The AI Earnings Mirage: BlackRock’s Equity Bias and the Structural Blind Spots Beneath the Hype

The hype is a lagging indicator. BlackRock’s global chief investment strategist, Wei Li, recently stated that AI-driven earnings growth will reshape investment strategy, tilting the scales in favor of US equities over government bonds. The statement made headlines. The logic, however, deserves a forensic audit. As a researcher who has spent years dissecting cross-border payment rails and tokenomics, I find the assertion structurally incomplete. It rests on a series of unverified assumptions about the quality, breadth, and sustainability of AI-related profits. Liquidity evaporates faster than hype. So does unsupported optimism. Context is critical. BlackRock is not a casual observer in this narrative. The firm manages over $10 trillion in assets. It has aggressively launched spot Bitcoin ETFs, signaling a pragmatic embrace of digital assets, but its core revenue engine remains traditional asset management. When a firm of this scale signals a preference for equities based on an AI earnings thesis, it moves markets. The claim is not merely analytical; it is performative. It shapes portfolio flows, influences retail sentiment, and validates a specific risk-on posture. The timing is also telling. We are in a bear market for crypto, but a bull market for AI narratives. The divergence is instructive. AI has become the new liquidity magnet, absorbing the capital that once flowed into speculative digital assets. The core of my analysis focuses on the actual mechanics of AI earnings growth. Wei Li’s thesis assumes that current AI adoption translates into durable, scalable earnings. The data suggests otherwise. First, the concentration problem. AI-related revenue growth is overwhelmingly concentrated in a handful of companies. Nvidia’s data center business, Microsoft’s Azure AI, and Google Cloud’s AI offerings account for the vast majority of observable AI revenue. The market cap of the “Magnificent Seven” now trades at 30-35 times forward earnings. The broader S&P 500 trades at roughly 21-22 times, well above the historical average of 16-17 times. The equity risk premium, the extra return investors demand for holding stocks over risk-free bonds, has compressed to near-zero. This is not a sign of rational optimism. It is a structural red flag. Second, the quality of earnings is suspect. I audited tokenomics during the 2017 ICO boom. I saw projects raise $50 million on the strength of whitepapers that ignored slippage. The current AI earnings cycle has a similar flavor. A significant portion of AI-driven “revenue” is internal transfer pricing. Cloud providers book AI services consumed by their own AI divisions. Startups like OpenAI and Anthropic are burning cash at unprecedented rates to acquire enterprise customers, often at a loss. The revenue is real, but the profitability is deferred. The market is pricing in a future that may not materialize. Volatility is the fee for entry. But the entry ticket here is overpriced. Third, the capex cycle is a double-edged sword. Hyperscalers are spending over $200 billion annually on AI infrastructure. This spending creates immediate revenue for Nvidia and TSMC. It also creates massive depreciation costs, energy bills, and a looming supply glut. My research on DeFi yield farming taught me a simple lesson: artificially inflated yields decay into value destruction. The same applies to capital expenditure cycles. The build-out phase is profitable for suppliers. The utilization phase is where the value is either realized or destroyed. We have not reached that inflection point. The market is paying peak prices for pre-peak earnings. The contrarian angle is uncomfortable but necessary. The market is treating AI earnings growth as a decoupling event, a new paradigm that justifies higher multiples and lower risk premiums. I have seen this playbook before. In 2020, DeFi yields were the new paradigm. In 2021, NFTs were the new paradigm. In 2022, algorithmic stablecoins were the new paradigm. The pattern is always the same: a narrative-driven repricing, a liquidity injection, a reflexive feedback loop between price and narrative, and then a correction when the fundamentals fail to catch up. Code is law until the wallet is empty. AI earnings are the new code. The question is who holds the wallet. The sustainability of AI earnings growth hinges on two variables the market is ignoring. First, the ROI on enterprise AI spending. Gartner predicts AI budgets will reach 10% of IT spend by 2025. But the realization rate is low. Most enterprise AI deployments remain in the pilot or proof-of-concept phase. The second variable is regulatory risk. The EU AI Act is now in force. Copyright lawsuits against OpenAI and Google are advancing. Data governance rules are tightening. Regulation lags, but penalties lead. The legal costs alone could erode the thin margins