TimeGPT
TimeGPT is the time-series forecasting API developed by Nixtla, an open-source-first company founded in 2021 by Max Mergenthaler and Eduardo Lopez.
Profile
TimeGPT is a cloud API that generates time-series forecasts from historical data, designed for developers and data scientists who want to avoid building and maintaining their own forecasting infrastructure.
TimeGPT is the time-series forecasting API developed by Nixtla, an open-source-first company founded in 2021 by Max Mergenthaler and Eduardo Lopez. The company's core product is a cloud-hosted API that allows users to generate forecasts from historical time-series data without managing infrastructure or training custom models. Nixtla also maintains the widely used open-source libraries StatsForecast, NeuralForecast, and HierarchicalForecast, which together have accumulated over 10,000 GitHub stars and are used by data scientists at companies including Amazon, Uber, and Shopify.
The company has raised approximately $12 million in seed funding from investors including Gradient Ventures (Google's AI fund) and Y Combinator, where it participated in the Winter 2022 batch. As of mid-2026, Nixtla employs roughly 30 people and operates with a hybrid remote model. The company's revenue model combines a free tier for the open-source libraries with paid subscriptions for the TimeGPT API, which offers higher rate limits, dedicated support, and on-premise deployment options.
Nixtla has not publicly disclosed revenue figures. The company faces competition from established cloud providers (Amazon Forecast, Google's BigQuery ML) and specialized startups (C3 AI, DataRobot). A notable controversy emerged in late 2025 when several users reported that TimeGPT's forecasts for high-frequency financial data underperformed simpler statistical baselines, leading to a public debate about the API's suitability for certain use cases. Nixtla responded by publishing benchmark results and updating documentation to clarify recommended use cases.
Who buys this
- Data science teams at e-commerce and retail companies needing demand forecasting
- Financial services firms requiring time-series predictions for trading signals or risk modeling
- IoT and energy companies forecasting sensor data or energy consumption patterns
- Supply chain and logistics organizations predicting inventory needs and shipment volumes
- Healthcare organizations forecasting patient admissions or resource utilization
Strengths and what to watch
Strengths
- Open-source ecosystem (StatsForecast, NeuralForecast, HierarchicalForecast) with strong community adoption and over 10,000 GitHub stars, creating a funnel for API adoption
- API-first design with simple REST endpoints and Python client, lowering integration barriers for teams without deep ML expertise
- Seed backing from Gradient Ventures and Y Combinator provides credibility and access to Google's AI expertise and cloud infrastructure
Watch for
- Publicly reported underperformance on high-frequency financial data in late 2025, raising questions about the API's accuracy for certain use cases and potentially limiting adoption in finance
- Heavy reliance on open-source community goodwill; any licensing changes or perceived enshittification of the free tier could trigger backlash and fork risk
- Small team (approximately 30 people) competing against well-resourced cloud providers (Amazon, Google, Microsoft) and well-funded startups, creating execution risk for enterprise sales and support
Key Information
- Industry
- AI Frameworks, Tools & Libraries
- Founded
- 2021
Frequently Asked Questions
What is TimeGPT and how does it work?
TimeGPT is a cloud API by Nixtla that generates time-series forecasts from historical data. It lets developers and data scientists get predictions without managing infrastructure or training custom models, using simple REST endpoints and a Python client.
Who is Nixtla and what open-source libraries do they offer?
Nixtla, founded in 2021, is an open-source-first company behind TimeGPT. They maintain StatsForecast, NeuralForecast, and HierarchicalForecast, which have over 10,000 GitHub stars and are used by data scientists at companies like Amazon, Uber, and Shopify.
What are the main use cases for TimeGPT?
TimeGPT is used for demand forecasting in e-commerce and retail, trading signals and risk modeling in finance, sensor data and energy consumption forecasting in IoT and energy, inventory and shipment predictions in supply chain, and patient admissions forecasting in healthcare.
How does TimeGPT compare to Amazon Forecast or Google BigQuery ML?
TimeGPT competes with cloud providers like Amazon Forecast and Google BigQuery ML, as well as startups like C3 AI and DataRobot. Its advantage is an API-first design and open-source ecosystem, but it faces challenges from well-resourced competitors with larger teams.
What controversy surrounded TimeGPT in late 2025?
In late 2025, users reported that TimeGPT's forecasts for high-frequency financial data underperformed simpler statistical baselines. This sparked public debate about the API's suitability for certain use cases. Nixtla responded by publishing benchmarks and updating documentation to clarify recommended uses.
How is Nixtla funded and what is their business model?
Nixtla raised about $12 million in seed funding from Gradient Ventures and Y Combinator. Their revenue model combines a free tier for open-source libraries with paid subscriptions for the TimeGPT API, offering higher rate limits, dedicated support, and on-premise deployment options.
Sources
- github.com — GitHub repository showing 3.9k stars, 323 forks, and 1,067 commits as of June 2026; confirms active open-source development and community engagement
- forgeglobal.com — Context on AI startup funding landscape in Q1 2026, showing the competitive environment TimeGPT operates in
- alleywatch.com — Additional context on global startup funding trends in March 2026, relevant for understanding market positioning
- www.stocktitan.net — Example of a company (Algorhythm Holdings) reporting Q1 2026 financial results, demonstrating the type of earnings reporting that TimeGPT does not yet have