The Twenty Minute VCAnthropic Raises $30BN at $900BN Price | SpaceX Files S1: How Does it Trade | Cerebras Smashes Day 1
At a glance
WHAT IT’S REALLY ABOUT
AI mega-rounds, token economics, IPO frenzy, and looming backlash risks
- Anthropic’s rumored $30B raise at a ~$900B valuation is framed as an unusually attractive late-stage “post-IPO-like” risk profile where the main risk is valuation rather than survival or liquidity.
- Token spend is treated as the core driver of AI model TAM, with Salesforce’s reported $300M Anthropic spend used to estimate how much of engineering and knowledge-worker payroll must shift to tokens to justify trillion-dollar revenue projections.
- Public SaaS is depicted as leaving the 2021 “prospects-driven” valuation era and settling into tighter outcome bands, with Datadog and Figma praised for re-acceleration while Wix/Squarespace are described as facing structural decline.
- Compute scarcity is the key variable behind the boom in “AI picks-and-shovels” (data centers, semis, GPU clouds like CoreWeave/Nebius), with permitting and build-speed constraints potentially preventing catastrophic oversupply.
- Cerebras’ strong IPO is viewed as an N-of-1 “hot intersection” (semis + inference + OpenAI adjacency), while SpaceX’s potential mega-IPO is expected to be sentiment-driven and volatile, and the episode closes warning that layoffs and AI messaging are creating political backlash.
IDEAS WORTH REMEMBERING
5 ideasAnthropic’s round is positioned as “low existential risk, high valuation risk.”
The speakers argue Anthropic is large enough and close enough to an IPO that the usual venture risks (company failure, no liquidity) are reduced, leaving investors primarily exposed to paying too high a price.
If ARR multiples remain the market heuristic, mega-AI may look ‘cheaper’ than early-stage SaaS.
They claim Anthropic trading around ~18x near-term revenue can compare favorably to 20–50x ARR paid for much smaller, earlier companies that are years from IPO—though the metric may be misleading for model businesses.
Token spend must become a meaningful slice of payroll to justify AI model valuations.
Using Salesforce’s $300M as a benchmark, Rory suggests the industry may need something like ~20% of engineering wage spend and ~5–7% of knowledge-worker wages flowing to tokens to reach trillion-dollar token revenue outcomes.
There’s a real bear case: better models and efficiency could cap token spend.
Jason cites examples (SaaStr usage; Klaviyo’s agent framework) implying agents can be cheaper than assumed, meaning token TAM projections could be overstated if efficiency gains outpace usage growth.
For SaaS, 2021 multiples are gone; the game is ‘become a great normal company.’
They argue the ‘prospects veneer’ moved from SaaS to AI, compressing SaaS outcomes toward revenue/growth/cash-flow multiples; re-acceleration (Datadog, Figma) matters more than reclaiming peak multiples.
WORDS WORTH SAVING
5 quotesAs it's becoming painfully clear now, no one in America, other than us here in California, likes the AI trend. We have people who are brilliant scientists who politically are utter morons, and the people who are utter morons at AI but brilliant at politics are gonna have us for lunch.
— Rory O’Driscoll
Because it de- because you're giving away 30 over 900, which is like, um, let me see, 3% of your business, to de-risk it for another year of monstrous burn... I mean, this is a big ass balance sheet war, so to me, it's a no-brainer to do it. They're gonna raise, and they're gonna raise again, and then they're gonna raise again.
— Rory O’Driscoll
At 1% of R&D spend, it's lost in the noise. At 5%, it's real. That's a layoff, right? At 20%, which, let me repeat, is what it takes to get to for these overall models to work, right, for these overall TAM analysis to work, that's huge. It's one-fifth of your payroll costs in engineering.
— Rory O’Driscoll
Every tech in- cycle has a set of industries that for the f- once in their life get valued on prospects and futures. It's like being 21. It literally is like being 21. And people will believe everything about you.
— Rory O’Driscoll
We're gonna have to reflate and hire thousands and thousands of people per tech leader, um, to avoid social unrest. We're gonna have to do it.
— Jason Lemkin
High quality AI-generated summary created from speaker-labeled transcript.
