The coding whispers truths only the silent can hear.
Last week, I sat through a data dump that most analysts would scroll past: BCA Research’s chief strategist Dhaval Joshi framing the AI boom not as a single supernova, but as a rolling bubble. A cascade of localized overvaluation, each layer igniting, cooling, then passing the torch. The market’s reaction? A collective shrug. But in the red noise of CAPEX sheets and GPU rental rates, I caught a quiet signal that should keep every crypto participant awake.
Context: The Narrative of Fragmented Hype
Since 2023, the AI narrative has been a monolithic beast. Nvidia’s market cap eclipsed $3 trillion. OpenAI’s valuation hit $150 billion. Yet Joshi’s framework—one I’ve quietly tested against my own audit of on-chain capital flows—suggests this is not a single bubble destined to pop, but a sequence of bubbles sliding across the tech stack: infrastructure (chips, data centers) → model layers (LLM training) → tooling (middleware, dev frameworks) → application (enterprise AI solutions). Each layer attracts a wave of speculative capital, inflates, then partially deflates as the next layer takes the spotlight.
This is not a new pattern. The 1990s internet bubble was itself a rolling series: semiconductors (Intel) → portals (Yahoo) → e-commerce (Amazon) → fiber optics (Global Crossing). Each sub-bubble burst locally, but the aggregate narrative kept the whole system afloat until the final collapse. Joshi’s warning is that AI is repeating this script—and the crypto market, which mirrors high-risk narrative cycles, should be listening.
Core: The Mechanism of Mispricing and the Crypto Parallel
Let me decode the rolling bubble’s anatomy. At its heart is capital misallocation: money flows to the layer with the hottest narrative, not the highest return on capital. In AI, infrastructure layer (GPU farms, data centers) captured the first wave. Venture capital poured into chip startups and cloud providers, driving up CAPEX far beyond current demand. Nvidia’s data center revenue grew 400% year-over-year in 2024, but GPU rental prices (e.g., H100 spot rates) have already started to soften. The signal is clear: the infrastructure bubble is starting to cool, and capital is rotating to the model layer and beyond.
Trust is a variable, not a constant.
I’ve seen this script before in crypto. In 2020, DeFi summer’s liquidity mining APYs were the infrastructure layer’s equivalent: subsidized TVL that vanished when incentives stopped. Compound’s governance narrative was a social contract, but the code of whale dominance betrayed the ideal. The same mispricing is now playing out in AI. The question is not if the bubble will roll, but when it will reach crypto’s doorstep.
Why should crypto care? Because the rolling bubble creates an inter-market correlation risk. When AI infrastructure valuations compress, capital doesn’t disappear—it rotates. Some of it will flow into the next high-beta narrative. Crypto, with its own layered structure (L1 → L2 → DeFi → NFTs → AI agents), is a natural recipient. But the rotation is not a blessing; it’s a delayed systemic risk. If the AI bubble eventually cascades into a synchronized collapse (as Joshi’s model implies when the rolling stops), the crypto market, already in a bear cycle, could face a liquidity shock.
In the red, I found the quiet signal.
Let me ground this in my own experience. In 2022, during the FTX crash, I retreated from public analysis. The narrative collapse was emotionally brutal—but it taught me to read the structure behind the noise. Today, I’m applying the same lens to the AI bubble. The rolling pattern means that the market’s fragility is not visible in the aggregate indices. It’s hidden in the internal capital flows between layers. For crypto, the key metric is not BTC dominance or TVL, but the cross-chain capital rotation from AI-related tokens (like RNDR, FET, AGIX) to DeFi infrastructure. This rotation is already happening, but it’s quiet.
Contrarian: The Bubble That Never Pops?
Here’s the counter-intuitive angle: a rolling bubble may actually delay the ultimate crash. By fragmenting the overvaluation across multiple layers, the system avoids a single, catastrophic blow-off. Each local deflation acts as a pressure release valve. This is precisely what happened in 2023-2024: GPU stocks corrected 20% in mid-2024, but AI model companies (like Anthropic) raised massive rounds, keeping the narrative alive. The crypto market, too, can benefit from this if it positions itself as the next layer.
Fragility breaks the loudest voices first.
But the contrarian trap is to assume that rolling means harmless. The truth is that capital misallocation accumulates. The longer the roll continues, the more capital is mispriced across all layers. When the final rotation fails—when no new layer can absorb the overvaluation—the entire stack collapses. The 1990s internet bubble ended this way. The infrastructure layer (fiber) had real long-term value, but the timing of the collapse caught even the best projects. Today’s AI infrastructure (GPU farms) has similar durability, but the capital misallocation is far larger. The risk is not a crash today, but a crash tomorrow—with a steeper cost.
I’ve seen this fragility in crypto’s own rolling bubbles: DeFi → NFT → GameFi → Meme coins. Each sub-bubble burst, but the market kept rolling until the FTX black swan. The current AI rolling bubble is a macro mirror of that pattern. The crypto market must prepare for a scenario where AI capital rotation finally hits our shores, not as a bullish catalyst, but as a risk transfer.
Takeaway: The Next Narrative
So where does the roll land next? The data suggests that capital is moving from AI infrastructure to the model layer and then to AI applications. But the crypto-native layer—decentralized compute, AI agents on blockchain, and verifiable inference—is still a tiny fraction of the flow. The opportunity lies in the intersection: proof-of-use for AI compute. If the rolling bubble forces investors to seek real ROI, crypto’s transparency and on-chain verification could become the next narrative.
To hold firm is to understand the void.
The code whispers. The signal is in the mispricing. The market is not crashing; it’s rotating. The question is not whether the bubble will burst, but whether you are positioned in the layer that capital will touch next—or the one that will be left empty.
Listen to the quiet chains. The truth is in the red.