Overview of 20VC: $1BN ARR in 18 Months; The Untold Story of Higgsfield
In this episode of 20VC, Harry Stebbings interviews Alex Mashrabov, founder of Higgsfield, about the company’s explosive growth, its AI video strategy, and the unusual path that took him from competitive programming in Kazakhstan to building one of the fastest-growing consumer AI businesses in the world. The conversation covers Higgsfield’s reported $1B annualized revenue run rate, its heavy use of models, the decision to prioritize distribution and product over “moats,” and why Mashrabov believes the future of content is increasingly AI-generated.
Key Takeaways
-
Higgsfield’s growth has been extreme
- Mashrabov says the company went from $1M to $1B in annualized revenue in 18 months.
- Revenue is calculated as live revenue over the last four weeks, annualized, and not from signed-but-not-yet-realized contracts.
- The company is still early in its video adoption curve, but already seeing significant expansion from customers.
-
The company’s core thesis is AI-native video for marketing and social media
- Higgsfield was built around the idea that video AI will become a central production layer for social content, advertising, and brand storytelling.
- The biggest customer use cases include:
- High-volume direct-to-consumer ad creative generation
- Short-form drama / serialized content
- AI-assisted commercial production
- Brand asset management and searchable content libraries
-
Distribution and content are the real growth engine
- Higgsfield reportedly uses no paid advertising.
- Instead, it relies on:
- An in-house creative team of 150+
- Tutorials, product launch videos, and open-sourced projects
- Creator-led and content-led distribution
- Mashrabov argues that owning distribution matters more than conventional marketing in AI.
-
Internal model usage is enormous
- The company spends over $4M per month on models.
- Average spend is said to be over $10K per employee per month.
- Some individuals spend $30K+ in a week on AI models during intense experimentation.
- Mashrabov sees this as evidence that AI is becoming embedded in real production workflows.
-
Higgsfield is becoming a model-routing business
- The company initially tried to build more of its own models, but Mashrabov says this was partly a mistake.
- Today, the emphasis is on:
- Choosing the right model for the task
- Using open-weight models and post-training where useful
- Optimizing for cost, steerability, and customer outcome
- He frames this as “tokenomics”: selecting the most efficient tokens and models for the job.
Founder Story and Origin
From Kazakhstan to Silicon Valley
- Mashrabov grew up in Central Asia, where economic opportunity was limited and competitive achievement was a major route to success.
- His parents, both engineering professors, pushed him early into:
- Programming competitions
- Math/data structure camps
- International academic contests
- By age 19, he was reportedly top 3 in the world in competitive programming.
Early AI Work and the Snap acquisition
- Before Higgsfield, Mashrabov worked on early neural-network systems and language translation.
- He co-founded AI Factory, which was acquired by Snap for over $160M.
- At Snap, his team helped build face filters, which he describes as one of the most widely used consumer media AI products.
- That experience shaped his belief that AI-generated video would become a dominant phone-native medium.
Product Strategy and Go-to-Market
Finding product-market fit
- Higgsfield did not have immediate PMF.
- The team reportedly burned more than $10M of its $16M seed round searching for the right product.
- The breakthrough came after speaking directly with creative directors, who said the biggest missing feature in AI video was camera control.
- Once Higgsfield launched its product in March of last year, Mashrabov says PMF was immediate.
Consumer and enterprise split
- Revenue is now slightly over 50% business/enterprise, with the rest split across consumer and prosumer use cases.
- Pure consumer mobile use is still a relatively small share.
- He believes the company will not win by competing for $20/month subscriptions alone, because horizontal giants like OpenAI and Google will likely dominate those entry-level tiers.
- Instead, Higgsfield aims to move users up to $1,000+ annual spend via utility and workflow depth.
Retention and expansion
- Mashrabov admits there is a large early drop-off in consumer retention, largely because users don’t immediately understand the product’s value.
- But once customers are active, expansion is strong:
- He cites a customer that went from $99/month to a $6M annual deal
- He says month-12 NRR is over 300%
- His view: churn early, expand later if the product becomes embedded in workflows.
Views on AI Models, Benchmarks, and Moats
Benchmarks are overrated
- Mashrabov is skeptical of AI benchmark culture.
- He argues that benchmarks often don’t reflect real-world workflows, especially in video.
- For video, he says:
- The prompt length can be 3,000+ words
- Each scene may use 10+ image references
- Video generation is more like a modern rendering engine than a simple text-to-video task
- In his view, benchmark obsession can encourage bad behavior and doesn’t translate into product success.
Moats are limited
- He says traditional moat thinking is “mostly bullshit.”
- The two real sources of moat, in his view, are:
- Delivering a real outcome
- Network effects
- For Higgsfield, the moat is not just model quality; it’s the ability to help customers sell more through AI-created content and to build a community flywheel via open source.
Open models and model routing
- Mashrabov expects open-weight/open-source models to continue gaining share.
- He believes many companies don’t need frontier proprietary models and will prefer:
- Cheaper
- More steerable
- Workflow-optimized systems
- Higgsfield already chooses the model itself in over 40% of cases, depending on the task.
Team, Culture, and Hiring
Big teams, heavy usage
- Higgsfield has roughly 400 people.
- Around 50 are in California, while 300+ are in Kazakhstan, with additional remote talent elsewhere.
- The company is unusually distributed and leans heavily on talent from:
- Kazakhstan
- Europe
- Other parts of Asia
Hiring philosophy
- Mashrabov is strongly meritocratic in his hiring views.
- He says the company is trying to build a place where the best people can do the best work, with minimal “corporate theory.”
- He’s inspired by leaders like:
- Jensen Huang
- Elon Musk
- Nick (likely a reference to a hard-charging product/operator leader)
- His preferred management style is direct, detail-oriented, and execution-heavy.
Work-life tradeoffs
- He is candid about the sacrifices involved:
- Around 80–90 hours/week on Higgsfield
- Limited family time
- He frames this as a combination of:
- Immigrant drive
- Fear of missing the opportunity
- Deep conviction in the technology and market
Geography, Asia, and the Future of Content
Asia as the innovation signal
- Mashrabov repeatedly points to Asia as the source of major content trends:
- Short-form drama
- AI-native DTC marketing
- AI-assisted production workflows
- He says Higgsfield learns a lot from Asian behavior, even though:
- 70%+ of revenue comes from the West
- He believes Asia is often earlier in adopting direct-to-consumer and AI-enabled content strategies.
The future of content
- He believes most social content will become AI-generated or AI-assisted.
- Authentic, human-made content will still exist, but it may be more valuable and command higher CPMs.
- He predicts new roles will emerge around:
- AI-native creative direction
- Real-time story generation
- Managing multiple AI-produced variations
Notable Quotes and Ideas
- “Moats are bullshit.”
- “Distribution matters.”
- “When you know, you know.” — on product-market fit
- “The best product is going to win.”
- “Most of the content on social media is going to be AI generated.”
- “Hire the best people, empower them, and retain them.”
Bottom Line
This episode presents Higgsfield as a highly unconventional but rapidly scaling AI company built on video generation, distribution-first growth, and heavy model usage. Mashrabov’s core belief is that the biggest opportunity in AI is not just model development, but owning the workflows that help businesses create more content, sell more, and move faster. The result is a company that he believes can become infrastructure for the future of advertising and content production.