Alex
Mashrabov

Speed Is the Strategy

When Higgsfield's consumer app launched on March 31, 2025, something broke - or at least that's what it looked like from the inside. Traffic spiked 20x within a month. The team thought they were under a DDoS attack. They weren't. It was just people, a lot of them, wanting exactly what Alex Mashrabov had built.

By month five, Higgsfield had 11 million users and $50M in annual recurring revenue. By month nine, $200M ARR, 15 million users across 240 countries, and a $1.3 billion valuation. Cashflow positive the whole time.

That kind of velocity is not accidental. It's engineered. Higgsfield ships 4-7 feature updates per week, sometimes one per day. New AI models get integrated within 24 hours of public release. The release schedule runs 300+ times a year. In the AI video space, where competitors run quarterly update cycles, this is not an iteration advantage - it's a different category of operation.

From Andijan to the World Finals

Mashrabov grew up in Andijan, Uzbekistan, where his parents are both mechanical engineering professors. He started programming at 10. By 20, he was ranked in the top three competitive programmers on the planet.

At the ACM ICPC World Finals - the Olympics of competitive programming - he competed twice: 53rd in 2014 with Moscow Institute of Physics and Technology, then 38th in 2016 with Innopolis University. Along the way: a bronze medal at CodeChef SnackDown 2015, 8th at the VK Cup, and a handle (map on Codeforces) that still shows up in the competitive programming Hall of Fame.

That background is not incidental to the rest of his career. Competitive programming is precision work under pressure, where the difference between first and second is often one edge case caught at the last second. Mashrabov brought that instinct to product development: the discipline of shipping fast and getting it right.

The Snap Chapter

At 18, he turned down an internship offer from Meta. Later, he'd describe it as one of the best decisions he ever made.

Instead, he co-founded AI Factory, a startup building neural networks for mobile devices and AR-powered video. In 2020, Snap quietly acquired AI Factory for $166 million. The deal was the foundation for Snapchat's Cameos feature - deepfake-adjacent tech that let users insert their face into video content. Mashrabov's neural networks were now running on one of the world's largest social platforms, at scale, globally.

He stayed at Snap as Director of Generative AI, eventually becoming Head of Generative AI. His team built the AI systems underlying Snapchat's face filters - tools that reached hundreds of millions of users. This was not a niche role at a small company. This was production AI at the bleeding edge of consumer scale, where everything that breaks, breaks visibly, and immediately.

Building Higgsfield

Higgsfield AI launched in 2023 as a bet on a specific thesis: the future moment when most social media video is AI-generated. Not a bet on models. A bet on infrastructure. There are hundreds of AI video models. Higgsfield's game is being the orchestration layer that makes them all usable, fast, and professional.

The platform aggregates 30+ video and image generation models - Sora 2, Kling 3.0, Veo 3.1, Seedance 2.0, and more - and wraps them in production tooling: Cinema Studio, Marketing Studio, viral presets, Adobe Premiere and After Effects integrations, a canvas workspace, and a proprietary supercomputer orchestrator that runs 8x cheaper and 3x faster than alternatives. The platform generates approximately 4.5 million videos daily.

Early on, a catalog of 100+ cinematic templates with pre-defined camera controls cracked the code on user adoption. The templates made AI-generated video look cinematic rather than experimental. Creators who had never touched a camera control could suddenly produce content that looked deliberate. That move catalyzed the initial growth spike.

The Canva Playbook

When asked which company he models Higgsfield after, Mashrabov does not say OpenAI or Runway. He says Canva - a $4B ARR business with 20 million paying users, built from Australia without being in the center of the AI hype cycle. Canva's lesson was not about technology. It was about discipline: a focus on paying users, operational efficiency, and building a durable business rather than chasing valuation.

That philosophy shows up in Higgsfield's numbers. The company is cashflow positive. Its 300 employees - many of them top-tier competitive programmers recruited from Almaty, Kazakhstan - are operating in a flat hierarchy where everyone does substantive work. Mashrabov expects 10-12 hour days, and makes no apologies for it.

The enterprise side is accelerating. Customers with marketing budgets exceeding $100 million are now generating 90% of their social media ad creative through Higgsfield. Broadcast-quality video output, which cost roughly $100,000 per minute in the early days of AI video, now runs around $500 per minute on the platform - a 200x cost reduction that changes what's possible for brands at scale.

The Long Game

The $1.3 billion valuation from Accel, AI Capital Partners, Menlo Ventures, and GFT Ventures is not the destination. Mashrabov has said publicly he sees Higgsfield as a company with billions of users in 10-20 years. The potential market - professional video creation running at hundreds of billions of dollars annually - is not a niche play. It's a platform bet.

He spoke at CES 2026 on how generative AI is transforming the creative industry. He was on the panel because Higgsfield had already done it - not because it planned to. That's the distinction. The competitive programmer from Andijan, who wrote neural network code before he was old enough to drive, is not projecting. He's reporting.