Reflection AI, a rapidly growing AI startup founded last year by former DeepMind researchers, has raised $2 billion in funding at an $8 billion valuation. This represents a remarkable 15-fold increase from its $545 million valuation just seven months ago. Originally focused on autonomous coding agents, the company is now positioning itself as both an open-source alternative to closed frontier labs like OpenAI and Anthropic and a Western counterpart to China’s DeepSeek.
Founders with Deep AI Expertise
Reflection AI was launched in March 2024 by Misha Laskin and Ioannis Antonoglou. Laskin led reward modeling for DeepMind’s Gemini project, while Antonoglou co-created AlphaGo, the AI system that defeated the world champion in Go in 2016. Their experience building advanced AI systems forms the foundation of Reflection AI’s vision. The startup believes that top AI talent can develop frontier models outside traditional tech giants.
The team now consists of approximately 60 experts, mostly AI researchers and engineers specializing in infrastructure, data training, and algorithm development. Many have previously worked at DeepMind and OpenAI, bringing significant expertise to the company’s ambitious plans.
Building Open Frontier AI
With the new funding, Reflection AI has developed a state-of-the-art training infrastructure for large-scale Mixture-of-Experts models, a leading architecture powering today’s most advanced language models. The company plans to release its first frontier language model next year, trained on tens of trillions of tokens.
According to Laskin, the company has achieved what was once possible only within the world’s top labs. Reflection AI first demonstrated its approach in autonomous coding, and it now aims to expand into general agentic reasoning.
Responding to Global AI Competition
Laskin emphasized the urgency of competing with Chinese AI firms. He mentioned DeepSeek, Qwen, and other Chinese models as a wake-up call for the U.S. “If we do not act, the global standard of intelligence will be set by others,” Laskin said. He noted that enterprises and governments often avoid Chinese models due to legal and strategic concerns, which creates both a challenge and an opportunity for Reflection AI.
American technologists have welcomed Reflection AI’s mission. David Sacks, the White House AI and Crypto Czar, praised the company for advancing open-source AI. He highlighted the advantages of open models in terms of cost, control, and customization. Clem Delangue, CEO of Hugging Face, emphasized the importance of sharing AI models and datasets quickly to support innovation.
The Open Model Philosophy
Reflection AI’s approach to being “open” centers on public access to model weights, which determine how AI systems function, while keeping training datasets and pipelines proprietary. Laskin explained that providing model weights allows researchers and developers to experiment and innovate.
The company’s business model reflects this balance. Researchers can freely use the models, while revenue comes from large enterprises building products on top of Reflection AI’s models and governments developing sovereign AI systems. Laskin noted that enterprises prefer open models because they can run them on their own infrastructure, control costs, and customize solutions for specific workloads.
Future Plans and Funding
Reflection AI has not yet released its first model, which will initially focus on text-based tasks, with multimodal capabilities planned in the future. The latest funding will be used to acquire the compute resources necessary to train these models, with the first release expected early next year.
Investors in this funding round include Nvidia, Disruptive, DST, 1789, B Capital, Lightspeed, GIC, Eric Yuan, Eric Schmidt, Citi, Sequoia, CRV, and others.
Conclusion
Reflection AI is positioning itself as America’s open frontier AI lab, aiming to ensure that the development of frontier artificial intelligence remains open, collaborative, and aligned with Western standards. With its deep technical expertise, top-tier talent, and strong financial backing, the company is poised to compete with global AI leaders while supporting innovation through accessible model weights.
