In a funding round that confirms investors’ enthusiastic appetite for generative AI tools, Higgsfield, a San Francisco-based startup specializing in AI-driven video production, secured an $80 million Series A extension and boosted its valuation to more than $1.3 billion. Founded by Alex Mashrabov, a former Snap executive who led the social media giant’s generative AI efforts, the company claims to have generated $200 million in annual revenue just nine months after launching. The expansion builds on an initial Series A of $50 million, bringing total funding to $130 million.
Higgsfield’s rapid rise comes amid an exploding demand for AI video generation, where tools can generate high-quality clips from text prompts, transforming marketing, content creation, and social media. The startup, which emerged from stealth in April 2025, now boasts 15 million users and partnerships with major brands. Investors including Accel led the latest tranche on metrics comparable to existing players in the sector.
From Snap to Startup Helm
Mashrabov, who worked at Snap Inc. for more than a decade and was promoted to head of generative AI, saw an opportunity in video compositing during his tenure there. At Snap, he oversaw features such as AI-generated lenses and avatars, but the constraints of being a publicly traded company limited more daring experiments. “We were building for hundreds of millions of users, and now we can move faster,” Mashrabov told TechCrunch.
Mashrabov, who launched Higgsfield in early 2025, assembled a team of former Snap engineers and AI researchers from OpenAI and Google DeepMind. The platform’s core innovation lies in a diffusion model that is fine-tuned for cinematic video, allowing output of up to 30 seconds at 1080p resolution. Early adopters include advertising agencies and TikTok creators, who appreciate its realism and customizability.
revenue rocket fuel
Higgsfield’s $200 million run rate comes from a freemium model. The free basic generation attracts users, while the pro tier, which costs between $20 and $100 per month, offers unlimited credits, 4K export, and branded kits. Enterprise deals average $500,000 per year and are targeted at Fortune 500 marketers with scalable video production needs. “The entire creative team will be replaced,” a company spokesperson said, according to Reuters.
Competition from OpenAI’s Sora, Runway, and Pika Labs is fierce, but Higgsfield differentiates itself with “character consistency” (ensuring people and objects are consistent across clips) and integration with Adobe Premiere. Boosted by celebrity deepfakes and viral X posts showcasing product demos, user growth reached 15 million by Q4 2025.
Investor enthusiasm for AI video
Margaret Francis, Accel partner, highlighted Higgsfield’s traction: “Revenue per user is five times the industry average and churn is less than 3%.” The round values the company at $1.3 billion post-money, implying 10x run-rate revenue. Other backers include Thrive Capital and former GitHub CEO Nat Friedman. TechCrunch’s post about X amplified this news, garnering thousands of views and sparking a discussion about AI disruption.
This funding reflects broader momentum in VC. AI video startups raised more than $2 billion in 2025, according to PitchBook data. Higgsfield’s path reflects a leap forward in Mistral AI’s reputation, betting on multimodal models as the next frontier for post-text and images.
The technical edge under the hood
At its core, Higgsfield employs a proprietary “Video Latent Diffusion” architecture trained on 100 million hours of licensed footage. Unlike competitors that rely on public datasets that are prone to bias, Higgsfield partners with Hollywood studios to obtain clean data. Mashrabov emphasized safety: “We watermark all output and block harmful prompts during inference.”
The model’s efficiency in generating a 10 second clip in less than 30 seconds on the H100 GPU comes from a custom quantization technique. As reported by BitcoinWorld , the upcoming release promises audio sync and multi-shot narrative, positioning Higgsfield as a Hollywood pilot that is already being discussed.
Monetization mastery
Corporate successes include Coca-Cola and Nike, who have used Higgsfield for personalized ad variations. One campaign generated 10,000 unique 15-second spots, reducing costs by 80%. “The run rate has essentially reached $200 million. Marketing costs are zero,” Mashrabov said in an interview with Tech Startups.
Challenges still exist. High computing costs consume $50 million a year on Nvidia chips, which is offset by revenue but squeezes margins. Regulatory scrutiny is looming, with EU AI laws classifying video generation as “high risk.”
road to domination
Looking to the future, Higgsfield is eyeing a Series B at a $3 billion valuation by mid-2026, according to people close to the company. There are rumors of acquisitions of smaller tools like Luma AI. Mashrabov envisions “AI as a new camera” that will democratize film production. As detailed by Forbes, brands now consider this their core infrastructure, generating thousands of assets every week.
Risks include commoditization of models due to the emergence of open source alternatives. Still, Higgsfield leads the industry with a 70% gross margin and 300% quarter-over-quarter growth. X sentiment is positive, with creators praising its breakthrough in Photoshop for video.
Broader market ripples
Higgsfield’s success shows that video AI is moving from being just a tool to being a business imperative. Traditional VFX companies love Industrial Light & Magic integration testing and fear obsolescence. Investors flooded in, with similar rounds taking place at Runway ($300 million) and Kling AI ($1 billion).
For insiders, the real story is defense. Higgsfield’s moat combines data, distribution through the Snap alumni network, and constant iteration. As Mashrabov told NewsBytes, “We’re not just building tools. We’re redefining creation.”
