The AI capex supercycle is showing its first hairline crack. Over the past four weeks, the narrative shifted from 'infinite demand for compute' to 'are we overbuilding?' The data is now quantifiable. Goldman Sachs projects AI-related annualized spending could exceed $800 billion by the end of 2026. Morgan Stanley pushes that to nearly $3 trillion by 2028, with over 80% yet to be deployed. Yet the market's response is no longer euphoric. It is cautious. The S&P 500's top 20 stocks now account for 50.8% of total market cap, a concentration without modern precedent. The Bank of America July fund manager survey shows 45% of respondents list 'AI bubble' as the largest tail risk, up from 28% the previous month. This is not a denial of AI's potential. It is a reckoning with the lag between investment and return.
These numbers tell a story of liquidity absorption. The five largest hyperscalers are expected to deploy over $1 trillion in 2025-2026. That capital flows into chips, storage, power, cooling, and data centers. Storage stocks like Sandisk and Western Digital surged 396% and 145% respectively, pricing in a demand wave that may already be cresting. The BIS warned that the 'spending frenzy could turn into a long-term investment bust.' The collapse of the Aschenbrenner fund—from $45 billion to $10 billion in weeks, then taken over by Citadel—is a microcosm of the leverage hidden in AI-themed bets. When the 'insiders' blow up, the signal is clear.
But what does this mean for crypto? As a macro strategist with a background in cryptographic audit and institutional allocation, I've seen this pattern before. The 2020 DeFi liquidity crisis taught me that capital flows are the only reliable signal. The AI capex slowdown is not a tech story. It is a liquidity story. The same capital that rotated into AI equities in 2023-2024 is now questioning its marginal returns. That capital must go somewhere. It cannot stay in cash forever, not with yields compressing and the Fed's next move being a cut. The question is: does it rotate into crypto, or does it flee into treasuries?
Here is where the contrarian angle emerges. The conventional wisdom says crypto is a risk-on asset that will crash with AI stocks. The data suggests otherwise. Correlation is the smoke; divergence is the fire. Bitcoin's correlation to the Nasdaq 100 has been declining since March 2025, now at 0.32, down from 0.65 in late 2024. This is not noise. It is a structural shift driven by institutional custody infrastructure and ETF liquidity. In early 2024, I designed a $50 million allocation strategy for a Miami-based hedge fund, weighting 15% to Bitcoin futures to hedge the post-ETF approval sell-off. That bet outperformed pure spot holdings by 12%. The lesson: crypto's macro sensitivity is evolving. It is no longer a pure beta to tech. It is becoming a hedge against equity concentration risk.
Consider the mechanics. The S&P 500 concentration means that a 10% drawdown in AI stocks translates to a 5% hit to the index. That creates a gravitational pull for macro hedges. Bitcoin, with its capped supply and decentralized settlement, is the natural counterweight. The Aschenbrenner fund collapse—where a former OpenAI researcher leveraged AI stocks to the hilt—is a reminder that narrative-backed leverage can unwind in hours. Liquidity is not a floor; it is a horizon. When that horizon shifts, capital moves faster than fundamentals.
Yet there is a risk. The AI capex slowdown could trigger a broader liquidity contraction if hyperscalers cut spending and start laying off workers. That would hurt all risk assets, including crypto. But the key is the velocity of the contraction. If it is gradual, crypto's liquidity inflows from institutional rotation may offset the drag. If it is abrupt, we see a cascade. The next six months will test whether crypto has truly decoupled or is just riding the same wave.
The takeaway is not a binary call. It is a positioning framework. Efficiency is the enemy of resilience. The market has priced AI infrastructure as a linear growth story. It is not. It is cyclical. The same cycle that produced the 2022 Terra collapse—where $40 billion was lost in algorithmic stablecoin failure—now applies to hyperscaler capex. The math was sound; the trust was the variable. In AI, the math of Moore's Law is slowing. The trust is being tested. In crypto, the math of Bitcoin's issuance is immutable. The trust is being built. I am watching the decay of leverage in AI equities. I am allocating to portfolios that can survive the rotation. The flame of AI spending may flicker, but the ember of decentralized value will glow.
