이미지: TechCrunch Startups
Summary
- Higgsfield raised a $400 million Series B at a $5.4 billion valuation, more than quadrupling the $1.3 billion valuation it held eight months earlier.
- The company cited $700 million in annualized revenue, 30 million users, and partnerships with 390 Fortune 500 companies as evidence.
- CEO Alex Mashrabov named the massive compute costs of video processing as a key use for the new funding.
- 이번 유치액
- 4억달러 시리즈B
- 신규 기업가치
- 54억달러
- 직전 기업가치(8개월 전)
- 13억달러
- 연환산매출
- 7억달러
- 사용자 수
- 3000만명, 200개국
- 포춘500 고객사
- 390곳
- 라운드 주도
- DST 글로벌
- 설립
- 2023년, 알렉스 마슈라보프(전 스냅 임원)
Valuation Quadruples in Eight Months
Just eight months ago, Higgsfield was valued at $1.3 billion. This week, the company rewrote that figure to $5.4 billion after closing a new $400 million Series B round. According to an announcement made via PR Newswire, the round was led by DST Global, with participation from Goldman Sachs Alternatives, Valor Capital, and Tribe Capital.
| Timing | Valuation | Multiple |
|---|---|---|
| 8 months ago (late 2025) | $1.3 billion | 24 |
| Latest round (August 2026) | $5.4 billion | 100 |
A Startup That Rents Out the Whole Studio
Higgsfield is a startup that offers entire AI-powered image and video production studios as a service. It was founded in San Francisco in 2023 by Alex Mashrabov, who previously led Snapchat's generative AI organization. The company has built its own models — Soul, which produces realistic human portraits, and DoP, which generates cinematic camera movements — while also drawing on models from OpenAI, Google, and ByteDance. Its flagship products are Cinema Studio, built for filmmakers, and Marketing Studio, aimed at advertising and marketing teams.
Last year, the company drew attention by showcasing AI-made films at Cannes and in New York. According to a Wall Street Journal report, $400,000 of one film's $500,000 production budget went toward AI compute costs. In other words, generating a single image or a single video clip is itself a cost-generating process.
Revenue and Enterprise Customers
The company highlighted three figures in its announcement: $700 million in annualized revenue, 30 million users across 200 countries, and partnerships with 390 Fortune 500 companies.
| Metric | Figure |
|---|---|
| Annualized revenue | $700 million |
| Users | 30 million across 200 countries |
| Fortune 500 partners | 390 |
Enterprise growth stands out in particular. Mashrabov said enterprise adoption of video AI will become more deeply embedded in everyday marketing and creative work.
Why Pour Money Into Compute
Beyond typical uses like hiring and product development, a substantial portion of the new funding will go toward securing compute. Mashrabov explained that "video is one of the most compute-intensive areas in AI." According to him, processing a single one-minute video requires roughly the same amount of computation as processing 60,000 words of text. Securing stable compute has become an essential cost for competing with rivals like Synthesia and Runway, he said.
Editor's Take
A startup quadrupling its valuation in eight months is rare. What this pace signals is simple: money flowing into the AI video market hasn't dried up — if anything, investors are racing to lock in the winners early. This round sits atop a broader trend of capital, once cautious after Sora's early stumbles, now flowing toward players with actual revenue like Synthesia, Runway, and Higgsfield.
The shift looks different depending on where you're standing. A year ago, AI video tools were still seen as "fun demos." Now, 390 Fortune 500 companies have built them into real workflows. That gap has less to do with model performance than with whether a company can simultaneously run compute operations, enterprise sales, and in-house model development. That's precisely why Higgsfield sticks with a strategy of blending its own models (Soul, DoP) with outside ones — relying solely on one company's models would hand over both compute leverage and quality ceilings to someone else.
The practical lesson for marketing and content teams is clear. The era of piloting AI video on a single project is already over. The question to ask now is: how many pieces will we produce per month, and what's the compute cost per piece? The fact that 80% of a Cannes-featured film's budget went to AI compute shows that adopting video AI isn't a "free experiment" — it's a new fixed cost. Leave that out of the budget, and the bill will come as a surprise later.
What happens in the coming weeks isn't hard to predict. Runway and Synthesia are likely preparing rounds of similar scale, and Nvidia and cloud providers selling compute will absorb a large share of this new funding right back. The valuation race among AI video startups is increasingly narrowing down to a single question: who can secure compute more cheaply and more reliably.
