Who this is for: ecommerce and POD owners, and whoever runs their paid social, trying to decide how much creative budget should move into AI video tools.

On September 24, 2026, Higgsfield AI said its annualized revenue had passed $1 billion, eighteen months after its browser platform launched. For a company most marketers first noticed through viral camera-motion clips, that’s a remarkable number. It also needs careful reading, because “annualized revenue” is not a year of sales, and nearly every figure in the announcement comes from the company itself.

Quick answer: the Higgsfield AI milestone is credible as a snapshot and lines up with Higgsfield’s earlier disclosures, but it’s a company-reported run rate, not audited revenue. For brands, the more useful signal is where the money comes from: businesses paying for AI-generated ads at volume.

What Happened

Founder and CEO Alex Mashrabov announced the milestone in a post on X and LinkedIn, writing that Higgsfield “crossed $1B in annualized revenue” and has grown 20x since September 2025. The company’s own account framed it as 18 months after launch, adding that more than 30 million people now use the platform and that enterprise adoption has grown tenfold since June.

Bloomberg reported the news the same day and explained the method: Higgsfield takes its most recent four weeks of revenue and multiplies by 13. Mashrabov told Bloomberg that direct-to-consumer businesses drove much of the growth, describing a snowball effect where brands that see a return on AI-generated ads keep producing more of them.

The timing had a twist. That same day, OpenAI switched off the Sora 2 API, months after closing the Sora app, which took one of the best-known names in AI video out of the market.

What’s Confirmed and What’s Company-Claimed

Here’s how Higgsfield AI’s headline numbers hold up against independent sources:

None of this means the figures are wrong. It means no auditor or regulator has checked them, which is normal for a private company but worth remembering before you repeat them in a board deck or a pitch.

How Higgsfield AI Got Here

Higgsfield AI was founded in 2023 by Mashrabov, who previously led generative AI at Snap after Snap acquired his earlier startup, and CTO Yerzat Dulat. The San Francisco company launched its web platform in March 2025. Much of what it sells sits on top of other companies’ models, including Google’s Veo, Kuaishou’s Kling, ByteDance’s Seedance, and image models like Nano Banana and Flux, with Higgsfield’s own templates, camera presets, and in-house models layered over them. Its Marketing Studio, which turns a product photo or product-page URL into UGC-style ads, try-ons, and product shots, is aimed squarely at ecommerce teams.

The revenue climb has been steep, and it was tracked publicly along the way:

Mashrabov told reporters in February that he was aiming for a $1 billion run rate by the end of 2026. By the company’s account, it got there about three months early. That consistency is the strongest argument for taking the number seriously: each milestone matches a dated funding round or earlier interview, and the January Series A extension was reported with the same $200M figure at the time.

Is It Really the Fastest Ramp in AI?

The “fastest ever” framing comes from a chart Higgsfield circulated, which OfficeChai described as measuring each company’s climb from $1 million to $1 billion in run rate. On that chart, Higgsfield AI took about 18 months, against roughly 22 for Anthropic and Cursor.

Source: Higgsfield, via OfficeChai

Two cautions apply. Every company on that list reports run rate its own way, so the race is between self-reported numbers. And Higgsfield AI sells tools to creators and marketers, while Anthropic sells model access at a far larger scale today, so matching milestones doesn’t mean matching businesses. A fair reading: Higgsfield’s ramp is among the fastest ever reported, and “fastest in history” is its own claim.

What “Annualized Revenue” Tells You, and What It Doesn’t

A run rate is a short snapshot stretched into a year. Here’s the math with the assumptions showing:

• If $1B equals the last four weeks × 13, those four weeks brought in roughly $77M ($1,000M ÷ 13).
• If a strong ad season lifts one four-week window 30% above normal, annualizing that window inflates the run rate by the same 30%. (An illustrative assumption, not a Higgsfield figure.)

That second point is why AdExchanger said the milestone “comes with an asterisk. Ad-driven businesses swing with the seasons, and Q4 is the heaviest ad period of the year. A growing share of Higgsfield AI’s revenue also comes from on-demand credits rather than subscriptions, and usage revenue moves around more from month to month.

A run rate says nothing about profit, either. “Positive gross margin since the beginning of this year” implies margins were negative before, and gross margin still sits well above net profit on an income statement.

Who’s Affected

DTC and Shopify brands are the stated growth engine. If AI ad production is paying off for your competitors, your creative testing pace is being measured against theirs whether you use these tools or not.

Creative agencies are the biggest paying group, reportedly around 70% of revenue by company figures. The agencies buying at scale tend to treat AI generation as a production layer, not a substitute for strategy.

UGC creators and small production studios feel the most direct pressure on routine work like simple product demos and talking-head variants.

The wider market points the same way. According to the IAB’s 2026 video ad spend report, nearly two in three ad buyers now use generative AI for digital video creative, up from half in 2025, and they expect about a third of this year’s ad assets to involve it.

What Higgsfield AI’s Growth Means for Your Creative Budget

The practical shift is that producing ad variants keeps getting cheaper, so testing discipline becomes the edge. On Meta and TikTok, creative fatigue pushes acquisition costs up, and a tool that turns one product page into twenty variants only helps if someone decides which few deserve spend, understands why a hook works, and checks that the output still looks like your brand.

