Higgsfield $400M Series B: AI Video Hits $5.4B

The AI video market, written off by some as a niche of viral demos, just minted another near-unicorn on the strength of real revenue. Higgsfield, a two-year-old startup that builds AI video and image tools for creators, brands, and studios, announced a $400 million Series B at a $5.4 billion valuation on August 17, 2026. The round was led by DST Global, and Higgsfield said its annualized revenue has reached $700 million this month — up from roughly $20 million a year ago.

The valuation marks a roughly four-fold jump from the $1.3 billion Higgsfield commanded earlier this year, when it raised $80 million. The pace, and the revenue behind it, make the round one of the loudest signals yet that generative video has crossed from novelty into a business with paying customers at scale.

The round

DST Global led the financing, with a long list of participants that blends growth investors and strategic backers: Growth Equity at Goldman Sachs Alternatives, Intel Capital, Tribe Capital, Smash Capital, Fifth Wall, Valor Capital, Liberty Global Tech Ventures, Mirae Asset Capital, and NTT DOCOMO Ventures. The mix of a marquee crossover fund (DST), a Wall Street growth arm (Goldman), and a chipmaker’s venture unit (Intel) reflects how many corners of the market now want exposure to the applications layer of AI rather than only the models underneath it.

Higgsfield was co-founded and is led by Alex Mashrabov, who previously founded AI Factory — the computer-vision company whose technology powered Snapchat’s face filters after Snap acquired it. “Every business needs visual content, but creating it at the quality, speed and scale companies demand remains complex and expensive,” Mashrabov said in announcing the round. “The next wave of value will be created by the applications that put this technology to work, and Higgsfield is leading that shift in visual media.”

The numbers behind the valuation

What distinguishes this raise from the froth of earlier generative-AI rounds is that the multiple is attached to revenue, not just to a demo reel. Higgsfield’s climb from $20 million to $700 million in annualized revenue over twelve months is the kind of curve investors will pay a premium to own. At a $5.4 billion valuation against $700 million of annualized revenue, the company is priced at roughly 8 times revenue — aggressive by traditional software standards, but modest next to the multiples the AI-model labs command.

The usage figures the company disclosed are the engine of that revenue. Following the rollout of its Supercomputer product in May 2026, users of Higgsfield’s agentic products grew 42-fold in three months, and the platform now processes more than 20 million content generations per month. Higgsfield says it powers visual production for 390 of the Fortune 500 — the enterprise footprint that turns consumer-scale usage into durable, contracted revenue.

Those metrics point to a specific bet: that the winning products in generative media are not the raw models but the agentic applications wrapped around them — tools that take a brief and orchestrate generation, editing, and iteration end to end, rather than handing a user a single prompt box. Higgsfield’s growth since the Supercomputer launch is its evidence that creators and brands will pay for that orchestration layer.

A crowded, capital-hungry field

Higgsfield is competing in a market that has attracted some of the best-funded efforts in AI. Chinese labs have pushed hard on video generation, from ByteDance’s Seedance 2.5 model to Alibaba’s open-source Wan-Animate 2 character-animation system, while Western players include Google’s Veo line, OpenAI’s Sora, and Runway. The open-weight frontier is moving quickly too, with image and multimodal systems like Black Forest Labs’ Flux 3 lowering the cost of the underlying generation.

That competition cuts two ways. It validates the category — the biggest labs in the world are pouring compute into video — but it also means Higgsfield does not control the models it depends on, and the price of raw generation keeps falling as open-weight alternatives improve. The company’s answer is to compete on product, workflow, and enterprise distribution rather than on owning a frontier model. The $400 million now on its balance sheet is largely a bet that it can out-execute on that applications layer faster than the model providers can move up the stack into finished creative tools.

For founders and investors watching the round, it is also a data point about how startup funding rounds are being priced in the current AI cycle: revenue growth of this steepness, attached to a clear enterprise wedge, is commanding valuations that would have looked implausible a year ago.

What it means

Higgsfield’s raise is a marker that AI video has an economic model, not just a technical one — and that the money is flowing to the application layer as aggressively as it once flowed to the models.

Who wins. Higgsfield wins the capital and the credibility to lock in enterprise customers and outspend smaller rivals on product. Its investors win exposure to one of the fastest revenue curves in applied AI. Creators and brands win cheaper, faster visual production — and, for now, a well-funded vendor unlikely to disappear. Intel’s venture arm, in particular, gains a strategic window into a compute-hungry workload.

Who’s exposed. The risk is the same one that shadows every application-layer AI company: dependence on models it does not own, in a market where the cost of raw generation is falling and the largest labs are building competing end-user tools. An 8x revenue multiple assumes the growth curve holds and that Higgsfield’s product moat is real rather than a temporary lead. If the frontier labs ship polished creative suites, or if open-weight video closes the quality gap, the applications layer gets squeezed from both directions.

What to watch next. First, retention and gross margins — whether the 20-million-generations-a-month usage translates into sticky, profitable enterprise contracts or churns as novelty fades. Second, how much of the $400 million goes to compute versus growth, a tell about how commoditized the underlying generation has become. Third, the competitive response from Google, OpenAI, and the Chinese video labs, any of which could reset customer expectations overnight. And fourth, whether this round pulls more capital into the applications layer broadly — because if Higgsfield’s curve is repeatable, it will not be the last AI-media company to quadruple its valuation in a single year.