Etzioni’s number is $29 billion. The number worth putting next to it: $380 billion.
That’s the confirmed post-money valuation from Anthropic’s most recent round, confirmed in our May 17 brief. Anthropic isn’t a virgin unicorn. It has paying enterprise customers, confirmed government contracts, and a commercially deployed product line. The $380 billion valuation is large, but it’s sitting on a monetization structure.
The $29 billion Etzioni estimates for his pre-revenue cohort is sitting on something different. Differentiating between these two capital structures is the practical problem Etzioni’s warning creates for anyone trying to read the AI funding landscape.
What Etzioni Is Actually Claiming
Etzioni’s analysis identifies approximately a dozen pre-revenue AI research labs with an estimated aggregate of $29 billion to $30 billion in VC funding and approximately $127 billion to $130 billion in paper valuations. These are his estimates, not disclosed market data, not regulatory filings, not independently aggregated figures. The cohort reportedly includes Safe Superintelligence (SSI). Per Let’s Data Science (a T4 source), Etzioni also referenced a lab called “Project Prometheus,” reportedly valued at $38 billion with $16.2 billion raised. That specific claim requires a higher-tier source confirmation before being treated as established fact.
The “virgin unicorn” coinage is useful precisely because it’s blunt. A lab can reach a $1 billion-plus valuation on the strength of researcher talent, published papers, and investor conviction about long-term research trajectories, without building anything a customer pays for. Etzioni’s argument is that approximately $127 billion in paper valuation is currently resting on that foundation, and that the biotech precedent, where pre-revenue companies faced sharp corrections when commercialization timelines extended, is instructive.
The biotech parallel has surface validity. It also has limits that are worth naming.
The Confirmed Capital Landscape: Infrastructure vs. Research
Hub-verified data from May 2026 gives a clearer picture of where the large-scale AI capital is actually going.
Anthropic’s $380 billion valuation confirmed on May 17. OpenAI’s $122 billion round confirmed May 18. Hark’s $700 million Series A confirmed May 22. Modal Labs’ $355 million round confirmed May 21. The SpaceX S-1 Anthropic compute deal reported May 23.
These rounds share structural features. They involve companies with deployed products, paying customers, or confirmed infrastructure use cases. Anthropic’s API is in production. OpenAI’s revenue is disclosed and growing. Modal and Hark are infrastructure-layer companies with B2B revenue models. The SpaceX-Anthropic compute relationship reflects compute demand from a commercially active AI deployment program.
The pre-revenue lab cohort Etzioni identifies is structurally different. SSI has explicitly stated it won’t pursue near-term revenue, a philosophical commitment to long-horizon safety research before commercialization. That’s a legitimate research strategy. It’s also a capital structure that depends entirely on investors accepting indefinitely deferred returns.
The Biotech Parallel: Does It Hold?
Etzioni’s biotech framing argues that a market loaded with pre-revenue companies eventually faces correction when commercialization timelines extend beyond investor patience. In 2021 biotech, that correction arrived when interest rates rose and speculative capital retreated. Pre-revenue biotechs that had raised at premium valuations faced sharp multiple compression.
The AI parallel has structural differences worth examining.
First, the revenue pathways are different. A pre-revenue biotech requires FDA approval, clinical trials, and market launch, a 5–10 year process with regulatory gatekeepers. A pre-revenue AI lab can generate revenue through API access the day it decides to. The commercialization switch is faster. That doesn’t mean they’ll flip it, but the structural barrier is lower.
Second, the investor base is different. Anthropic’s round included sovereign wealth funds. OpenAI’s included Microsoft and SoftBank. These are not hedge funds that need quarterly marks, they’re long-duration capital sources with appetite for extended pre-revenue periods. SSI’s