Luma AI has raised $900 million in Series C funding to accelerate development of multimodal artificial general intelligence, the company announced Wednesday at the U.S.–Saudi Investment Forum in Washington, D.C.
The round, which includes SAFEs and two tranches of capital, was led by HUMAIN, the AI company backed by Saudi Arabia’s Public Investment Fund, with participation from AMD, Amazon, Andreessen Horowitz, Amplify Partners and Matrix Partners.
To support model training and deployment, Luma will become a customer of HUMAIN as it builds Project Halo, a 2-gigawatt AI compute supercluster in Saudi Arabia. The infrastructure is among the world’s largest planned buildouts and is designed to support training at peta-scale levels, allowing AI models to learn from significantly larger visual, audio and language datasets than those used in today’s leading LLMs.
Luma is developing World Models, a class of multimodal systems designed to simulate and operate within the physical world, representing what it describes as the next step beyond language models. CEO and co-founder Amit Jain said the company aims to train using what he equates to the digital memory of humanity, arguing that real-world simulation will require integrating data at unprecedented scale and complexity.
The company’s flagship model, Ray3, is already used commercially across studios, brands and advertising agencies, and has been deployed via Adobe’s global products. The new capital will support expansion of applications beyond entertainment and marketing towards simulation, design and robotics.
The partnership also extends to HUMAIN Create, an initiative to build culturally aligned AI models trained on Arabic and regional data. The programme aims to deliver sovereign AI solutions for enterprises and governments across the Middle East and North Africa.
Luma said the combined financing and compute access will accelerate its trajectory toward multimodal AGI capable of supporting real-world human tasks at scale.