of AI application companies. The infrastructure layer will survive. The application layer will be squeezed. The earnings growth Wei Li cites is disproportionately from the infrastructure layer. That is a cyclical bet, not a structural one. My takeaway is a cautionary one, framed as a question. The AI earnings narrative has driven equity markets to record highs. The implied assumption is that earnings growth will outpace the risk-free rate for the foreseeable future. Historical precedent suggests otherwise. Every technological revolution has an investment cycle that peaks before the productivity gains materialize. The railroad boom, the telecom bust, the dot-com collapse. The AI boom will follow the same arc. The question is not whether AI will transform the economy. It will. The question is whether the current prices reflect the transformation or the speculation. Based on my analysis, the prices reflect speculation. The earnings quality is too concentrated, the capex cycle is too front-loaded, and the regulatory environment is too uncertain. The prudent move is not to abandon equities but to question the concentration risk. I am reminded of my 2022 Terra-Luna post-mortem. I spent three weeks reverse-engineering the death spiral. The conclusion was simple: the mechanism was elegant, the economics were broken. The same analysis applies here. The AI narrative is elegant. The economics, at current valuations, are broken. There is a reason I remain skeptical of the “AI will save the stock market” thesis. It ignores the basic accounting principle that earnings must eventually be backed by cash flow. The market is projecting cash flows five to ten years out at discount rates that assume no major disruptions. That is a fragile assumption. Disruptions are not the exception. They are the norm. The path forward is not to short the market or to abandon AI-related investments. The path is to demand better evidence. I want to see enterprise AI renewal rates, not just subscription announcements. I want to see ROI case studies, not just pilot program press releases. I want to see the breakdown of AI-related revenue between internal transfers and external sales. Until that data is available, the AI earnings thesis remains an act of faith, not an act of analysis. And faith, in financial markets, is priced at a premium. That premium is the risk. We are entering a phase where the macroeconomic background will test the AI thesis. Interest rates remain elevated. The US fiscal deficit is expanding. The 10-year Treasury yield is hovering near 4.5%. If AI-driven inflation emerges, or if the deficit continues to widen, yields will rise. Rising yields compress equity multiples. The current equity risk premium of near zero leaves no room for error. The market is effectively saying that equities are as safe as bonds. That is a structural impossibility. Volatility is the fee for entry. The fee is being deferred, not eliminated. I write this not as a bear but as a realist. My background in financial engineering taught me to stress-test assumptions. My experience in crypto taught me that narratives are powerful but ephemeral. My current research on cross-border payments has shown me that infrastructure is the last thing to be commoditized. The AI build-out is real. The AI earnings are real. But the prices are ahead of the fundamentals. This is a classic early-cycle phenomenon. The smart money is not in the narrative. The smart money is in the data. And the data, at this moment, does not support a structural repricing of risk. A final observation, drawn from my 2026 work on AI-agent payment protocols. The most critical vulnerability in any economic model is the fee-burning mechanism. In AI, the fee is the capex. The burn rate is accelerating. The question is whether the revenue side can keep up. For the infrastructure layer, the answer is yes, for now. For the application layer, the answer is unclear. For the broader market, the answer is no. The AI earnings miracle is a concentrated phenomenon. It does not justify a broad equity preference. It justifies a narrow, high-conviction bet on a specific set of companies. And that is a very different strategy from the one Wei Li articulated. The difference matters. It is the difference between investing and speculating. The market has chosen speculation. I have chosen to audit the claims. The audit is ongoing. The findings are not reassuring. The takeaway is simple. AI will reshape the economy. The earnings will follow, but on a different timeline and with a different distribution than the market currently prices. The prudent investor should not abandon equities but should abandon the assumption that the AI thesis is a substitute for fundamental analysis. The thesis is a starting point, not a conclusion. The conclusion requires data, time, and a willingness to accept that volatility is not a bug. It is a feature. And it is the fee for entry.