Energize Capital is proud to lead the $16 million Series A investment in Nixtla, the company building foundational AI models for time series forecasting. The round was led by Energize alongside continued participation from existing investors including Greatpoint Ventures and True Ventures. With this financing, Nixtla has raised approximately $22 million to date. Energize partner Juan Muldoon will join the Nixtla board, and Energize senior associate Agustina Soriano Sergi will join as a board observer.

The Challenge

Time series data is the heartbeat of industrial operations. From power grids and batteries to factories, fleets, supply chains, and financial markets, critical systems all rely on massive volumes of data indexed to time. At a utility, this data could look like megawatts levels flowing through the grid throughout the day. For a retail company, it could look like sales per hour. Across industries, this fundamental information is utilized to track operations, analyze performance, and – crucially – make accurate forecasts.

Yet despite its importance, forecasting time series data remains one of the most challenging problems in applied machine learning. Today, in order to make strong predictions, most organizations rely on highly specialized data science teams to build bespoke forecasting models, spending weeks or months to clean and structure historical data, experiment with dozens of statistical and machine learning models, and troubleshoot models to improve accuracy.

Within energy, industrial, and climate settings, the forecasting challenge is even more pronounced. Volatility in energy demand, increasing penetration of renewables, changing supply chain dynamics, and erratic pricing environments are all driving a need for forecasting models that can be faster and more accurate.

Introducing Nixtla

Nixtla was built to address this problem. As the first foundational model for time series data forecasting, Nixtla helps teams generate predictions quickly, accurately, and efficiently. Their enterprise product, TimeGPT, acts as the data-based equivalent to a large language model (LLM), i.e.: where ChatGPT ingests terabytes of text, detects patterns, and predicts language-based outputs, TimeGPT trains on time series data and produces time-based forecasts. Through this process, Nixtla is able to deliver production-ready, customizable models, that deliver faster and better forecasts across industry verticals.

The company launched in 2021 as a collection of open-source libraries (StatsForecast, NeuralForecast, and others) that quickly became favorites of data scientists and engineers. Today, their collection called Nixtlaverse has received over 42,000 GitHub downloads and 10,000 GitHub stars.

This rigorous foundation developed into a commercial platform that enterprises use to deploy forecasting systems worldwide. Since 2025, Nixtla’s enterprise platform has served startups and Fortune 500 companies alike, touting customers like Microsoft, Decathlon, and Zalando. Nixtla has driven results across power demand forecasting, energy price prediction, battery dispatch optimization, inventory planning, anomaly detection in equipment, and more. The platform offers both API-based deployments and self-hosted implementations, allowing enterprises to maintain control of sensitive data while accessing state-of-the-art models.

The enterprise model’s extensive pre-training and “zero-shot” inference works out of the box, eliminating the need for bespoke tailoring, multiple test models, or granular troubleshooting. This not only increases time to value by 10 times but also provides up to 42% accuracy improvements over traditional methods.

The Energize Angle

Nixtla is the latest proof that the future of industrial and climate impact will be driven by foundational digital innovation. We have long invested in the data layer of energy and industrials with portfolio companies like Tyba and Amperon, where we have seen firsthand how accurate market predictions can lead to tangible climate impacts: When clean energy sources like solar or wind are accurately predicted, grid operators can reliably integrate renewables into the power mix. Similarly, when energy demand is correctly modeled, batteries can store and dispatch power more efficiently. This unlocks more clean electricity and supports a balanced, resilient grid.

But this goes beyond energy. Accurate forecasting – no matter the industry – inherently reduces waste. It allows manufacturers to avoid overproducing inventory, moves fleets more efficiently, and identifies equipment anomalies before they become failures.

We also see this as a solution for compute efficiency. The AI arms race is accelerating, requiring immense energy and infrastructure to train and deploy new models. A zero-shot solution like Nixtla’s collapses compute time required for testing and training forecasting models, thereby accelerating innovation and significantly curbing resource demand.

The Team

Nixtla was founded by Max Mergenthaler Canseco (Chief Executive Officer) and Cristian Challu (Chief Scientific Officer), longtime collaborators who met while studying engineering in Mexico. Both founders come from deeply technical backgrounds in machine learning and time series analysis, with experience spanning academic research, consulting, and real-world deployment.

Today, Nixtla is headquartered in San Francisco with a globally distributed team of 17 engineers and researchers. In addition to industry-leading customers, the company has also garnered many awards and recognitions, including G2 recognitions in Best Results, Best Est. ROI, Users Most Likely to Recommend, Easiest to Use, and more.