Client Success Story
GridStor
Discover how GridStor leveraged Nixtla's TimeGPT to generate precise, long-term electricity price forecasts, reduce AWS costs by 50%, and drive smarter operational decisions
50%Cost Reduction
TimeGPT cut GridStor's operational expenses by reducing reliance on costly AWS instances
1 secForecast Workflow
Integrated API delivers forecasts in seconds, significantly accelerating decision-making
+15%Accuracy Improvement
Enhanced long-term and hourly forecasts provided GridStor with reliable data for strategic planning
Overview
GridStor, headquartered in Portland, Oregon, is a rapidly growing developer, owner, and operator of grid-scale battery energy storage systems (BESS). The company focuses on integrating renewable energy sources to enhance grid reliability and efficiency.
Backed by Goldman Sachs Asset Management, GridStor is well-capitalized to pursue its mission of building a more resilient and sustainable energy infrastructure across the United States. In the past year, it expanded aggressively into ERCOT and CAISO regions, including a new 150 MW/300 MWh facility in Texas
To succeed in increasingly volatile electricity markets, GridStor needed a forecasting solution that could keep pace with changing dynamics. With Nixtla, GridStor transformed its ability to make financially sound, data-backed decisions in real time
The Critical Nature of Energy Price Forecasting
In today's renewable-driven power markets, electricity prices are more volatile than ever. With daily price changes up to 20× greater than stock market fluctuations, and intraday price spikes exceeding 1000% volatility, accurately forecasting these dynamics is mission-critical for energy storage operators like GridStor
- · Need to decide when to charge (when power is cheap) and when to discharge (when prices spike)
- · Even a 1% improvement in forecast accuracy can save millions in operational costs
- · Traditional forecasting models struggled with extreme price volatility driven by renewable generation
TimeGPT: Transforming Forecasting Precision
GridStor partnered with Nixtla to leverage TimeGPT, an AI-driven time series forecasting engine. By integrating vast supply-demand data from ERCOT markets and renewable output into TimeGPT's models, GridStor gained the ability to predict price fluctuations with far greater precision and horizon than traditional methods
- · Significantly improved accuracy for month-ahead hourly forecasts—previously a major challenge
- · Enhanced visibility into future prices allows optimal battery charging/discharging scheduling
- · Enabled long-term planning with quarterly and yearly price trends for new project development
Studies show that advanced forecasting and optimization can increase battery revenue by 10–20%
Business Outcomes
[01]
Operational Efficiency
- 50% reduction in cloud computing costs
- Streamlined forecasting with automated AI processes
[02]
Strategic Forecasting
- Rapid 2-second forecast generation
- Significantly improved month-ahead hourly forecasts
[03]
Improved Long-Term Planning
- Data-driven decisions for new project development
- Better visibility for long-term business planning
TimeGPT's onboarding is remarkably smooth and fast. Within seconds, we have actionable predictions ready to go. This is extremely impactful for our business. TimeGPT has been a game-changer for our budgeting process, operations, and board-level reporting. The speed, accuracy, and reliability have made it an essential part of our mission-critical workflows
Senior Manager of Market Analytics, GridStor