Luma AI Raises $900 Million In Series C Funding Round

Listen to this article

Luma AI has raised $900 million in a Series C round, marking a significant escalation from its prior $90 million funding. The round is led by HUMAIN, with the company’s post money valuation reaching approximately $4 billion. Funds will be used primarily to accelerate the development of ‘World Models’ for multimodal AGI, targeting applications in robotics, entertainment, advertising, gaming, and education.

Luma AI, a Palo Alto-based startup founded in 2021, specializes in generative AI for video, image, and 3D content creation. Its flagship products, like the Dream Machine platform and Ray3 model, enable users to produce cinematic videos from text or image prompts, with capabilities for realistic motion, 16-bit HDR rendering, and visual annotations. This latest funding round positions Luma as a leader in multimodal AI, extending beyond text based large language models (LLMs) to systems that process video, audio, and images for real world simulation.

The $900 million Series C dwarfs Luma’s previous raises, bringing total funding to over $1 billion. Led by HUMAIN, launched in May 2025 to deliver full stack AI solutions, the round values Luma at around $4 billion, up from an estimated $3.2 billion in August 2025 discussions. CEO Amit Jain emphasized the capital intensive nature of scaling multimodal models, noting that early stage tech development had to precede such large infusions.

Key participants include:

• HUMAIN: Provides not just capital but compute access via Project Halo.
• AMD Ventures: New strategic investor.
• Existing Investors: Andreessen Horowitz (a16z), Amplify Partners, and Matrix Partners, signaling continued confidence from Silicon Valley VCs.

This round follows a $90 million Series C in December 2024 (with Amazon and AMD involvement) and a $43 million Series B in January 2024, reflecting rapid growth from $70 million total funding pre 2024.

A core element is Luma’s role as the primary customer for Project Halo, a 2-gigawatt supercluster set to become one of the world’s largest AI compute facilities by 2030. This infrastructure, developed in collaboration with AMD, Cisco, and HUMAIN, will enable training on “peta-scale multimodal data”, 1,000 to 10,000 times more than typical LLMs, using a quadrillion tokens from humanity’s digital footprint.

The partnership extends to product development in “HUMAIN Create,” focusing on:

• Model Training: Scaling World Models for multimodal AGI that simulates physical reality.
• Sector Expansion: From current deployments in entertainment (e.g., Adobe integrations) and advertising to robotics, design, gaming, and personalized education.
• Global Roadmap: Joint efforts in interactive storytelling and frontier capabilities.

Jain highlighted the need for “immense compute” to advance beyond LLMs, with Ray3 already benchmarking higher than competitors like OpenAI’s Sora 2 and on par with Google’s Veo 3.

Co-founded by Amit Jain (ex Apple Vision Pro) and Alex Yu (ex UC Berkeley AI researcher), Luma AI has grown to 51-100 employees, with plans to double its team post 2024 funding. Its tech stack includes:

• Ray3: A reasoning video model for generating clips with motion effects and annotations.
• Photon: Image generation tools.
• Genie: 3D object creation from text in seconds.
• Dream Machine: Consumer app launched in November 2024, generating $8 million in annualized revenue by December 2024.

Luma’s innovations, like DDIM for faster image generation, target creative industries, with millions of users via its 2021 smartphone app.

This raise occurs amid an AI funding boom, with Anthropic’s $13 billion Series F (valuing it at $183 billion) and Cohere’s $7 billion valuation in 2025.
For Luma, the deal mitigates compute shortages, but risks include escalating costs from competition (e.g., OpenAI, Google) potentially compressing margins. Positively, it accelerates AGI progress, with Ray3’s commercial traction (e.g., in studios and brands) validating scalability.

This infusion cements Luma’s trajectory toward multimodal leadership, blending innovation with infrastructure to redefine AI’s physical world applications.

