AI startups now hit $100M revenue in 18 months, not seven years

For two decades, software investors carried the same clock in their heads: a good SaaS company takes roughly seven years to reach $100 million in annual recurring revenue. In the technology economy taking shape now, that clock is broken.

The clearest dataset comes from Bessemer Venture Partners’ State of AI report, which tracked twenty high-growth AI startups and found a class it calls “Supernovas” averaging roughly $40 million ARR in their first year of commercialization and about $125 million by year two. In Bessemer’s summary, that is $100 million ARR in about 18 months, against the seven years the old benchmark allowed.

Two speeds, two kinds of company

Bessemer’s data splits fast-growing AI companies into two archetypes. Supernovas sprint: consumer-adjacent products, often built close to foundation models, with gross margins near 25 percent and sometimes fragile retention. “Shooting Stars” look more like classic SaaS: about $3 million ARR in year one, scaling to roughly $103 million by year four, with healthier 60 percent margins.

Institutional money is reading the same chart. Hamilton Lane’s 2026 Market Overview shows time-to-revenue for AI companies running far ahead of historical software cohorts, and the firm’s executives have described AI companies scaling five to ten times faster than historical norms. One driver sits on the demand side: enterprise buyers are pushing agentic AI deployments into production budgets at unusual speed.

What the new clock means for money

Three consequences follow. Diligence windows shrink, because a company can define a category in two quarters. Revenue quality matters more than revenue speed, because a 25 percent margin business at $100 million ARR is a very different asset from a 60 percent one at the same size. And the benchmark language itself is changing: investors have started replacing the old “triple, triple, double, double, double” growth shorthand with steeper curves.

What to watch

Retention data will decide which Supernovas were real businesses and which were usage spikes wearing a subscription. Watch the next round of renewals at the fastest scalers, and watch whether margin profiles improve as inference costs fall. Speed has been proven. Durability has not. In healthcare, Bessemer’s separate health AI data shows the same compression with a five-year clock. Cognition’s coding agent Devin is a case study in the pattern: its valuation more than doubled to $26 billion on revenue growth that outpaced its headcount by a wide margin.