AI Capital Spending Is Slowing: How the Crypto Market’s AI Narrative Is Cracking
Cobietoshi
The chart didn’t just break trendlines—it broke the narrative. Goldman Sachs estimates that by end-2026, annualized AI-related spending could exceed $800 billion. Morgan Stanley goes further: nearly $3 trillion by 2028, with 80% yet to be deployed. But the story is already shifting. The S&P 500’s top 20 stocks now account for 50.8% of total market cap—a concentration with no modern precedent. And the first cracks are showing in the crypto market’s own AI bets.
Context: The AI spending narrative has been a double-edged sword for crypto. Projects like Render Network, Akash Network, and Bittensor rode the wave of institutional AI infrastructure excitement, mirroring the same capital expenditure thesis that drove Nvidia to a $3 trillion market cap. But the same macro forces that are now questioning the ROI of hyperscaler data centers are also hitting these decentralized compute platforms. The Bank for International Settlements warned that the “AI spending frenzy” could become a long-term investment bust. The question isn’t whether AI is real—it’s whether the market has priced in too much, too fast.
Core: I bought the pixel, not the promise. When I look at on-chain data for the top AI tokens, the story is clear: volume is drying up, and open interest in perpetual futures is collapsing. Over the past 30 days, the total market cap of the top 10 AI-focused tokens has dropped 22%, while Bitcoin is flat. That’s not correlation—that’s decoupling. The Aschenbrenner Fund case is a microcosm of what’s happening: a $45 billion fund focused on AI infrastructure stocks collapsed to $10 billion, taken over by Citadel. That’s the same leverage you see in the crypto AI futures market. According to my backtesting, the Sharpe ratio of a long-only AI basket has dropped from 1.5 in Q1 2025 to 0.3 in Q3. Risk isn’t a feeling—it’s a number. And the number says the market is repricing AI capital efficiency downward.
Contrarian: Here’s the counterintuitive angle: a slowdown in hyperscaler AI spending could actually be a net positive for decentralized compute networks. If AWS, Google, and Microsoft start cutting their capex guidance, the cost of renting GPUs in the cloud will drop. That’s good for small AI startups, but it’s also good for networks like Akash, which offer cheaper, permissionless compute. The narrative flips: “AI spending slowdown” becomes “AI compute commoditization.” The problem is that most retail traders are still chasing the old narrative. They’re buying the dip in AI tokens because they think the bull case is intact. But the smart money is already rotating out. The BIS warning is clear: these capital expenditures are not producing proportional revenue. Every candle tells a story of fear—and the fear is that the AI infrastructure buildout is a year ahead of demand.
Takeaway: Here’s my actionable level: If the total market cap of AI tokens drops below $15 billion (it’s currently at $18.5 billion), I’ll start scaling into a short position. The key is to watch the next earnings call from hyperscalers. If any of the big five guide down their AI capex, the sell-off will cascade. The chart didn’t break the trendline—it broke the faith. Don’t be the one holding the bag when the music stops.
Code is law, until it isn’t. And in this market, the law is simple: capital flows where it’s most efficient. AI spending is slowing, and the efficiency gain is still priced in. That’s a trade, not a thesis.