The AI Rolling Bubble: A Macro Stress Test for Crypto's Next Cycle

You’ve heard it a thousand times: AI is a bubble about to burst. The pundits point to NVIDIA’s eye-watering PE ratio, the vacant office parks of overbuilt data centers, and the endless parade of me-too chatbots. But here is the trap. The data suggests something more insidious: a rolling bubble, not a single collapse. This is not a comforting alternative. It is a structural failure mode that postpones the reckoning, reallocates capital in chaotic waves, and ultimately leaves the entire risk-asset ecosystem—including crypto—far more fragile than the headlines admit. Dhaval Joshi, chief strategist at BCA Research, recently laid out this thesis with a clarity that cuts through the noise. His argument: AI’s valuation mania is not monolithic. It rotates across the technology stack—infrastructure, models, tools, applications—each layer inflating as the previous one cools, creating a sequence of local bubbles rather than one global explosion. This is not a new phenomenon. The 1990s internet boom followed a similar pattern: semiconductor hype, then portal mania, then e-commerce, then optical networking. Each rotation minted new fortunes and buried old ones. The difference today is the scale of global liquidity funneling into a single narrative. In 2024, the four largest cloud hyperscalers alone committed over $200 billion in capital expenditure, most of it to GPU clusters and AI infrastructure. That number dwarfs the total market cap of all crypto assets before the 2021 peak. But here is the macro watcher’s first question: where is this capital coming from, and what is it crowding out? The answer lies in the global liquidity map. Central bank balance sheets, after a brief tightening, are pivoting toward easing. The Fed’s rate cuts, the BOJ’s cautious normalization, and China’s stimulus have created a new wave of cheap money. That money searches for yield, and AI is the current favorite. Yet the rolling bubble structure means that capital is not evenly distributed. It pools in one layer—say, infrastructure—until the next earnings season reveals a slowdown in GPU utilization or a price war in cloud compute. Then the narrative shifts, and the capital rotates to the next layer, leaving behind a trail of overvalued startups and stranded assets. As a macro strategy analyst who has spent years connecting on-chain data to traditional economic indicators, I see a direct parallel to crypto’s own narrative cycles. In 2020, DeFi Summer was the infrastructure layer. Then NFTs became the application layer. Then L2 scaling solutions became the tooling layer. Each rotation created localized bubbles, and each left behind a residue of washed-out projects and disillusioned retail investors. The difference is scale: AI’s capital rotation is an order of magnitude larger, and its macro implications are more profound. Based on my work synthesizing Fed rate hikes with stablecoin supply changes ahead of the Bitcoin ETF approval, I can state with confidence: the AI rolling bubble will drive crypto’s next liquidity cycle, not decouple from it. Let me stress-test this claim. The prevailing narrative among crypto optimists is that AI and crypto are separate asset classes, that a crash in AI will trigger a “great rotation” into crypto as capital seeks alternative high-beta bets. This is the decoupling thesis. It is seductive, but the data says otherwise. Look at the on-chain flows: stablecoin supply has been flat to declining since early 2024, even as AI stocks rocketed. If capital were rotating from AI to crypto, we would see a surge in stablecoin minting. We don’t. Instead, liquidity is being absorbed by AI infrastructure debt—corporate bonds, convertible notes, and private placements. Crypto is not the beneficiary of AI’s excess; it is the fellow traveler in a macro liquidity cycle that is now tilting toward AI. When the rolling bubble finally hits a layer that cannot sustain the rotation—when the application layer fails to deliver ROI, for example—the entire risk-asset complex will feel the withdrawal. Chaos is just data that hasn’t been parsed yet. My contrarian angle is this: the real risk is not that AI crashes and crypto benefits. It is that the rolling bubble transitions into a synchronized collapse. If the macro environment shifts—a sudden spike in real rates, a geopolitical shock, or a corporate earnings miss from a major AI player—the rotation stops. All layers contract simultaneously. The infrastructure debt, the model licensing fees, the application subscriptions—all dependent on cheap capital—unwind. Crypto, already starved for fresh liquidity, will suffer a liquidity crisis worse than 2022. The 2022 bank run forensics I conducted on Celsius and Three Arrows taught me that opaque counterparty risk in a leveraged system always surfaces when the tide goes out. The AI bubble is simply a larger, more opaque leveraged system. What does this mean for a crypto investor positioning for the next cycle? Stop thinking about AI as a separate narrative. Start treating it as a macro variable that dictates the availability of risk capital. The key signal to watch is not AI stock prices, but the on-chain supply of stablecoins. If stablecoin supply begins to rise in tandem with AI infrastructure spending, it means capital is being created ex nihilo—a liquidity injection that will eventually find its way into crypto. If stablecoin supply remains flat while AI capex accelerates, it means capital is being rotated out of crypto. The second scenario is more likely today. The rolling bubble is a time bomb, not a windfall. Failure-mode stress testing is the only honest due diligence. So, where are we in the cycle? Based on BCA Research’s framework, the current rotation is at the infrastructure layer, with NVIDIA and cloud hyperscalers absorbing the bulk of capital. The next rotation will likely hit the model layer, where OpenAI, Anthropic, and others face valuation compression as margins shrink. The final rotation—application layer—will be the most dangerous, because that is where the ROI expectations are highest and the evidence thinnest. When the application layer fails to deliver, the rotation stops. The music ends. The question is not whether, but when. And when the tide turns, on-chain transparency will reveal the cracks faster than any balance sheet. My takeaway is not a recommendation to short AI or go long crypto. It is a call to understand the macro plumbing. The AI rolling bubble is a structural feature of the current liquidity environment, not a bug. It will shape the next crypto cycle more than any halving, any ETF, or any regulatory clarity. The investors who survive will be those who watch the on-chain stablecoin supply, track the Fed’s real rate, and remember that in every rolling bubble, the last rotation is fatal. As I wrote in my 2024 macro synthesis: 'When the liquidity tide recedes, the only thing that remains is the code.'