The numbers don't add up. OpenAI, once valued at $300 billion in private markets, now shows a post-money valuation of 852 billion RMB (~$117B) in the latest IPO rumor cycle. Anthropic, its primary rival, sits at 965 billion RMB (~$133B). The tether snapped before the press release hit the wire.
This is not a pricing anomaly. It's a narrative leak. The market is pricing a future that hasn't been coded yet. And as a narrative hunter, I can see the structural weakness in the story.
Let me be clear: I've been watching this trend since 2020, when I audited Uniswap v2 and found liquidity manipulation vectors that were later exploited. The same pattern applies here. The narrative is built on hype without the code to back it up.
Context: The AI IPO Pipeline
The article from BlockBeats outlines a clear timeline: OpenAI and Anthropic targeting 2026 Q4, Perplexity in 2027, Chinese firms like DeepSeek, Moonshot AI, Baichuan, and StepStar aiming for 2027-2028 on A-shares or Hong Kong. The valuations are staggering—OpenAI at 8520 RMB, Anthropic at 9650 RMB, DeepSeek at 710 RMB (pre-money). These numbers mirror the DeFi summer of 2020, when every new protocol was worth billions before a single user swapped tokens.
What the article fails to mention—and what any serious analyst should flag—is the complete absence of financial fundamentals. No revenue, no EBITDA, no client concentration data. Just a valuation tag and a press release. This is a soft launch of a narrative, not due diligence.
Core: The Narrative Dissonance
Let's break it down by the numbers that matter.
First, the valuation-to-funding ratio. OpenAI has raised 1800 RMB (cumulative) and is valued at 8520 RMB—a 4.7x multiple. Anthropic has raised 1320 RMB and is valued at 9650 RMB—a 7.3x multiple. The market is giving Anthropic a higher growth premium despite OpenAI's revenue dominance. Why? Because the narrative says "safety-first" is the next big premium, not capability. This is a sentiment-reality dissonance. I've seen this before in DeFi: the "safety narrative" for L2s that were actually centralized sequencers.
Second, the China discount. DeepSeek raised 70 RMB and is valued at 710 RMB—a 10x multiple. That is higher than both US firms on a funding-to-value ratio, but lower on absolute dollar basis. The market is pricing in both technical inferiority and regulatory risk. But here's the hidden signal: the discount is too shallow. If China's chip export controls tighten, DeepSeek's model training costs could spike 300% overnight. The narrative does not price that tail risk.
Third, Perplexity. 210 RMB valuation on a search tool that competes with Google. Its current funding round is only 200 million RMB—a token round. This suggests the company is not desperate for cash, which could mean it is cash-flow positive. But if it is, why is it raising at all? The answer: to pay for the IPO narrative. The tether is thin.
Contrarian Angle: The Real Leak Is in the Business Model
Every investor is looking at the IPO timeline. They are asking: when can I buy? But the real question is: will the business survive the transition from private to public market scrutiny?
Look at the Chinese firms. DeepSeek is open-source, Moonshot has a paid chatbot, Baichuan is B2B, StepStar is also B2B. None of them have disclosed their unit economics. In AI, the cost per inference is the hidden killer. If inference costs are above API revenue, the company burns cash on every token. This is the same as writing an immortal smart contract with infinite gas consumption.
In my 2022 LUNA investigation, I saw the same pattern: UST depeg was mathematical, but the narrative said "it will hold." Three days before the crash, on-chain data showed the velocity of capital was collapsing. The same is happening here. The IPO narrative is a story of infinite growth, but the actual on-chain metric—call it revenue per GPU hour—is likely negative.
Takeaway: The Next Narrative Isn't IPO Dates—It's Earnings Reality
By 2028, the market will shift from "who is going public" to "who is profitable." The companies that survive will be those with strong unit economics, not just strong narratives. The contrarian play is to short the narrative now and long the earnings later.
As I've said before, "We hunt the signal in the noise of consensus." The signal here is the absence of data. The noise is the valuation.
Tracing the code back to the source of the leak. The leak is not in the IPO date. The leak is in the price that assumes profitability without proof. The tether will snap when the first S-1 filing reveals a burn rate that scares institutional investors. That will be the inflection point.
Watching the tether snap, not just the price drop. The price drop will come only after the tether breaks. But the tether is already cracked. The narrative is the only asset that doesn't depreciate—until it does.