Video-generation startup Luma AI announced that it has secured $900 million in a new funding round, led by Humain, an artificial intelligence company backed by Saudi Arabia’s Public Investment Fund.
The round also saw participation from Advanced Micro Devices’ venture arm and existing backers Andreessen Horowitz, Amplify Partners, and Matrix Partners, with the announcement made at the U.S.-Saudi Investment Forum on Wednesday.
The fresh investment pushes Luma’s valuation to over $4 billion.
Luma builds multimodal “world models” capable of learning from text, video, audio, and images to replicate real-world environments. CEO Amit Jain explained that these models go beyond text-only large language models and are more effective in “helping in the real, physical world.”
“With this funding, we plan to scale and accelerate our efforts in training and then deploying these world models today,” Jain said.
In September, Luma AI introduced Ray3, described as the first reasoning video model able to interpret prompts and generate videos, images, and audio. Jain noted that Ray3 currently performs better than OpenAI’s Sora 2 and is comparable to Google’s Veo 3.
Humain, launched in May, aims to deliver full-stack AI solutions to strengthen Saudi Arabia’s presence as a global AI leader. The company is headed by industry expert Tareq Amin, formerly CEO of Rakuten Mobile and previously leading Aramco Digital.
During the forum, Humain also revealed multiple collaborations, including standalone data center projects with AMD, Cisco, GlobalAI (built on Nvidia infrastructure), and Elon Musk’s xAI.
Luma AI and Humain will additionally collaborate to construct a 2-gigawatt AI supercluster in Saudi Arabia, called Project Halo, which Jain said will be among the world’s largest GPU deployments.
Across the tech industry, major companies are investing heavily in supercomputing resources to train advanced AI systems. In July, Meta announced its 1-gigawatt Prometheus supercluster, while Microsoft recently rolled out the first supercomputing cluster built with the Nvidia GB300 NVL72 platform.
“Our investment in Luma AI, combined with HUMAIN’s 2GW supercluster, positions us to train, deploy, and scale multimodal intelligence at a frontier level,” Amin said in a release. “This partnership sets a new benchmark for how capital, compute, and capability come together.”
Their partnership also includes Humain Create, an initiative dedicated to developing sovereign AI models trained on Arabic and regional datasets. In addition to working on what is expected to be the world’s first Arabic video model, Jain said Luma AI’s technologies will be applied across businesses in the Middle East.
He pointed out that because most AI models scrape data from the open internet, regions outside the U.S. and Asia tend to have limited representation in AI-generated content.
“It’s really important that we bring these cultures, their identities, their representation—visual and behavioral and everything—to our model,” Jain said.
AI content-generation tools have faced pushback from entertainment companies over copyright issues. Luma’s text-to-video platform Dream Machine encountered claims of IP copying earlier this year, but Jain emphasized that the company has strong protective systems in place.
“Even if you really try to trick it, we are constantly improving it,” he said. “We have built very robust systems that are actually using models we trained to detect them.”
