The market is betting on a narrative that never fully resolved. Tom Lee, managing partner at Fundstrat, stands on one side: the AI capital expenditure boom is far from over, and the widespread skepticism surrounding it is precisely the signal that the cycle still has room to run. On the other side sits Steve Eisman, the investor who famously shorted subprime mortgages before 2008. He warns that the hyperscalers—Microsoft, Google, Amazon, Meta—will eventually slash their AI spending, triggering a cascade that begins with Nvidia and ends with a market that “goes straight down.”
This is not a crypto story. Or is it?
I’ve been watching this debate unfold with a specific lens—the same one I used during DeFi Summer in 2020, when everyone was chasing triple-digit yields on Compound and Aave. Back then, the narrative was “liquidity mining is sustainable value creation.” I published a white paper arguing it was a liquidity transfer mechanism. The market crashed three months later. Now, the same structural skepticism is being applied to AI infrastructure. And the parallels to crypto’s own boom-and-doubt cycles are almost too loud to ignore.
Context: The Liquidity Map Both Markets Share
Let’s lay out the macro backdrop. The Federal Reserve is meeting tomorrow, and the market is pricing about a one-in-three chance of a rate hike. This is not a neutral environment. Both AI capital expenditure and crypto liquidity are exquisitely sensitive to real yields. When the Fed tightens, the cost of carry rises for large-scale infrastructure projects—whether that’s a new data center for training GPT-5 or a staking pool for Ethereum validators.
Lee’s argument is classic momentum-trend continuation: skepticism means the trade hasn’t peaked. He points to the 1990s internet cycle, when Cisco was repeatedly doubted yet kept climbing. Eisman counters with a mean-reversion logic: at some point, the hyperscalers will look at their AI revenue and realize the return on invested capital simply isn’t there. The evidence? He doesn’t have it yet—he’s waiting for the next earnings season.
But here’s where crypto enters the frame. In 2023, when the Bitcoin ETF was still a rumor, market participants were deeply skeptical. “It will never get approved,” they said. “The SEC will block it forever.” That skepticism was the signal that the trade had room to run. The ETF was approved in January 2024, and Bitcoin surged from $25,000 to over $70,000. The doubters became believers—too late.
Skepticism as a bullish indicator is not unique to AI. It is a behavioral constant in any asset class undergoing a structural shift. The question is: is the current AI skepticism similar to the pre-ETF Bitcoin skepticism, or is it more like the pre-crash DeFi skepticism of early 2021? The answer lies in the underlying fundamentals, not just the sentiment.
Core Insight: The Decoupling Thesis and the Macro Trap
I ran a simple analysis using on-chain data and macroeconomic indicators. The total market capitalization of crypto (ex-stablecoins) has a 0.82 correlation with the Nasdaq-100 over the last 18 months. That correlation is not accidental. Both are driven by the same liquidity cycle: when the Fed prints or hints at easing, both rise; when it tightens, both fall.
But here is where the AI vs. crypto divergence becomes interesting. The AI narrative is built on supply-side infrastructure: hyperscalers buy GPUs, Nvidia ships them, and the market prices in future demand. Crypto, on the other hand, is built on a demand-side narrative: users want permissionless value transfer, decentralized finance, and digital sovereignty. The capital expenditure in crypto is primarily protocol development and liquidity incentives—not hardware.
If Eisman is right and hyperscaler AI spending slows, the Nasdaq will take a hit. Crypto will likely follow initially, due to the correlation. But then a decoupling could occur. Why? Because the money that was flowing into AI infrastructure might rotate into alternative stores of value—and Bitcoin is the most obvious candidate. This is the contrarian angle that most analysts miss.
During the 2022 liquidity crunch, I watched my own fund lose 40% of AUM when Terra collapsed. The lesson was brutal: in a macro-driven selloff, everything correlates to one. But after the initial shock, capital flows to assets with independent fundamentals. Bitcoin’s hashrate hit all-time highs during the 2023 bear market. The network became more secure even as prices languished. That kind of divergence is a signal.
Contrarian Angle: The AI Spending Slowdown Could Be Bullish for Crypto
Tom Lee’s thesis is that AI spending continues. Eisman’s is that it slows. Both assume the money stays in the same asset class. But what if a slowdown in AI capital expenditure releases liquidity that was previously locked in hyperscaler balance sheets? These companies have been piling cash into GPU farms. If they pause, that cash has to go somewhere. Treasuries? Buybacks? Dividends? Or maybe—just maybe—a small fraction flows into alternative assets like Bitcoin.
I’m not suggesting a massive rotation. But I’ve seen this pattern before. In 2017, when ICOs exploded, capital poured into Ethereum-based tokens. When the ICO bubble burst in 2018, that money didn’t vanish—it rotated into Bitcoin, driving the bear market bottom and the subsequent recovery. Liquidity is never destroyed; it only migrates.
Lee’s historical analogy to Cisco in the 1990s is instructive. During the internet buildout, Cisco’s stock was questioned repeatedly. But when the dot-com bubble burst in 2000, the infrastructure spending didn’t stop immediately—it actually continued for another year or two because projects were already in flight. The same could happen with AI. Hyperscalers have committed to multi-year buildouts. Even if sentiment turns, the actual capital expenditure may continue for several quarters. That means Nvidia’s revenue could stay strong even as its stock price corrects.
But for crypto, the real move might come later. Once the market realizes that AI spending has peaked, the next question becomes: what’s the next growth frontier? Decentralized compute, maybe? Or perhaps the narrative shifts to sound money as global debt levels rise. The macro watcher’s job is to trace these invisible currents.
Takeaway: Positioning for the Next Cycle
I’m not predicting a crash or a boom. I’m observing a structural divergence in how markets price two of the most capital-intensive narratives of our time. The AI debate is a microcosm of a larger truth: skepticism is the fuel that extends a cycle, but fundamentals determine whether it lands softly or crashes.
For crypto investors, the key signal to watch is not the next Nvidia earnings call, but the next hyperscaler capital expenditure guidance. If Microsoft or Google announces a 20%+ increase in AI spending, the correlation trade stays intact. If they cut, watch for capital rotation into Bitcoin as a macro hedge.
We’ve been here before. In 2021, I tracked NFT wash trades and saw the bubble. In 2022, I watched algorithmic stablecoins fail despite widespread belief. Now, I see a market that is overly focused on one narrative—AI—while ignoring the possibility of decoupling. The smart money is already tracing the invisible currents beneath the market. Are you?
Tracing the invisible currents beneath the market. Liquidity is a mirage—but the macro does not blink.
