The AI Rolling Bubble: A Crypto Narrative Hunter's Reading of Capital Misallocation

Hype fades; structure remains. Dhaval Joshi's 'rolling AI bubble' thesis is not merely a warning about artificial intelligence—it is a structural blueprint for understanding how capital flows deform entire technology stacks. As a Web3 Research Partner who has tracked narrative cycles since the ICO boom, I see a familiar pattern: a market that refuses to crash cleanly, instead rotating its excesses from one layer to the next. The question is not whether the bubble exists, but where it lands next—and whether crypto becomes its final destination. Context: The Rolling Bubble Thesis Joshi, chief strategist at BCA Research, argues that AI is not headed for a single, catastrophic burst. Instead, the valuation excess is migrating across the AI stack—from infrastructure (chips, data centers) to models (large language models) to tools (frameworks, middleware) and finally to applications. Each layer experiences its own mini-boom and mini-bust, but the overall market never collapses because a new hotspot ignites just as the old one cools. The hidden risk lies in capital misallocation: money pours into layers that may never generate proportional returns. This is not a theory of denial—it is a theory of delayed reckoning. Based on my audit experience during the 2017 ICO frenzy, I manually evaluated 45 whitepapers and found 38 with zero technical differentiation. The same pattern repeats today. The AI stack's four layers mirror the crypto stack's segmentation: L1 blockchains, L2 scaling solutions, middleware, and DeFi applications. In both worlds, the 'hot' layer attracts disproportionate capital, while the underlying fundamentals lag. The difference is that AI's rolling bubble has a longer time horizon—and a deeper pool of institutional liquidity. Core: The Mechanism of Narrative Rotation Joshi's thesis hinges on a simple observation: market sentiment does not disappear; it relocates. In 2023, the narrative centered on NVIDIA and GPU scarcity. In early 2024, it shifted to OpenAI's valuation and Anthropic's fundraising. By mid-2024, the spotlight moved to AI-powered enterprise software like Palantir. This is not random—it follows a logic of narrative saturation. Once a layer becomes 'priced in,' speculators search for the next untold story. I modeled this behavior during DeFi Summer 2020, when I tracked yield farming strategies across Uniswap and Compound. I discovered that 70% of 'yield' was merely inflationary token rewards, not genuine value accrual. The same dynamic applies to AI: much of the perceived value in AI startups is narrative inflation. The infrastructure layer, for instance, has seen CAPEX from Microsoft, Google, Amazon, and Meta exceed $200 billion annually. Yet the revenue from AI products—while growing—remains far below that spending. The gap is covered by faith in future adoption. This is where crypto's narrative cycle becomes a useful lens. In the blockchain space, we saw capital rotate from L1s (2021) to L2s (2022) to liquid staking (2023) to restaking (2024). Each rotation created temporary winners and left behind overvalued projects. I recall tracking the NFT explosion in 2021: I analyzed 1,200 Bored Ape transactions and found that community sentiment metrics showed increasing isolation and toxicity, not the utopian connection promised by founders. The narrative was a status symbol, not a utility. The same is true for AI models today—they are status symbols for companies, not proven productivity tools. Joshi's 'capital misallocation' risk is the crux. The AI stack's infrastructure layer may be overbuilt relative to current demand. GPU rental prices (H100 spot rates) have already softened in 2024. If the next rotation moves capital away from chips and toward applications, infrastructure companies will face a valuation correction—but the overall AI market will not crash because money will flow into the next narrative. This is structurally identical to crypto's rotating bubbles. Contrarian: The Crypto Receiving End The contrarian angle that Joshi's analysis does not explicitly address is the spillover effect on crypto. If AI's rolling bubble continues, where does the capital go when the AI narrative saturates? History suggests that high-risk, high-narrative assets attract overflow. During the 2020-2021 cycle, as tech stocks peaked, retail and institutional capital rotated into crypto. The same pattern could emerge now. But there is a darker interpretation. The AI bubble's rolling structure means that the overall market's risk profile is increasing, not decreasing. Each rotation adds a new layer of overvaluation, and the underlying capital misallocation compounds. When the music finally stops—if macroeconomic conditions tighten or a black swan appears—the simultaneous deflation of multiple overvalued layers could trigger a synchronized crash across both AI and crypto. The rolling bubble becomes a ticking time bomb, not a safety valve. I learned this lesson during the LUNA and FTX collapses in 2022. I suffered severe burnout and retreated from public discourse for three months. During that solitude, I re-evaluated my core values, deciding to focus only on infrastructure projects with sustainable economic models. The same discipline should apply to AI. The rolling bubble creates an illusion of safety—'it's not all going to crash at once'—but that illusion is dangerous. The risk is not a single event; it is a slow accumulation of systemic fragility. Takeaway: The Next Layer to Watch If Joshi is right, the next layer of AI's rolling bubble is likely to be applications. That means capital will flow into AI-powered software, AI agents, and decentralized AI networks. For crypto, this creates a unique opportunity. Decentralized AI infrastructure—protocols that offer compute, data, and model training on blockchain—could become the next narrative hotspot. Projects like Render, Akash, and Bittensor have already seen increased attention. But the contrarian truth is that most of these projects will suffer the same fate as their centralized counterparts: they will be overvalued relative to actual usage. Efficiency is not empathy. Code doesn't feel. The rolling bubble will eventually force a reckoning—not through a single crash, but through a series of quiet collapses. The crypto market, with its own history of narrative rotations, should prepare for a capital influx from AI, but also for the eventual hangover. The smart strategy is to focus on infrastructure with real utility, avoid the hype layers, and watch for the moment when the next rotation begins. That moment is closer than most think.