Luma AI, the frontier artificial intelligence company working on multimodal AGI, has secured a massive $900 million Series C funding round, led by HUMAIN — a PIF-backed AI company building large-scale, full-stack AI infrastructure. The raise marks one of the most significant investments into next-generation AI systems that move beyond language models and toward systems capable of understanding, simulating, and interacting with the physical world.
Strategic participants included AMD Ventures, alongside existing investors such as Andreessen Horowitz, Amplify Partners, and Matrix Partners, underscoring growing confidence in Luma AI’s mission to build “World Models” — considered the next leap forward after Large Language Models (LLMs).
Project Halo: A 2-Gigawatt AI Supercluster in Saudi Arabia
As part of the partnership, Luma AI will become a primary compute customer of HUMAIN, as it builds Project Halo, a colossal 2-gigawatt AI supercluster in Saudi Arabia — positioning it among the largest AI compute infrastructures in the world.
Announced at the U.S.–Saudi Investment Forum in Washington D.C., on the occasion of HRH Crown Prince Mohammed bin Salman’s official visit, this initiative signals Saudi Arabia’s ambition to become a global hub for AI compute, talent, and innovation.
The infrastructure will support Luma’s training of peta-scale multimodal data — video, images, audio, and text — at a scale 1,000 to 10,000 times larger than today’s frontier LLMs. These capabilities are critical to creating AI systems that can understand and simulate real-world physics, enabling applications across robotics, immersive entertainment, gaming, advertising, simulation, design, and personalized education.
From Language Models to World Models
While LLMs have transformed information processing, Luma AI is focused on what comes next — models that can perceive, reason, and act in the physical world.
“Our ambition is to build AI that can help humanity in the physical world and expand our understanding of the universe. To achieve that, we need systems that can learn from a quadrillion tokens of data — the collective digital memory of humanity,” said Amit Jain, CEO and Co-Founder, Luma AI.
These next-generation systems — called World Models — aim to learn from the entire digital visual and auditory history of humanity, creating AI that can generate experiences, simulate reality, and support intelligent decision-making in dynamic, real-world environments.
Ray3: Reasoning Video AI in Action
Luma’s latest model, Ray3, already represents a shift toward reasoning-based visual intelligence. It is the world’s first reasoning video model capable of generating physically accurate scenes, visual effects, and dynamic animations.
The technology is already being used by global studios, creative agencies, and brands and is integrated within platforms such as Adobe’s creative ecosystem.
With this new funding, Luma plans to accelerate the expansion of Ray3 into robotics simulation, industrial design, and digital twin applications, further narrowing the gap between digital imagination and physical execution.
HUMAIN Create: Sovereign AI for the Arab World
A significant element of the partnership is HUMAIN Create — a regional initiative focused on building AI models trained on Arabic language, visual culture, and regional data sets. This will allow governments, enterprises, and creators across the MENA region to adopt AI that is culturally aligned, sovereign, and context-aware.
“This is not just about funding — we’re building the entire value chain that makes next-generation AI possible,” said Tareq Amin, CEO of HUMAIN.
“Together with Luma AI and our 2-gigawatt supercluster, we are creating a new global benchmark — where capital, compute, and capability converge.”
A Trillion-Dollar Opportunity Unfolding
By combining U.S.-based frontier AI innovation with Middle Eastern compute infrastructure and sovereign AI ambitions, the Luma-HUMAIN partnership is shaping up as a potential trillion-dollar opportunity spanning:
Entertainment and content creation
Marketing and brand experiences
Robotics and automation
Simulation and digital twins
Education and knowledge modeling
Smart cities and infrastructure
It also highlights the geoeconomic repositioning of AI infrastructure, where nations are now competing not just for talent, but for compute dominance — the critical fuel for next-generation artificial intelligence.
