The Covenant of Value: When the Market Stopped Believing in Magic
There is a quiet shift happening in the marble halls of capital allocation. It is not announced with the thunder of a rate cut or the whisper of a new all-time high. It is happening in the silence between quarterly earnings calls, in the pause before an analyst asks about customer retention. Over the past several weeks, I have watched the market's attention drift from the macro to the micro, from the color of the tide to the shape of the boat. The CITIC Securities report on AI stock adjustments is not just another sell-side document; it is a confession. The market is no longer paying for imagination. It is paying for execution.
For years, we in the crypto world have lived this truth. We called it 'the flippening' or 'the merge,' but the underlying principle was always the same: narrative without throughput is just a prayer. The report's core thesis—that AI pricing has shifted from a macro-driven model to an industry-fundamentals-driven one—resonates with a pattern I have seen in every cycle since 2017. The ICO whitepaper that promised a decentralized utopia but delivered a token with no utility was priced on hope. The AI company that promises AGI but delivers a chatbot with declining engagement is priced on the same hope. The market has finally learned to ask the question we should have been asking all along: where is the value, and who is paying for it?
My code was the covenant, not just the contract. This is the lens through which I read the report's three pricing variables: commercialization pace, compute conversion efficiency, and model gap evolution. The first variable is the most tangible. The report correctly identifies that AI revenue growth is still driven by new customer acquisition rather than deep monetization of existing users. OpenAI's annualized revenue crossing $4 billion sounds impressive until you measure it against the inference costs. Anthropic's growth is real, but gross margins are under pressure. This is the 'revenue for market share' phase, and it is unsustainable by definition. In DeFi, we called this 'liquidity mining APY'—the project subsidizing TVL numbers. Stop the incentives, and the real users vanish. The same principle applies here. The market is beginning to understand that a customer acquired through aggressive pricing is not a customer; it is a rental.
The second variable, compute conversion, is where my technical background forces me to pause. The report frames compute as a moat, and it is. But I have audited enough smart contracts to know that a moat without a drawbridge is just a hole in the ground. Compute advantage only converts to market share through productization, distribution, and service. Google has the best compute infrastructure in the world, yet its AI commercialization lags OpenAI. Why? Because compute is a necessary condition, not a sufficient one. The report's implicit acknowledgment of this—that 'compute advantage does not directly create value'—is a subtle but profound admission. It is the same lesson we learned in the Layer 2 wars: the Data Availability layer is overhyped. 99% of rollups don't generate enough data to need dedicated DA. The technology is necessary, but the application is what matters.
The third variable, the model gap, is where the report introduces its most controversial concept: 'anti-distillation.' This is the idea that leading model makers will use technical means—output watermarking, API usage restrictions—to prevent competitors from training on their outputs. The report calls this the 'largest potential variable,' and I believe this is an understatement. Anti-distillation is not a technical feature; it is a declaration of war on the open innovation model. It is the equivalent of a DeFi protocol forking itself and then suing anyone who uses the fork. The implications are staggering. If anti-distillation succeeds, the 'catch-up path' for smaller AI companies is severed. They cannot stand on the shoulders of giants because the giants have installed anti-climb spikes. The industry would accelerate from 'a hundred flowers blooming' to 'oligopoly.'
In the silence of the bear, we heard the truth. This is what the report is really saying, even if it does not use those words. The bear market of 2022 taught me that the market's patience is finite. The 'patience window' for AI commercialization is closing. If the leading players cannot deliver above-expectation commercial data in the next two to three quarters, the valuation system will shift from PS multiples to PE logic. This is not a technical adjustment; it is a philosophical one. It is the market saying, 'I no longer believe in your potential. Show me your present.'
But here is the contrarian angle that the report misses. The report assumes that anti-distillation is a rational strategy for the incumbents. I am not so sure. In my experience building 'The Commons,' a community for ethical Web3 builders, I have seen the power of open systems. The reason Ethereum survived the bear market while so many 'Ethereum killers' did not is not because of superior technology. It is because the open, permissionless nature of the network created a resilience that closed systems cannot replicate. If the AI incumbents succeed in closing their systems, they may win the battle for model supremacy but lose the war for ecosystem dominance. The open-source models—Llama, Qwen, Mistral—are not just competitors; they are the seeds of a future the incumbents cannot control.
The report's analysis of the 'K-shaped divergence' and its potential convergence is also worth examining. The suggestion that a weaker dollar and reduced rate hike expectations could trigger a rebalancing of capital from US AI leaders to other markets, including A-shares, is a trading signal. But it is also a values signal. It suggests that the market is beginning to recognize that AI value is not geographically concentrated. The compute advantage may be in the US, but the application of AI is global. The report's bias assessment is honest about its own limitations—it is a sell-side document with potential conflicts of interest. But the framework it provides is useful, even if the conclusions are incomplete.
Every broken token taught me how to hold value. This is the lesson of the current market. The AI stocks that will survive the 'expectation verification period' are not necessarily the ones with the best models or the most compute. They are the ones with the clearest path to monetization, the most efficient use of resources, and the strongest customer retention. The report identifies three variables, but I would add a fourth: resilience. The ability to survive the inevitable disappointments, the missed quarters, the delayed product launches. This is not a technical metric; it is a cultural one. It is the difference between a project that pivots when the market shifts and one that doubles down on a failing narrative.
As I read the report's final recommendations, I am struck by the absence of one word: trust. The report talks about commercialization, compute, and model gaps, but it never mentions the fundamental issue that underpins all of them. Trust is compiled, not claimed. The market is not just looking for execution; it is looking for integrity. It is looking for companies that do not overpromise, that do not hide their costs, that do not use 'narrative' as a substitute for substance. This is the covenant that the market is demanding. And it is the same covenant that we in the Web3 space have been trying to build for years.
The question is not whether the AI market will correct. It is whether the correction will be a cleansing or a collapse. The report suggests it will be a cleansing—a shift from 'paying for imagination' to 'paying for execution.' I hope it is right. But I also know that the market is a fickle god, and its patience is measured in quarters, not decades. The AI industry has a window to prove itself. The question is whether it will use that window to build something lasting, or whether it will squander it on the same hubris that has felled every empire before it. The silence of the bear is not a warning. It is an invitation. The question is whether we are ready to listen.