Finding the signal in the static of the new wave.
It started with a single line buried in a strategy note from BCA Research’s Dhaval Joshi: “AI is not a single bubble, but a rolling series of mini-bubbles.”
The phrase landed like a stone in the crypto pond. For months, the dominant narrative across both Wall Street and Web3 has been binary: AI is either the next internet (evergreen growth) or the next dot-com (imminent implosion). Joshi’s frame is a third path—one that has profound implications for capital flows, risk appetite, and the very structure of the next crypto cycle.
I’ve been watching this static since mid-2023, when I was tracking Render and Akash during my “AI-Crypto Convergence Hunter” phase. Back then, the narrative was simple: decentralized compute would capture spillover from centralized AI. Fast forward to 2026, and the picture is far more complex. The AI infrastructure bubble has already inflated, deflated in pockets, and now appears to be rotating into new layers. The question for crypto isn’t whether AI will crash—it’s how the rolling bubble will redirect capital, and whether crypto is the next stop on the rotation.
Context: The Anatomy of a Rolling Bubble
Joshi’s core insight is grounded in a structural reality: the AI tech stack is not monolithic. It’s a layered cake—silicon (Nvidia, AMD), cloud infra (AWS, Azure, GCP), foundation models (OpenAI, Anthropic, Meta), developer tools (LangChain, Weights & Biases), and applications (Palantir, Jasper, Copy.ai). Each layer has its own capital cycle, its own hype curve, and crucially, its own moment of investor attention.
From 2023 to early 2024, the spotlight was on silicon. Nvidia’s market cap tripled, data center CAPEX exploded, and GPU rental prices spiked. Then, in mid-2024, the narrative shifted to models—OpenAI’s $150B valuation, Anthropic’s $5B round, and the “AGI race” narrative. By late 2025, the momentum had started to drift toward applications, with Palantir and other AI-native SaaS companies seeing multiple expansions.
This isn’t just sector rotation. It’s a structural mechanism that prevents a single, simultaneous crash. As Joshi puts it, “capital misallocation is the real risk, not an immediate blow-up.” The misallocation happens because money chases the hottest layer, leaving other layers undercapitalized relative to their eventual role. And when the heat moves, the previous layer cools—but doesn’t necessarily crash—because the story isn’t broken, just postponed.
Core: The Signal-in-Noise Filter for Crypto
Now, let’s connect this to the static of the new wave—crypto. If AI bubbles are rolling, then the liquidity that flows into AI does not disappear; it rotates. The question is: where does the rotation go next?
Based on my experience tracking narrative cycles since 2020, I’ve observed a pattern: when a dominant tech narrative (like DeFi summer or the NFT mania) begins to show signs of fatigue, capital doesn’t exit the risk asset class entirely. Instead, it seeks the next frontier narrative that offers a similar mix of novelty, asymmetry, and escape velocity. In 2021, it was NFTs. In 2023, it was AI. In 2025, it was AI agents.
Here’s the contrarian layer: the rolling AI bubble could actually be a net positive for crypto in the medium term. Here’s why:
- Capital Spillover: When AI infra valuations become too rich relative to growth, institutional investors will look for cheaper, higher-beta alternatives. Crypto—specifically decentralized compute, AI-focused L1s, and tokenized AI agents—offers exactly that. We saw this in late 2024 when Render’s token price surged 400% after Nvidia’s earnings miss.
- Narrative Substitution: The same human psychology that drives AI hype—the fear of missing out on the next big thing, the desire for a “revolutionary” story—drives crypto cycles. When AI’s marginal narrative excitement fades, the next narrative (often crypto + AI) becomes the new plaything for momentum capital.
- Structural Demand: AI models need compute, and decentralized compute networks (Akash, io.net, Ritual) offer a compelling alternative to centralized cloud for specific use cases (inference, fine-tuning, privacy-preserving model training). The rolling bubble means that even if centralized AI infra sees a cooling, demand for decentralized alternatives could accelerate as capital seeks differentiation.
But there’s a catch. This spillover only works if crypto’s own infrastructure is ready. We’ve seen this movie before: in 2022, when Terra collapsed, the entire crypto market was treated as a single risk bucket. The same could happen if AI’s rolling bubble suddenly reverses into a synchronous crash, triggering a broad risk-off move that drags down crypto along with it.
Contrarian: The Blind Spot of the “Rolling” Thesis
Let me push back against Joshi’s framework—because as a Narrative Hunter, I know that every framework has blind spots.
The rolling bubble thesis assumes that the layers of the AI stack are sufficiently independent to decouple their valuations. In reality, they are deeply interconnected. Nvidia’s revenue depends on cloud providers buying GPUs, which depends on model companies raising capital, which depends on applications generating revenue. If the application layer fails to monetize, the entire chain collapses. The rolling effect merely delays the day of reckoning.
Second, the thesis ignores the macro environment. We are in a bear market for risk assets, where liquidity is scarce and central banks are still tightening. In such an environment, rolling bubbles tend to converge—because when the tide goes out, all boats are exposed. The 2022 crypto crash was a perfect example: DeFi, NFTs, L1s all crashed in unison, despite different narratives.
Finally, the thesis underestimates the energy bottleneck. AI data centers are consuming massive amounts of electricity, and grid constraints are already limiting new builds. This is a physical cap on the infrastructure layer that no amount of narrative rotation can solve. If the infra layer hits a hard wall, the entire stack feels it.
So where does this leave crypto? The real risk is not that AI steals crypto’s narrative—it’s that AI’s rolling bubble eventually synchronizes with crypto’s own cycle, creating a double-whammy of capital destruction. We saw this in March 2025 when the AI sector correction (driven by NVIDIA’s guidance) coincided with a crypto sell-off (driven by ETF outflows). The correlation is growing.
Takeaway: The Next Chapter Loading
Finding the signal in the static of the new wave.
Joshi’s framework is useful, but it’s incomplete. The rolling AI bubble is not a guarantee of salvation for crypto—it’s a structural feature that will amplify both the highs and the lows. The next 12 months will be a test: if AI applications start generating real revenue, capital will stay in AI, and crypto will be starved. If AI applications disappoint, the rolling bubble will accelerate its rotation, and crypto could be the next beneficiary.
My playbook: Watch the GPU rental spot price. Watch the token unlock schedule of AI-related crypto projects. Watch the quarterly earnings of Palantir and Snowflake. The signal will come from the intersection of real-world AI adoption and on-chain capital flows.
