Renewables have won on cost. Solar and wind are now the cheapest sources of electricity ever created. But they've introduced a problem the industry hasn't fully solved: intermittency. Power comes when the weather allows, not when people need it.
This has turned energy trading into one of the most complex real-time markets on Earth. Prices swing 10x in hours. A single weather forecast error can cost millions. The people managing this complexity are among the sharpest in any industry. They've built careers on pattern recognition, market intuition, and the ability to synthesize vast amounts of information under pressure.
But even the best traders face a bottleneck that talent alone can't solve. The tools haven't kept pace with the market's evolution. Most platforms were architected for a grid that ran on predictable baseload generation. They show you what happened. They don't tell you why, or what to do about it. They don't reason.
Meanwhile, the data keeps multiplying. More weather models, more forecast updates, more price signals, more generation sources. A trader today might need to monitor fifteen different weather models, track price movements across dozens of market zones, and understand how all of it affects their specific positions. The synthesis required is enormous. The time available is not.
The real constraint in energy trading was never data or algorithms. It's thinking time. The research, the pattern recognition, the cross-referencing across multiple systems. That cognitive work lives in the heads of experienced traders, and there simply aren't enough hours in the day.
We built Athena to give that time back.
What Athena Does
Athena is an AI system that monitors every weather model, every price signal, and every position in your portfolio. Then it tells you what actually matters. You ask questions in plain language. You get answers you can act on.
Last week, a customer watched Athena complete an analysis in 2 minutes that usually takes their team 5 hours.
Here's what working with Athena looks like in practice.
You open your morning and ask: "What changed overnight and should I care?" Athena scans all fifteen-plus weather models it has access to, identifies the meaningful shifts, and ranks them by impact on your positions. It might tell you that EPT-2 Rapid Refresh is showing a 15% wind speed increase over North Sea wind farms, affecting day-ahead spreads between Germany and the Netherlands. It quantifies the exposure. It recommends what to review. The kind of synthesis that used to require pulling data from five different systems and thirty minutes of analysis happens in seconds.
You can go deeper. "Where's my exposure if wind underperforms by 10%?" Athena runs the scenario across your entire portfolio, stress tests the positions, and shows you exactly where the risk concentrates. "How would this strategy have performed last winter?" It backtests against two years of historical data and walks you through the results.
The questions that used to require a quant team and three days of lead time can now be answered before your morning coffee gets cold.
Athena has native access to all of Jua's proprietary models, including EPT-2 Rapid Refresh which updates 24 times per day and EPT-2 Ensemble which extends forecasts out to 60 days. It also integrates leading third-party AI models from Microsoft, Google, and ECMWF, plus established numerical weather prediction models. Fifteen-plus models in total, analyzed continuously, with automatic detection when they start diverging from each other.
Beyond weather, Athena fuses power generation forecasts, solar irradiance data, temperature anomalies, precipitation patterns, price signals, spreads, and demand indicators across all ENTSO-E zones. Every answer comes with interactive visualizations you can explore and export.
The system watches 24/7. It generates automated pre-market briefings so you start each day knowing exactly what shifted and why it matters. It surfaces signals you would have missed. A junior analyst using Athena can now ask the questions that previously required years of experience to even know to ask.
Why It Works
Most AI weather products take existing numerical models and layer machine learning on top. We went a different direction.
Over the past three years, we built our own foundation models from scratch. Trained on petabytes of atmospheric data. Designed to simulate the actual physics of weather systems, not just find statistical patterns in historical numbers. When you understand the physics, you can reason about causation. You can answer "why" questions, not just "what" questions.
In independent benchmarks, our EPT-2 models outperform the European Centre for Medium-Range Weather Forecasts. ECMWF has been the gold standard for fifty years. We also outperform recent AI weather efforts from Google and Microsoft. The accuracy gains compound when you're making decisions multiple times per day across volatile markets.
The results from teams using Athena reflect this. 80% reduction in manual research time. 10x faster from signal to decision. Continuous analysis without adding headcount. We're already serving major utilities and trading firms across four continents, running predictions for portfolios that cover more than 50 gigawatts.
Where This Goes
Weather is a trillion-variable fluid dynamics problem. Getting it right requires AI that can genuinely simulate physical systems at scale. We've built that. But weather is a proving ground, not the destination.
The same architecture that models atmospheric dynamics can model any domain where complexity overwhelms traditional simulation. Energy infrastructure planning. Climate risk assessment. Industrial process optimization. Supply chain logistics. The physical world is an information problem, and most of it remains unsolved because the tools to reason about it at scale didn't exist.
Athena is the interface to that intelligence. We started with energy traders because the need is urgent and the value is immediate. The companies that deploy this will move faster and see further than their competitors. In a market defined by volatility, that edge compounds.
But we'll end up somewhere much larger.
Marvin Gabler, CEO