← Home AI & LLM topic The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch Guest profile

Alex Mashrabov

The Twenty Minute VC (20VC): Venture Capital | Startup Funding | The Pitch

20VC: $1BN ARR in 18 Months; The Untold Story of Higgsfield | Spending $4M Per Month on Models | Why Moats in AI are BS | Scaling a Content Team to 150 People with Alex Mashrabov

Published: September 28, 2026
Duration: 1h 9m
Summary source: description
Last updated: Sep 28, 2026

Discusses openai, anthropic, investing.

Summary

Alex Mashrabov is the Founder and CEO of Higgsfield, the third fastest scaling company to $1BN in ARR behind OpenAI and Anthropic. Reports suggest their latest funding round could place an $8BN valuation on the company. Prior to Higgsfield, Alex sold his prior company to Snap Inc for $166M. Alex was a competitive programmer as a kid, reaching third best i…

Alex Lim reveals how Higsfield hit $1B revenue in 18 months with 300 people from Kazakhstan, spending $4M/month on AI models, zero paid ads, and NRR over 300%.

Key takeaways

Higgsfield reached $1B in annualized revenue in just 18 months from $1M—faster than Cursor (24 months)—with 300 people based primarily in Kazakhstan, zero paid acquisition, and 80%+ gross margins on open-source model deployments.
Internal AI model spend at Higgsfield exceeds $4M/month (~$10K+ per employee), with top engineers and creatives spending $30K+ per week on models like Claude and Codex—signaling that AI inference cost is becoming a core operating line, not a tool budget.
NRR at month 12 exceeds 300% (one customer grew from $99/month to $6M/year), driven by agentic workflow adoption in video ad creation—while the company deliberately avoids the $20/month prosumer subscription market it believes OpenAI and Google will cannibalize.

Why this matters

Higgsfield's trajectory—$1B ARR, 80%+ margins on open-weight models, and 300% NRR—demonstrates that vertical AI companies built outside Silicon Valley on post-trained open-source models can outpace well-funded incumbents by owning a specific outcome (AI-generated ad performance) rather than competing on benchmark-driven model development.