Picture a hypothetical apparel brand whose best Meta ad fatigues every two to three weeks. AI generation can refill that pipeline much faster than a reshoot. What it can’t do on its own is write the offer, catch an off-brand color or a warped logo, or notice that a generated “customer” breaks a platform’s disclosure rules.

That judgment layer is where AI-heavy ad programs either scale or stall, and it’s the part our graphic design and ad creative team handles for clients, directing and finishing AI output so it’s usable at volume. The same holds for print-on-demand sellers using AI try-ons and lifestyle mockups: the tool saves the photoshoot, but print-ready POD designs still need a human pass. We walk through where people fit into this kind of pipeline in our guide to AI content workflows.

The Risks Worth Pricing In

• IP protection is limited. Higgsfield AI’s updated terms, effective August 27, 2026, say users own their outputs and can use them commercially. Protection against third-party IP claims, though, is offered on the Enterprise plan only, so on standard plans that risk sits with you.
• Uploads train its models by default. Product imagery and faces you upload can be used to improve Higgsfield’s models unless you delete them or have a plan with an opt-out.
• There’s a controversy record. In February 2026, a Forbes investigation reported offensive clips in creator marketing kits, creator payment complaints, and a suspension of the company’s X account. The company disputed parts of the criticism, and it later revised its terms after a separate backlash over content rights.
• It depends on other companies’ models. Google, ByteDance, and Kuaishou all sell their models directly and could compete harder for brand budgets. The upside of an aggregator is that losing one model, as happened with Sora, doesn’t break your workflow.

What to Do Now

If you own paid social creative, treat this as a reason to pilot, not to overhaul.

1. Run a four-week test. Produce AI variants alongside your existing creative rather than replacing it.
2. Put a person at the gate. A creative lead approves every asset for brand fit and likeness, plus any product claims, before it runs.
3. Measure what matters. Compare hook rate and CPA against your control ads, and track credits spent per asset you actually use.
4. Keep your inputs portable. Store prompts, reference images, and brand kits outside any single tool.

A few things would change this advice: audited revenue if Higgsfield raises again or files to go public, a sharp drop in its run rate after the holidays, pricing changes, or a need for IP indemnity that pushes you toward an Enterprise contract. And if you’d rather keep the creative engine running without managing it in-house, our design subscription covers AI-assisted ad and product creative with human review built in.

What to Watch Next

Higgsfield AI’s number gets its real test in the first quarter of 2027, when holiday ad budgets fade and the four-week window gets quieter. If the run rate holds through that dip, Mashrabov’s argument that application companies can capture real value on top of foundation models gets much harder to dismiss. If it slides, September will look more like a strong Q4 than a new baseline. Either way, AI-generated ads are already a routine line in many creative budgets, and the open question for your brand is who reviews what goes live.

Sources and dates

• Alex Mashrabov, post on X and LinkedIn, September 24, 2026
• Bloomberg, “AI Video Startup Higgsfield Says it Tops $1 Billion Run Rate,” September 24, 2026
• AdExchanger, daily news roundup, September 25, 2026
• OfficeChai, Higgsfield $1M to $1B ARR comparison, September 25, 2026
• Higgsfield Series B announcement ($400M at $5.4B), August 24, 2026
• TechCrunch, Series A extension, January 15, 2026
• Forbes, investigation into Higgsfield’s growth tactics, February 11, 2026
• Higgsfield Terms of Use update, effective August 27, 2026
• OpenAI Help Center, Sora discontinuation notice
• CNBC, Cursor funding and $1B annualized revenue, November 13, 2025
• IAB, 2026 Digital Video Ad Spend & Strategy report, July 14, 2026

FAQ

Is Higgsfield AI actually making $1 billion a year?
Not exactly. Higgsfield says its annualized revenue passed $1 billion, based on its last four weeks of revenue multiplied by 13. That’s a run rate, not recognized annual revenue, and it hasn’t been independently audited.

How long did it take Higgsfield AI to reach $1B?
About 18 months from the launch of its browser platform in March 2025, according to the company. It reported roughly $200 million in January 2026 and $700 million in August 2026.

Is Higgsfield the fastest-growing AI company ever?
Higgsfield claims its climb from $1 million to $1 billion in run rate beat Anthropic and Cursor. That comparison comes from the company’s own chart and rests on self-reported figures, so treat it as a claim rather than a settled fact.

Can I use Higgsfield AI videos in commercial ads?
Yes. Higgsfield’s terms say users own their outputs and commercial use isn’t restricted. Protection against third-party IP claims is only offered on the Enterprise plan, and uploaded content is used for model training by default.

What is Higgsfield AI used for?
Creators, marketers, and agencies use it to produce ads, product visuals, UGC-style videos, and cinematic clips, drawing on models such as Google Veo, Kling, and Seedance alongside Higgsfield’s own tools.

Does AI-generated ad creative replace a design team?
It replaces much of the routine production work. Brand direction, offer strategy, claims review, and choosing which variants deserve spend still need people.