The whispers started in the same tone we used to hear before the last cycle's top. A number, floating through private Telegram rooms and networking event chatter: $2 trillion. That's the valuation target allegedly circulating around Anthropic, the AI darling that positions itself as the "safe" alternative to OpenAI. The chart screams, but the order book whispers—and right now, the order book is saying something entirely different.
We've seen this movie before. It's the same script we watched play out in 2021, when every Layer-1 with a Telegram channel and a founder who wore sunglasses indoors was printing a "valuation" that dwarfed the GDP of small nations. The mechanics are different—this time it's about model compute instead of total value locked—but the emotional DNA is identical: narrative first, fundamentals never.
Here's the part that doesn't fit the press release. Anthropic's reported annual recurring revenue sits somewhere in the hundreds of millions, maybe pushing toward the billion-dollar mark. OpenAI is reportedly clearing $3 billion plus, and its own valuation is priced around $300 billion. So to get to $2 trillion, Anthropic needs to jump a valuation gap that makes the BTC-to-$1M crowd look conservative. Liquidity is just patience wearing a speedo, but this is not liquidity. This is pure, unadulterated narrative.
Context: The Old Playbook
The modern AI arms race operates on the same principle that drove the brutal crypto bear of 2022: capital efficiency has been sacrificed at the altar of market dominance. The Aave and Compound interest rate models of the world taught us that the true market price is rarely what the protocol says it is—it's what the marginal buyer is willing to pay when the music stops. In AI, the "protocol" is the model itself, and the "interest rate" is the price per million tokens.
Anthropic's current model is not wrong. It's just severely mispriced for the current velocity of capital. Post-Dencun, we saw rollup fees double and the immediate aftermath is that every builder started looking for alternative data availability layers. The same thing will happen here. When the cost of compute—which is Anthropic's COGS—remains stubbornly high, the valuation gap widens. We are looking at a fundamental mismatch between what it costs to operate and what the market is willing to underwrite.
From my audit experience of liquidity pools in 2020, I remember watching yield farmers chase the highest APY without looking at the underlying smart contract risk. They saw a high number and assumed a high opportunity. The same pattern is playing out with AI equities and private placements. The 2T number is a "hook," not a value. It's designed to grab headlines, attract the best talent, and create a hype loop that compels the Amazons and Googles of the world to keep writing checks.
The Core: The math doesn't back the story
Let's cut through the noise with a calculator. If Anthropic pulls a 50% compound annual growth rate (CAGR) over the next ten years—an absolutely unreasonable assumption in a mature market—they'd hit roughly $400 billion in revenue. For context, that's about the current revenue of a global mega-cap tech conglomerate. At a 10x price-to-sales multiple, you get to a $4 trillion valuation. That's the upper band of the "moon math."
To hit $2 trillion today, you'd need a P/S multiple of over 2,000. That's not an investment; it's a work of fiction. We are talking about a valuation that is more akin to the total market cap of the S&P 500's top five companies combined. It's like comparing a new Layer-2 with $10 million in TVL to Ethereum itself. The chart screams, but the order book whispers, and the whispers are saying something like "illiquidity."
What's more interesting is the intent behind the leak. It's a masterclass in negotiation. By leaking a $2 trillion target, Anthropic sets the anchor for its next round. If they walk in asking for $1 trillion, they look "reasonable" compared to the leaked target. It's the classic "shoot for the moon" pricing strategy. In the crypto world, we called it the "FDV game"—the fully diluted valuation of a token that doesn't have the volume to back it. It's the same trick of setting a high "whale rate" in a private sale to make the public sale look like a discount.
The contrarian angle: The safety tax
Here's the blind spot that nobody wants to talk about. Anthropic's entire brand is safety. The "Constitutional AI" ethos. The commitment to avoiding the "race to the bottom" of AI safety. That is a noble positioning, but in the market, it's a cost center, not a profit center. You're telling the market that your model is "safe" but the market doesn't pay for safety until after a black swan event.
The market whispers that "safe" is just code for "slower." While OpenAI is shipping GPT-5 and pushing boundaries into Agentic AI, Anthropic is holding back, checking the red lines. That's a huge differentiator in the minds of regulators, but it's a squeeze on the revenue curve. The market that is pricing in a $2 trillion valuation is betting that "responsible" will become a premium feature, not a standard. If AI safety doesn't become a paid mandate (i.e., a regulatory requirement with a fine attached), Anthropic is just a less powerful, less widely adopted version of OpenAI with a similar budget.
Takeaway: The next watch
Watch the next funding round. If the next raise comes in at $500 billion to $1 trillion, the leaked $2 trillion was a successful negotiating tactic. If it comes in at $200 billion, the market has rejected the narrative. If it comes in at $1.5 trillion, we are officially in a full-blown, 2021-style bubble.
Panic is just uncalculated opportunity in a hurry. The next few months will tell us if this is a headline grab or a fundamental shift in the AI narrative. From the rush to the slump, we kept moving. Speed kills, but hesitation bankrupts. The AI bubble—if it is a bubble—is moving fast. Keep your eyes on the actual revenue, not the headlines. The chart screams, but the order book whispers.
