AI Startups Are Now Adding Revenue Faster Than Any Software Cohort in History

The AI industry's revenue curves are not just steep — they are getting steeper, according to figures compiled by TechCrunch this week showing that leading AI companies are adding revenue faster than any software cohort before them.

The Numbers

Anthropic is the standout. In late May the company announced it had crossed a $47 billion revenue run rate — a milestone reached less than two months after passing $30 billion. That is acceleration, not just growth: the time to add each successive billion is shrinking rather than lengthening, the opposite of how software businesses have historically matured.

The pattern repeats down the stack. Sierra, which builds customer-service AI agents for enterprises, added another $100 million in ARR in just two quarters after reaching its first $100 million. Glean, the enterprise search and work-assistant company, crossed $300 million in ARR, needing only six months to grow from $200 million to $300 million.

Why the Curves Keep Bending Up

Two forces compound. First, agentic products bill on usage, so revenue scales with adoption in real time rather than through annual seat renewals. Second, enterprises have moved from pilots to production, wiring AI into workflows that consume tokens continuously. The result is a growth profile that looks less like traditional SaaS and more like a metered utility during a demand boom.

The Shadow on the Chart

The velocity comes with an asterisk. The same period saw Chinese open-weight models capture up to 46% of US enterprise token traffic on price, raising the question of how much of this revenue sits on premium tiers now facing an 80%-cheaper alternative. And the costs behind the revenue are enormous — multi-billion-dollar compute commitments that these run rates must eventually justify.

With OpenAI and Anthropic both reportedly targeting IPO roadshows this fall, the revenue-velocity story is the one they will tell investors. Whether the curve keeps bending up, or bends back as cheaper models erode the premium tiers, is the defining financial question of the AI market's second half.