NextFin news, On November 19, 2025, Luma AI, a U.S.-based artificial intelligence company specializing in multimodal AI, announced the closure of a $900 million Series C funding round. This round was led by HUMAIN, a Saudi Arabian PIF-backed AI firm, alongside significant participation from AMD Ventures, Andreessen Horowitz, Amplify Partners, and Matrix Partners. The announcement took place at the U.S.-Saudi Investment Forum in Washington, D.C., concurrent with the visit of Crown Prince HRH Mohammad bin Salman Al-Saud.
The funding will accelerate Luma AI’s development of multimodal artificial general intelligence (AGI) — AI systems capable of understanding and operating in the physical world through integrated multimodal data processing including video, images, audio, and text. Central to this effort is HUMAIN's commitment to build Project Halo, a 2-gigawatt AI supercluster in Saudi Arabia, designed to provide the computational power necessary for training petabyte-scale AI models far exceeding the scale of current large language models (LLMs).
Luma’s flagship technology includes Ray3, heralded as the world's first video reasoning model that can generate highly realistic video with motion and HDR effects from simple image inputs and user instructions. Their Photon image generation suite complements this by offering high-quality, cost-efficient image synthesis. Luma’s models are already integrated into commercial products, including Adobe’s Firefly design suite, marking considerable adoption in sectors such as film, advertising, and marketing.
The partnership reflects HUMAIN’s strategic approach to building an end-to-end AI value chain, combining capital, infrastructure, and technological capabilities. The collaboration also aligns with HUMAIN’s regional ambitions to foster AI sovereignty and tailored AI solutions for the Middle East and North Africa (MENA), through initiatives like HUMAIN Create, which develops culturally aligned AI models trained on Arabic and regional datasets.
From a strategic standpoint, this joint funding and infrastructure initiative advances the frontier of multimodal intelligence by pushing AI beyond text-based LLMs into the realm of 'World Models' — AI systems capable of simulating and reasoning about the real world at scale. Luma co-founder and CEO Amit Jain emphasized the vision to train AI on roughly a quadrillion tokens, akin to humanity’s collective digital memory, integrating diverse media to achieve AGI that can meaningfully assist in physical-world tasks and expand scientific understanding.
This development fits within a broader global trend where geopolitical actors and tech investors prioritize building foundational AI capabilities with localized adaptations and massive computing infrastructure. Saudi Arabia's investment via HUMAIN into Project Halo situates the Kingdom as a key player in the AI supercomputing race, leveraging its vision under Crown Prince Mohammad bin Salman’s leadership to diversify the economy and gain technological leadership.
Economically, Luma AI’s valuation has surged above $4 billion post-round, reflecting growing investor appetite and confidence in the multimodal AI segment's commercial potential, spanning entertainment, robotics, education, advertising, and simulations. The hardware-efficient models that Luma offers—achieving professional-grade image generation at under half a cent per image—highlight the growing trend of optimizing AI for cost-efficiency and scalability, crucial for mass market adoption and enterprise SaaS models.
Technically, the move toward multimodal AGI embodies a crucial evolutionary phase in AI research: integrating visual, audio, and textual information into coherent world models that can reason, simulate, and interact dynamically. Luma's development of video reasoning via Ray3 and 3D representation techniques such as NeRF (Neural Radiance Fields) demonstrates sophistication in generative AI that enables realistic dynamic content creation, crucial for next-gen digital experiences and robotics.
Looking forward, the successful deployment of Project Halo’s supercluster coupled with the collaboration between HUMAIN and Luma may catalyze significant breakthroughs in real-time AI inference at scale. This infrastructure enables training on petabytes of multimodal data—a scale 1,000 to 10,000 times larger than current frontier models—possibly setting new industry benchmarks and accelerating the timeline toward functional AGI systems.
Moreover, the geopolitical dimension cannot be overlooked. By anchoring this partnership in Saudi Arabia with its sovereign compute infrastructure and regional AI initiatives, HUMAIN is championing a model of AI development that balances global innovation with local cultural relevancy and regulatory alignment. This diversification strategy could inspire other nations to invest similarly, potentially reshaping global AI innovation hubs beyond Silicon Valley.
The $900 million investment thus represents more than capital infusion—it symbolizes the convergence of technological ambition, strategic geopolitical investment, and the practical business model of AI as a service. For stakeholders, including policymakers, tech investors, and industry players, the evolution of multimodal AI advancing toward AGI is one to watch closely, as it promises to redefine digital interaction, creative industries, and the automation of complex physical-world tasks.
According to The Information, this funding round underscores a competitive landscape where startups with multimodal AGI capabilities are securing large, supercluster-enabled investments to maintain innovation velocity. It also reflects the increasing role of cross-border strategic investments aligned with national digital transformation agendas under current global political dynamics.
Explore more exclusive insights at nextfin.ai.