Business of GTM

Higgsfield Claims a $1B Run Rate. Three Numbers a Buyer Needs Are Missing.

Higgsfield says it has passed a $1B annualized revenue run rate, calculated as four weeks of revenue times 13. It published no partner list, monthly churn rate, or customer count.

September 25, 2026

A Higgsfield film clapperboard in profile with every field on its slate left blank and red-striped clapper sticks.
A Higgsfield clapperboard with every field blank, like the numbers missing behind its $1B run rate claim.

Higgsfield, the generative video startup, told Bloomberg this week that its annualized revenue run rate has passed $1 billion. CEO Alex Mashrabov posted the milestone on LinkedIn the same day. Revenue from business customers is up tenfold since June. The company calculates the run rate by multiplying its last four weeks of revenue by 13.

AdExchanger's daily roundup ran the news with a warning, calling the metric "not a rock-solid" one. It noted that annualizing four weeks of holiday-season ad revenue would inflate the number. When the trade press is more careful with a number than the company announcing it, buyers should pay attention.

What the number leaves out

The $1B claim came without three numbers any serious buyer or investor would ask for before writing a check.

First, the partner list. Higgsfield sells to brands through advertising and creative agencies. Who those partners are and how long they stay is what separates a $1B run rate from a $1B business. AdExchanger's headline framed the limit as a partnership problem, and Higgsfield hasn't named its partners.

Second, monthly churn. When Higgsfield says business-customer revenue is up 10x since June, that's gross growth with no figure for the customers leaving. Ad-industry AI tools turn over quickly because creative teams experiment under real budget pressure. A 10x jump from June to September could hide 30% monthly churn if June's base was small and September includes early holiday spending.

Third, the customer count. Ten large accounts and 10,000 small ones look the same in run-rate math and nothing alike in an enterprise pitch. Without the count, you can't tell whether Higgsfield is embedded in brand teams or selling to lots of Shopify merchants. Both are legitimate businesses, but only the first matches the story the announcement tells.

The operator take that stopped being contrarian

On founder Twitter, the difference between ARR and run rate stopped sounding pedantic around the time of the 11x restatement. "ARR equals Annual Recurring revenue. Run rate equals revenue, any source," Sam Broner wrote in March 2025. Without contracts that reliably renew, a company doesn't have ARR.

That difference applies to most AI-vendor headlines this year. "When startups report ARR, I almost always assume it's annualized run rate," wrote Marina Temkin after the 11x restatement. Bloomberg's wording on Higgsfield is precise, using "annualized revenue run rate." The LinkedIn celebration and the lighter coverage use "ARR" freely, and that's where retail investors and enterprise CIOs start reading.

Margin is the other open question. Higgsfield builds its own product layer, but a meaningful share of its costs likely goes to the model providers underneath it. A business that mostly resells AI compute can reach $1B in revenue the way AWS resellers did. The cash is real in both directions, but the gross margins don't earn a software multiple. The announcement didn't include a margin figure either.

The counter-case worth taking seriously

The best case for Higgsfield is strong. The ad industry is a real entry point, and the CEO's disclosed trajectory is faster than any earlier generative video company's. The run rate went from $200M at the end of 2025 to $500M in June, $700M in August and $1B in September. The $400M Series B at a $5.4 billion valuation from DST Global also means at least one sophisticated growth investor did the diligence and set a price. That counts for something, but it doesn't settle the question.

DST wrote the check in August, and the $1B milestone came about six weeks later. DST's diligence may have covered partners, churn, and customer count. None of it reached the public claim, and outsiders can't price what private diligence saw. That gap between what the investor knows and what the market can check is the risk the announcement created. AI-vendor announcements will keep carrying it until disclosure standards tighten.

What this means for GTM leaders

Two moves, both cheap.

First, when a vendor pitches you on its run rate this quarter, ask three questions on the first call. How many customers does it have? What is its monthly churn? Which partners drive the number? A vendor that won't answer is telling you something. One that will is giving your CFO the facts needed before signing an enterprise license agreement. Vendors with the answers ready are the ones worth building a two-year integration around.

Second, if your buyers read TechCrunch and Bloomberg, assume they saw the $1B headline. They're probably asking internally whether AI-vendor revenue is real or reseller math. Address that question directly in your pitch. Your pipeline numbers will hold up, but a run-rate framing might not.

My bet is that AI-vendor announcements use a stricter definition of ARR by Q1 2027. Either investors push back on the next round of restatements, or one high-profile miss reprices the category. The trade press got there this week, and IPO prospectuses will follow.

Ten large accounts and 10,000 small ones look the same in run-rate math and nothing alike in an enterprise pitch.

Business of GTM

When a vendor pitches its run rate this quarter, ask on the first call how many customers it has, what its monthly churn is, and which partners drive the number.

By Business of GTM