Recommended: AI-Native Platform Psynth Automates The Psychological Report Creation Workflow, From Intake To Final Report

Luma AI’s $900 million Series C funding round, represents a pivotal moment in the generative AI landscape, underscoring the intensifying global race for multimodal intelligence and compute dominance. Founded in July 2021 by Amit Jain and Alex Yu in Palo Alto, California, Luma AI has evolved from a niche 3D modeling tool into a comprehensive platform for creating cinematic videos, images, and simulations from text, images, or audio inputs. With a team of 51-100 engineers, researchers, and designers, poised for expansion, the company now commands a $4 billion valuation, fueled by strategic alliances that extend far beyond traditional Silicon Valley funding.

Luma’s fundraising journey illustrates explosive growth in a capital hungry sector. The startup’s initial seed round in 2021 supported its smartphone app launch, which amassed over two million users by early 2024. Subsequent raises included:

• Series A (2023): Undisclosed amount, focusing on foundational generative AI research.
• Series B (January 2024): $43 million led by Andreessen Horowitz (a16z), valuing Luma at $200-300 million and funding GPU cluster development for model training.
• Series C (December 2024): $90 million with participation from Amazon, AMD, and others, totaling $157 million raised and enabling the Dream Machine app’s rollout.
• Current Series C (November 2025): $900 million, a 10x leap from the prior C round, bringing cumulative funding above $1 billion.

This progression reflects Luma’s maturation: early rounds prioritized core tech like DDIM (a diffusion model acceleration technique co-invented by team members), while recent capital targets infrastructure scale ambitions. As of August 2025, Luma was seeking at least $1.1 billion at a $3.2 billion valuation, with Middle East investors in talks, foreshadowing HUMAIN’s lead role.

HUMAIN, a May 2025 PIF initiative, is led by a philosophy of “building the full value chain” for AI, per CEO Tareq Amin. As Luma’s anchor customer for Project Halo, a 2-gigawatt supercluster rivaling global leaders, HUMAIN provides not just $900 million but unprecedented compute scale. This 1GW-phase rollout by 2030, in JV with AMD and Cisco, addresses Luma’s “immense compute” needs for quadrillion token training.

New entrant AMD Ventures adds hardware expertise, complementing prior AMD ties. Existing backers, a16z (deep AI portfolio), Amplify (early stage focus), and Matrix (multimodal bets), reinvest with 16 total investors including one angel. This mix signals broad ecosystem support, from VCs to sovereign funds.

Luma’s edge lies in “World Models”, foundational AI simulating reality via multimodal data, surpassing LLMs’ text limitations. Key offerings include:

• Ray3 (September 2025): World’s first reasoning video model, generating 16-bit HDR clips with motion and annotations; benchmarks above OpenAI’s Sora 2, equivalent to Google’s Veo 3. Deployed in Adobe products for studios and agencies.
• Photon Series: High fidelity image generators for design workflows.
• Genie: Rapid 3D asset creation, under 10 seconds from text.
• Dream Machine: Consumer facing app (launched November 2024), blending media for $8 million annualized revenue by December 2024.

These tools empower sectors like Hollywood (e.g., embedded in production pipelines) and advertising, with expansions into robotics simulation and personalized education. The funding accelerates R&D for AGI that “helps in the real, physical world,” per Jain, leveraging peta scale data for nuanced understanding.

The AI boom sparks bubble concerns: 2025’s mega rounds (e.g., Anthropic at $183 billion) amid high compute costs could strain economics if efficiency lags. Luma mitigates via HUMAIN partnerships, but competition from OpenAI and Google risks margin erosion. Positively, Ray3’s traction and revenue growth affirm viability, with the supercluster unlocking robotics/entertainment frontiers.

Access to Project Halo catapults Luma toward AGI leadership, potentially dominating $100 billion+ markets in generative media and simulation. Partnerships like HUMAIN Create could spawn interactive gaming/education tools, while Adobe integrations scale adoption.

Please email us your feedback and news tips at hello(at)superbcrew.com