Chasing ghosts in the digital art auction house. That’s what most traders are doing when they treat AI and crypto as separate beasts. But the real signal is in the rotation. Dhaval Joshi, chief strategist at BCA Research, dropped a bombshell that should echo through every crypto portfolio: the AI bubble is not a single explosive event. It’s a rolling cascade of mini-bubbles, each one inflating, deflating, and passing the baton to the next layer of the tech stack. For crypto, this is not noise—it’s a liquidity map.
Context: Why This Matters Now
We’re in a bull market, and the euphoria is blinding. Everyone’s chasing the next AI narrative—Nvidia’s data center GPUs, OpenAI’s valuation, Palantir’s government contracts. But Joshi’s framework, reported by Crypto Briefing, flips the script. He argues that capital misallocation is the real risk, not a sudden crash. The AI bubble is rolling through infrastructure (chips, cloud), then models (LLMs), then tools, then applications. Each layer gets its moment of absurd valuation before the next takes over. For crypto, this means two things: first, the AI bubble’s rot is localized, not systemic; second, when the AI hype cycle exhausts a layer, that capital doesn’t vanish—it migrates. And crypto is the natural orphanage for displaced speculative liquidity.
Core: The Rotating Structure and Crypto’s Hidden Leverage
Joshi’s thesis rests on a simple observation: AI capital expenditure has exploded, but revenue hasn’t kept pace. Microsoft, Google, Amazon, Meta—combined CAPEX exceeded $200B in 2024, while AI-related revenue growth, though strong, remains a fraction of that outlay. This is the textbook definition of capital misallocation. But Joshi claims it’s not a single bubble because the tech stack is layered, and each layer’s hype cycle is asynchronous. Infrastructure is overheated now; models will cool next; applications will get their turn. The key insight for crypto: rolling bubbles defer the crash but stretch the risk horizon. They don’t eliminate the eventual reckoning—they just make it a series of smaller, staggered corrections.
Here’s where crypto enters the picture. When the AI infrastructure bubble peaks and capital rotates out of GPU hyperscalers, that money doesn’t just sit in cash. It seeks new narratives. Crypto, with its own layered structure (L1s, L2s, DeFi, AI agents), is the perfect recipient. Already, we see AI-crypto convergence: decentralized compute networks like Akash Network, AI agents executing on-chain trades, and BRC-20 tokens trying to piggyback on Bitcoin’s security. But Joshi’s model warns: if the capital rotation is itself a bubble, crypto will simply be the next stop on the rollercoaster. The same misallocation that inflated GPU prices will inflate AI-token valuations, then collapse when the next rotation hits.
Based on my audit experience during the 2021 DeFi liquidity crisis, I’ve seen this pattern before. When Terra/Luna collapsed, liquidity didn’t evaporate—it migrated to other protocols, temporarily masking the systemic risk. The same is happening now. The AI bubble is a “deferred risk” vehicle. Crypto is not immune; it’s the next holder of the hot potato. The question is: when will the rotation stop, and who will be left holding the bag?
Contrarian: The Unreported Blind Spot—Crypto as the Final Bubble Layer
Most analysts view AI and crypto as separate risk assets. Joshi’s framework implies they are connected by a capital rotation chain. The contrarian angle is this: crypto might be the final layer of the rolling AI bubble, not a separate asset class. If the AI bubble rotates through infrastructure → models → tools → applications, and then runs out of narrative, the only remaining “new thing” is decentralized AI or crypto-native AI agents. That would make crypto the ultimate sink for misallocated capital before the entire stack collapses. The implication is stark: if you’re long AI tokens, you’re not diversifying—you’re just buying the last act of the same play.
Volume is the only truth the market respects. Look at the wash trading data from the NFT bubble in 2021—70% of Bored Ape volume was fabricated. The same pattern is emerging in AI-token trades. The “rolling bubble” narrative gives cover to pump-and-dump cycles disguised as rotation. When the faucet runs dry, the dryers crack. The market’s attention will shift, and liquidity will drain from the current AI darling, leaving only the true believers holding tokens with no real utility.
Takeaway: What to Watch Next
Joshi’s framework offers a roadmap, not a crystal ball. For crypto investors, the next signal is the peak of AI infrastructure spending. If Nvidia’s data center revenue growth slows, expect capital to rotate into AI model tokens (like those tied to OpenAI or Anthropic). That rotation will be the first domino. The real test comes when the rotation reaches crypto-native AI projects. If those tokens can sustain real on-chain activity (not just speculation), the rolling bubble might actually produce lasting value. If not, it’s just another ghost in the digital art auction house.
Leading the charge when the herd turns away. That’s the play. Monitor the AI infrastructure-to-application rotation, and position crypto as a hedge against the final, messy unwind. But remember: a rolling bubble still leaves a stain.