Understand any stock in simple English
Enter a ticker to learn what the company does and how it makes money.
Reflection AI, Inc.
Reflection AI builds open-source artificial intelligence models and sells access to them, aiming to be a Western alternative to Chinese open-source labs.
Reflection AI is an artificial intelligence company that builds "open" AI models — large language models whose underlying weights are released publicly — and makes money by selling access to them and to the computing power needed to run them. Its pitch is that the most powerful open models should come from the United States and its allies rather than from Chinese labs, so it positions itself as a Western champion of the open-source approach.
The company was founded in 2024 by two former Google DeepMind researchers, Misha Laskin and Ioannis Antonoglou. It came out of stealth in early 2025 with unusually loud ambitions: to build frontier-level AI in the open, compete with the biggest closed labs, and do it with a fraction of their spending. A big early bet was that the cheapest path was to train on enormous amounts of publicly available data rather than paying for licensed content. Reflection's business has two connected halves. It develops its own models and gives them away openly, which builds attention and a community of developers. It then earns revenue by hosting those models and selling inference — the computing work of running the AI — to companies that want the capability without building it themselves. The open model is the marketing; the hosted service is the product people pay for.
The company's rise has been fast and well-funded. It raised a large early round in 2025 at a multibillion-dollar valuation, backed by investors including Nvidia, and later secured a major computing deal with Nvidia to power its training. That combination of huge capital and cheap access to chips is what lets a young startup chase labs many times its size. Whether open models can truly match the closed frontier — and whether giving the models away leaves enough room to profit — is the central question hanging over the whole business.
Saved answer · last researched 10 Nov 2026