Key Takeaways
- AIX Ventures raised $28.0M (Seed) from AIX Ventures, NVIDIA Ventures (NVentures), Threshold Ventures.
- Sector: Artificial Intelligence (AI).
Analysis
Moonlake AI, an artificial intelligence research lab, has announced its emergence from stealth with a remarkable $28M in seed funding.
The funding was led by AIX Ventures, Threshold, NVIDIA Ventures, and was further supported by a host of leading AI researchers and founders. The lab, founded by Stanford AI Lab and NVIDIA researchers Fan-Yun Sun and Sharon Lee, is working on artificial intelligence models that can create real-time interactive content. This includes simulations and games and these models can enable anyone to create interactive worlds in just minutes.
Moonlake AI's models aim to lower the barrier to creative coding and open interactive world-building to developers, designers, and fans without the need for programming or domain knowledge. The company's innovative approach has attracted excitement from industry leaders such as YouTube founder Steve Chen who said he is personally excited by the potential Moonlake AI is working to unlock, comparing its platform's potential to that of YouTube in terms of interactive content creation.
Moonlake AI boasts the first reasoning model that allows the creation of immersive, interactive environments within minutes. Users and developers can simply describe their desired outcomes, whether it's for games or robotics simulations, and the system can compile it into a functional 2D or 3D scene that can be interacted with immediately. The platform uses multi-modal reasoning for spatial layout, tool use over modern engines to build assets and scenes, program synthesis for logic, a simulation layer for agents, and a real-time diffusion model to reskin the world. This makes it equally useful for game prototyping as it is for creating simulations to train reinforcement learning and embodied AI agents.
Creating interactive worlds can be a slow and expensive process, often requiring large teams and budgets, and even prototyping can take years of work. Moonlake AI is aiming to solve these issues with its innovative models. Initial use cases for Moonlake's platform include user-generated interactive worlds, interactive world creation for video game developers, studios, and IP holders, robotics and embodied-AI simulations for training and evaluation, creation of 3D environments for film and animation companies, and immersive learning experiences for teachers.
The founders of Moonlake AI, Fan-Yun Sun and Sharon Lee, have impressive backgrounds in AI research. Sun earned his PhD in Computer Science at Stanford whilst also working as a researcher at NVIDIA, leading efforts to generate large-scale 3D worlds for training AI agents. Lee was pursuing a PhD in generative AI and computer graphics, combining diffusion models with 3D engines. She also develops foundation models trained to reason about space, enabling richer interactivity and control in virtual worlds.
Moonlake AI is currently in private preview, with demos and additional information available on request.