind farms. The second stream forecasts cloud cover and incoming solar radiation. Grid operators use that data to estimate how much electricity solar panels will generate in the hours ahead.
Both streams refresh on an hourly cadence. DeepMind says that update frequency matters because weather conditions affecting output can shift within a single trading period on power markets.
The company framed the product as infrastructure for the energy transition. Giving teams advance notice of changing weather allows them to match clean energy supply to consumer demand. It also helps balance power grids when renewable generation is variable.
Also Read: Bitget Says Non-Crypto Trades Hit 40% As It Pushes Past Digital Assets
Why Grid Balancing Needs Better Forecasts
Electricity grids have historically relied on dispatchable generation, meaning power plants that can be switched on or throttled on demand. Coal, gas, and nuclear plants fall into that category.
Wind and solar do not. Their output depends on conditions that operators cannot control. A grid with a high share of renewables must anticipate generation fluctuations in advance rather than react to them.
Forecast errors compound across a grid. A wind farm overestimating its output can leave a system operator short of supply at peak demand. Underestimating it wastes capacity and forces costlier backup generation online.
Hourly-resolution AI forecasting addresses that gap directly. Higher accuracy in the one-to-twelve-hour window gives grid operators time to arrange storage dispatch, demand response, or backup capacity with minimal cost.
DeepMind has worked on weather modeling for several years. Earlier research produced GraphCast, a global weather model that outperformed European Center for Medium-Range Weather Forecasts benchmarks on several metrics.
WeatherNext 3 applies that foundation to energy-specific use cases.
The AI Infrastructure Angle For Crypto Readers
AI model deployment at this scale intersects with the decentralized compute narrative active in crypto markets. Networks including Bittensor (TAO) and Render (RNDR) have each framed their value proposition around the growing demand for distributed AI infrastructure.
A production weather forecasting product from a leading AI lab illustrates the real-world compute demands that decentralized networks are positioning to serve. DeepMind runs WeatherNext 3 on proprietary infrastructure. Whether open or distributed alternatives can match that performance remains an open question in the sector.
Elon Musk noted earlier Monday that AI data centers are contributing to lower electricity prices for consumers. That view runs parallel to DeepMind's framing, in which better AI forecasting reduces the cost of managing renewable supply.
Also Read: Grok 5 Emerges As Musk’s AGI Target While AI Safety Warnings Intensify
What Comes Next
DeepMind has not disclosed which grid operators or energy companies are actively using WeatherNext 3. Commercial partnerships in this space typically involve utilities, independent power producers, and energy trading desks.
The model's accuracy over extended forecast horizons, beyond six hours, will determine its utility for day-ahead power market bidding. That is where the largest volume of electricity contracts settles, and where forecast errors are most expensive.
Competing AI weather models from startups including Tomorrow.io and existing services from national meteorological agencies are also targeting the same market. DeepMind's entry raises the competitive floor for forecast accuracy across the sector.
Read Next: Revolut Hackers Start Publishing Client Files, Demand 10,000 Bitcoin
