The AI Capital Spending Slowdown: A Centralization Crisis That Crypto Has Seen Before

We audit the code, but who audits the conscience? Last month, the S&P 500’s top 20 stocks swallowed 50.8% of total market capitalization—a concentration JPMorgan called “without modern precedent.” This isn’t just a statistic; it’s a warning siren for any investor who has sat through a DeFi liquidation cascade. The same pattern that collapsed Terra’s UST peg is now playing out in the world’s largest equity index, only this time the collateral is AI infrastructure, not algorithmic stablecoins. Over the past seven days, I’ve been tracing the fallout from BeInCrypto’s analysis of “AI Spending is Slowing Down.” The report reveals a market that has become a leveraged bet on a single narrative: that trillions of dollars in capital expenditure (capex) will eventually yield equivalent revenue. Goldman Sachs estimates that annualized AI-related spending could exceed $800 billion by the end of 2026. Morgan Stanley pushes that figure to nearly $3 trillion by 2028, with over 80% yet to be deployed. But here’s the uncomfortable truth I learned during my 2020 audit of Harvest Finance’s yield optimization: when growth is driven by unsustainable token emissions, the music stops the moment the market questions the math. The core of the issue lies in the quality of that spending. Mac10, a quantitative analyst cited in the report, argues that record forward earnings growth is largely a “one-time event” as companies push unprecedented cash through their income statements to fund AI. This is not organic operational improvement—it’s a front-loaded investment that creates a phantom profit peak. Meanwhile, the Bank for International Settlements (BIS) has warned that the “spending frenzy” may turn into a “long-term investment bust.” Sound familiar? I saw the same pattern in 2021, when NFT platforms raised millions on promises of artistic revolution, only to deliver liquidity crunches and rug pulls. The difference is that AI capex is orders of magnitude larger, and the institutions deploying it are the same ones that crashed in 2008. But let’s dig deeper into the contrarian angle. The report’s most striking data point is the Aschenbrenner fund implosion: a $45 billion AI-focused hedge fund that collapsed to roughly $10 billion before Citadel took over. The fund’s manager, a former OpenAI researcher, used heavy leverage and concentrated bets on AI infrastructure stocks. This is the same behavior I analyzed in 2017 when I audited the 1Balance DAO’s governance model—centralized risk hidden behind a veneer of technical sophistication. The market is now asking: if the “insiders” lost their shirts, what hope does the average retail holder have? The BIS warning about a “credit event” triggered by Big Tech AI spending is eerily similar to the credit event that felled Three Arrows Capital in 2022. Yet, there is a counter-narrative that the report underplays. BlackRock argues that current AI leaders generate real profits and strong balance sheets, funding most capex from internal cash flows. This is not entirely wrong—Microsoft and Nvidia are not Enron. But as I wrote in my “Quiet Chain” newsletter during the 2022 bear market, resilience in downturns comes from distribution, not concentration. The S&P 500’s historic concentration means that any AI spending miss will reverberate across the entire index, magnifying losses by a factor of four or five compared to a more diversified market. The blockchain ethos—decentralization, permissionless access, and transparent ledgers—offers a structural alternative. Protocols like Filecoin (storage) and Render (compute) are building decentralized infrastructure that distributes risk across thousands of nodes, not a handful of hyperscalers. Build not for the peak, but for the plain. The AI capex slowdown is not a death knell for technology; it’s a recalibration of expectations. The real question is whether the market will learn from its centralized mistakes. In my experience auditing DeFi protocols, the ones that survived the 2022 crash were those with sustainable tokenomics and community governance, not those that chased the highest yields. Similarly, the AI infrastructure that will endure is not the one that spends the most today, but the one that builds for the long tail of use cases—edge computing, privacy-preserving inference, and open-source models. The blockchain community has already proven that decentralized networks can scale without sacrificing resilience. The next cycle will test whether traditional markets can do the same. History doesn’t repeat, but it often rhymes. The S&P 500’s concentration is a bug, not a feature. As we audit the code of the AI boom, we must also audit the conscience of the institutions driving it. The technology is real; the question is whether the financial architecture built around it is sound. I suspect the answer will be found not in the boardrooms of Wall Street, but in the public, transparent, and auditable networks that blockchain has pioneered.