Last week, a rumor surfaced that Anthropic is seeking a $2 trillion valuation in a private bid round. If true, this would exceed the market cap of 99% of publicly traded companies, including Meta. But the numbers don't add up. The bid is a signal, not a fact. And in a bear market for attention, signals matter more than fundamentals.
Context
Anthropic, the AI lab behind Claude, is reportedly positioning itself for a $2 trillion valuation ahead of an autumn IPO. Alongside this, CoreWeave—the GPU cloud provider that nearly collapsed in 2023—has staged a comeback, with its stock rebounding 40% over the past quarter. The narrative: AI infrastructure is the new gold rush, and the IPO window is open. Crypto Briefing, a media outlet that straddles crypto and AI, framed this as a story of "winners and losers." But as someone who has spent years auditing cryptographic protocols and assessing scalability claims, I see a pattern that resembles the 2021 DeFi boom: strong narratives, weak fundamentals, and a lot of capital ready to misallocate.
Core: The Valuation Math Doesn't Compute
Let’s start with the $2 trillion number. At a 20x revenue multiple—generous for a high-growth tech company—Anthropic would need to generate $100 billion in annual revenue. Based on my analysis of cloud API pricing and public inference costs, Anthropic’s current annualized revenue is likely around $2–3 billion. That implies a 33x growth requirement. Even if we assume a 50% CAGR, it would take 10 years to reach $100 billion. The implied terminal value is more than the entire current market cap of Nvidia. This is not a valuation; it’s a plea for liquidity.
In 2022, I conducted a comparative benchmark of Optimistic Rollups vs. ZK-Rollups, analyzing 10,000 transaction simulations. The key insight: upfront costs hid long-term scalability bottlenecks. The same applies here. A $2 trillion bid is an upfront cost that hides the bottleneck of real-world adoption. Anthropic competes with OpenAI, Google, and open-source models. The marginal cost of inference is dropping. The total addressable market is large, but the capture rate is uncertain. CoreWeave’s "comeback" is even more suspect. As a GPU lessor, CoreWeave’s business model is highly levered: it buys hardware with debt and rents it out. A 40% stock rebound does not reflect operational improvement; it reflects a repricing of risk. Based on my earlier analysis of Compound Finance during the Terra collapse, I know that leverage can mask fragility. A 15% drop in GPU rental prices could wipe out CoreWeave’s EBITDA. Scalability is a trilemma, not a promise.
Contrarian: The Blind Spot of Narrative-Driven Capital
The contrarian angle is that the $2 trillion bid may not be a real transaction. In my experience auditing Zcash’s Sapling upgrade in 2020, I found that subtle side-channel vulnerabilities only appear under load. Similarly, high-profile valuation bids often leak to test market appetite. If the bid is a marketing signal, it’s a bluff. The real test will come when the IPO filing is made public. CoreWeave’s comeback is also fragile—it depends on continued GPU supply from Nvidia and stable electricity prices. Code does not lie, but it often omits the truth. The omission here is the debt structure. CoreWeave’s balance sheet, like many crypto lending protocols, is opaque. The chain is only as strong as its weakest node—and the weakest node here is the assumption that AI demand will grow exponentially without interruption.
Takeaway: Watch for the Filing, Not the Noise
Forward-looking: I forecast that if Anthropic’s valuation is confirmed in a regulatory filing, it will signal a peak in AI capital sentiment. The autumn IPO window will attract a flood of AI companies, but the market will struggle to absorb them. This mirrors the 2022 crypto deleveraging, where too many projects IPOed into a liquidity drought. The smart money is watching the actual transaction data, not the headlines. My advice: wait for the S-1. Let the code and the numbers speak for themselves.