Key Takeaways

  • OpenAI applied targeted pressure on Anthropic from July through September, compressing Anthropic's summer metrics right ahead of its planned public offering.
  • Anthropic countered by releasing Claude 3.5 Opus, engineering a sharp month-over-month growth inflection from September into October.
  • Public market tech investors operate with the memory of goldfish, fixating on thirty-day acceleration rather than multi-quarter trends.
  • A strategic leak of Anthropic's October ARR to The Wall Street Journal at the low point of investor sentiment helped drive the order book to six to seven times oversubscribed.

The Short Memory of Institutional Capital

When momentum stalls ahead of an IPO, most executive teams freeze roadshows or accept steep valuation discounts. Anthropic faced that exact squeeze in late summer. OpenAI spent the months between July and September shipping updates specifically designed to box Anthropic out of the enterprise conversation. The offensive worked on paper: Anthropic's September numbers showed visible deceleration, rattling institutional allocators.

As Tae Kim noted on TBPN, the resulting panic misunderstood how public market analysts actually make decisions:

“Wall Street is not that smart. They're kind of like goldfish. They just, you know, what happened this month?” (play from 24:27)

Investors rarely evaluate enterprise software businesses on the smooth arithmetic of a three-year average. They want immediate velocity entering the first public quarter. By allowing September to look depressed, Anthropic set up an easy base against which any autumn recovery would look explosive.

Engineering the Growth Curve

The inflection came through product timing. Anthropic held Claude 3.5 Opus until the tail end of OpenAI's summer push. When Opus dropped, enterprise usage rebounded immediately, creating a hockey-stick contrast between late September and mid-October.

Kim explained how OpenAI's aggressive summer schedule actually set up Anthropic's win:

“I think part of OpenAI's strategy of really putting pressure on Anthropic from July to September kind of backfired. Because it made Anthropic's numbers look worse in September, and then Opus 3.5 came out. It was such a great model that their growth curve from the September quarter to October looks great.” (play from 24:02)

Metrics alone do not clear an order book; distribution of the narrative does. As rumors circulated that the listing was in jeopardy, Anthropic took control of the information flow.

“Someone very brilliantly at the bottom of the market leaked the October ARR to The Wall Street Journal,” (play from 24:45) Kim said. “Then sentiment completely flipped around, and now we're all there seven times.” (play from 24:53)

John Coogan agreed that the reset set up the transaction: “Even with all the turmoil and the chaos in with Nvidia, but also the market broadly, it's not the craziest time for Anthropic to go public next week. It feels like they reset the valuation and it's perfect.” (play from 23:42)

What to Do With This

Map your next fundraise or board review around second-order derivatives rather than steady-state revenue. If an incumbent attempts to box you out during a slow quarter, do not burn cash on marketing counter-punches. Hold your major feature or model release until the incumbent's news cycle peaks, ship when their campaign cools, and present the resulting thirty-day spike as your new baseline trajectory.