AI video company Higgsfield raised a $400m Series B led by DST Global at a $5.4bn post-money valuation, up from $1.3bn in January 2026. Goldman Sachs Alternatives, Valor Capital and Tribe Capital also participated. The company says it has $700m in annualised revenue, 30 million users across 200 countries, and 390 of the Fortune 500 as customers.
The arithmetic nobody ran
$5.4bn against $700m of annualised revenue is a multiple of about 7.7x. Frontier AI companies have been transacting at thirty to fifty times revenue and above. A sophisticated syndicate including DST and Goldman paying under eight times is not a bubble signal — it is a discount, and it is the most informative number in the announcement.
What the common framing gets wrong
The universal headline is the quadrupling. Two things sit underneath it. First, the $400m of new money is inside the $5.4bn post-money figure, so the pre-money valuation is $5.0bn and the genuine step-up is 3.8x, not 4.2x. Second, "annualised revenue" is a run rate — a recent short period multiplied out — not $700m earned. Third, "390 of the Fortune 500" is a logo count satisfied by a single seat at each company; it says nothing about spend. None of these figures are audited, and Higgsfield has published no release of its own about the round.
Why the discount is rational
The revenue is consumer subscription revenue in a category with negligible switching costs, where a competitor's better model is one cancellation away. Investors pricing at 7.7x are pricing churn. Founder Alex Mashrabov, formerly of Snap, supplied the other half of the case against himself: "Video is one of the most compute-intensive domains in AI. Just one minute of video is like processing 60,000 words." Quoted as a boast, it is a gross-margin warning.
A two-tier market
Set this beside the model layer, where Anthropic's last private mark of $965bn sat against a reported run rate in the tens of billions, and the shape of the market becomes visible. Capital is paying frontier multiples for model companies and consumer-software multiples for applications built on them — even when the application company is growing fast and has real revenue. The spread, not any single round, is where the risk is concentrated.