Key Takeaways

  • Keith Rabois warns that an IPO misstep or gross margin disappointment from OpenAI or Anthropic would trigger an immediate, sharp reset for private venture valuations across the sector.
  • Writing a check of $50 million at a $500 million valuation can be rational today because the potential upside for market winners is one to two orders of magnitude larger than in prior tech cycles.
  • Early revenue velocity matters less than team capability; building with the right engineers against the right hard technical problems trumps short-term growth clips.
  • Empirical labor data reveals zero aggregate job loss caused by AI, defying popular existential panic.

The OpenAI and Anthropic IPO Canary

Private market valuations currently float on the assumption that AI giants will produce historic public market debuts. If those public debuts reveal thin gross margins, crushing inference costs, or sluggish enterprise retention, private venture pricing will crater overnight.

Rabois puts the vulnerability bluntly: “the most devastating thing to the frothy hype and the traction that we're seeing with AI and the funding of it from VCs would be anthropic and or open AAI have a blip in their public exits, their public their IPOs.”

If you are raising on a multiple borrowed from foundation model hype, your pricing is leased on borrowed time. When the public markets inspect the real cost of compute and pricing power of the category leaders, every seed and Series A pricing model will adjust instantly.

The Math Behind a $500M Early Valuation

Paying top dollar for early startups looks like reckless investing to outsiders who compare today's rounds to SaaS multiples from 2018. Rabois argues that the arithmetic changes when the addressable outcomes expand by ten to one hundred times.

“It is possible now that at least one order of magnitude, possibly two orders of magnitude have been added to the upside potential of a startup, which does adjust the prudence and rationality of a VC investing $50 million at $500 million,” Rabois explains.

When software captured small software budgets, paying a 100x multiple on early revenue was financial suicide. When AI software replaces entire labor lines in massive industries, a company capturing a market becomes a trillion-dollar asset rather than a ten-billion-dollar one. High prices are rational if and only if the end-state prize is vastly larger.

Talent Density Beats Early Traction

Too many founders obsess over showing week-over-week revenue growth before their core product actually solves a difficult problem. In software layers powered by models, distribution tricks can fake early product-market fit, but that revenue evaporates the moment an underlying model update makes the wrapper obsolete.

Rabois looks past early revenue numbers to judge the team itself: “the critical density of talent does matter. Yeah. And so if a company is in the short term at low scale growing at you know xclipip that's not that critical if they have the right team assembled against the right challenges.”

True defensibility comes from engineers who can optimize models, build proprietary data loops, and execute technical workflows that commodity API wrappers cannot match. If you have the best minds solving the hardest friction point in a workflow, slow early traction is acceptable. If you have an ordinary team shipping a thin skin over GPT-4, high early growth will not protect you.

The Ghost of AI Unemployment

Tech pessimists and safety researchers frequently warn about catastrophic labor displacement and existential risk. Rabois dismisses both arguments as unsupported by facts.

“Every single piece of evidence there's, you could look through, you could be the Sherlock Holmes, you know, look with your magnifying glass. You can't find any evidence that AI is causing creating job loss,” Rabois notes. Instead of eliminating work, intelligent tooling increases aggregate output and shifts human attention to higher-leverage tasks.

What to Do With This

Audit your cap table and hiring bar before your next fundraise. If you are charging customers for a workflow layer, calculate your true gross margins after accounting for model API calls and compute overhead. If your margins are below 60 percent, fire your next two marketing hires and use that budget to recruit one senior systems engineer who can compress your token usage and optimize your local model inference.