I watched fortunes bloom and wither in real-time last week as the market digested a shift that most retail investors haven't fully registered yet. The narrative around AI stocks has fundamentally changed, and a new report from CITIC Securities just codified what many of us in the trenches have been feeling for months: we've entered the era of execution over imagination.
The report's core thesis is deceptively simple. It argues that the recent tech stock correction isn't primarily about Treasury yields or macro headwinds. Instead, it's about three internal industry variables: the pace of commercialization, the efficiency of compute conversion into market share, and the evolution of model gaps. And lurking beneath all three is a wildcard that could reshape the entire competitive landscape โ the concept of "anti-distillation."
Let me break down why this matters right now, because the timing is everything. We're at a critical inflection point where AI is transitioning from technical validation to่งๆจกๅ monetization. The market's sensitivity to commercialization speed has already exceeded its focus on raw model capability. I've seen this pattern before โ in 2021 with NFTs, in DeFi Summer, and now in the AI trade. The pattern is always the same: early believers get rewarded, late adopters get punished, and the repricing happens faster than anyone expects.
The Commercialization Gap
Here's the uncomfortable truth that the report dances around but doesn't fully articulate: there's a dangerous time mismatch between AI companies' technology investment curves and their revenue realization curves. The investment curve keeps climbing steeply, but the revenue curve hasn't hit its exponential inflection point yet. The capital markets are now repricing this mismatch in real-time.
OpenAI reportedly crossed $4 billion in annualized revenue, but inference costs remain stubbornly high. Anthropic's revenue is growing fast, but gross margins are under pressure. The industry is still in a "revenue for market share" phase, and unit economics haven't been validated. This is exactly what I warned about during the DeFi liquidity mining era โ when incentives stop, real users vanish. The same logic applies here: when the narrative stops, real revenue needs to appear.
The report correctly identifies that market expectations have shifted from "technical leadership equals commercial success" to "verifiable customer retention and willingness to pay." Microsoft's Copilot penetration debates and Salesforce's Einstein GPT adoption rates are case studies in this shift. Enterprise AI budgets are growing, but deployment is slower than early optimistic projections suggested.
What the report doesn't emphasize enough is the pricing power problem. Current AI service pricing models are still primarily cost-plus โ per token, per seat. We haven't seen mature value-based pricing emerge. This means AI companies haven't established pricing power directly tied to customer value creation. The quality of commercialization remains unproven.
The Compute Moat and the Distillation Threat
The report's transmission chain โ compute advantage to market share to model gap โ reveals the industry's most critical competitive logic: compute is the moat, and the moat is pricing power. This logic is reshaping value distribution across the AI supply chain. Infrastructure providers like GPU manufacturers and cloud services are gaining bargaining power, while model and application layers face margin compression from both directions.
But here's where the report gets truly interesting. The introduction of "anti-distillation" suggests the competitive landscape is shifting from a "model capability race" to a "data and knowledge asset protection" phase. This will profoundly affect AI technology diffusion paths and industry structure.
Based on my audit experience across multiple blockchain protocols, I've seen how technical moats can be weaponized. Anti-distillation โ whether through output watermarking or API usage restrictions โ could sever the "catch-up path" for smaller AI companies. If successful, the industry could accelerate from "a hundred flowers blooming" to "oligopoly." The report flags this as the "largest potential variable," and I agree, but I'd go further.
The Hidden China Question
The report's discussion of model gaps implicitly carries deep concern about China's AI industry. Under compute restrictions, will the US-China AI model gap widen? The positioning of anti-distillation as the biggest variable suggests the report's authors are deeply worried about this risk. If model gaps solidify due to anti-distillation, AI innovation diffusion will slow significantly โ and this hits China's AI industry, which relies on "open source plus distillation" paths, particularly hard.
But here's the contrarian angle that most analysts are missing: compute advantage itself doesn't directly create value. It only converts to commercial value through productization, distribution channels, and service systems. This explains why Google has top-tier compute but its AI commercialization lags OpenAI. Compute is necessary but not sufficient. The report hints at this but doesn't fully develop the implication.
The K-shaped Divergence Trade
The report's mention of "K-shaped divergence convergence" contains a trading signal that most readers will miss. Dollar weakness and reduced rate hike expectations could trigger capital rebalancing from US AI leaders to other markets, including A-shares. But this rebalancing's sustainability depends on whether AI industry fundamentals support valuation convergence.
This is where the "avoid excessive grand narratives" advice becomes crucial. The market's AI expectations are already loaded with grand narrative components โ AGI approaching, productivity revolution, and so on. Once these narratives fail to materialize as concrete business results, valuation correction risks amplify significantly. I've watched this movie before. It never ends well for those holding only narrative exposure.
The Repricing Has Begun
The report's core contribution is shifting the attribution framework for tech stock adjustments from external macro factors to internal industry variables. The implicit investment logic is that AI stocks have entered a "expectation verification period." Valuations will depend more on verifiable industry progress than macro liquidity. This means investment strategy needs to shift from beta-driven sector allocation to alpha-driven stock selection.
In 2023, AI stock valuations were anchored to "technology breakthrough expectations." Since 2024, the anchor has shifted to "commercialization realization." This shift has significantly reduced valuation tolerance for AI companies with only technology narratives and no commercial validation.
The report argues that Treasury yields aren't the root cause of tech stock adjustments. This is really saying: even if the interest rate environment improves, AI stocks lacking commercial validation won't easily recover their valuations. This judgment moves investment focus from macro trading to industry fundamentals.
What I'm Watching Now
Speed is survival, but empathy is the signal. In the next 2-3 quarters, I'm tracking three specific signals. First, the commercialization data in major AI companies' quarterly reports โ revenue growth, gross margins, customer retention rates. Second, whether leading model vendors introduce anti-distillation technical measures or clause changes. Third, GPU supply bottleneck relief progress.
The report's top risk is AI commercialization continuing to disappoint, triggering a systematic valuation downgrade as the market shifts from PS multiples to PE logic. I'd add that the "anti-distillation" scenario could solidify the industry structure, cutting off catch-up paths for smaller players and accelerating market concentration.
But here's what keeps me up at night: the report doesn't answer whether anti-distillation is technically feasible. Are there actually implemented solutions? If model gaps solidify, what structural impacts will this have on value distribution across the AI supply chain? And can China's AI industry, under compute constraints, find alternative paths around the "compute-to-model gap" transmission chain?
The Takeaway
The code didn't change, but the market's interpretation of it did. We've entered the verification phase where the market pays for execution, not imagination. The AI trade is no longer a simple beta play. It's a complex, multi-dimensional game where compute reserves, commercialization execution, and ecosystem lock-in effects collectively form competitive barriers.
Stability isn't the absence of change โ it's the ability to adapt when the repricing comes. The question isn't whether AI stocks will recover. The question is which ones deserve to. And that answer will be written in revenue reports, not press releases.
The next 6-18 months will separate the companies that built real businesses from those that built beautiful narratives. I've seen this before. The survivors won't be the ones with the best technology. They'll be the ones with the best unit economics, the strongest customer retention, and the ability to convert compute into cash. Everything else is just noise.
Watch the quarterly reports. Watch the gross margins. Watch the customer churn. The signal is in the numbers, not the narratives. And if you're not watching, you're already behind.