The AI Bubble Isn't Bursting — It's Rolling, and Crypto Should Be Wary

Over the past twelve months, Nvidia's market cap has surged past three trillion dollars, while AI startups raised billions at valuations that would make a 2021 NFT collector blush. Yet beneath the surface of this euphoria, a quiet warning from BCA Research's chief strategist Dhaval Joshi has been circulating through institutional circles: the AI bubble is not a single, monolithic explosion waiting to rupture. It is a rolling cascade of micro-bubbles, each inflating and deflating across different layers of the technology stack. This thesis, reported by Crypto Briefing, carries profound implications not just for AI investors, but for the entire architecture of decentralized capital — including the crypto markets we inhabit. I have spent the last decade building and auditing cryptographic systems, from the Parity Wallet vulnerability in 2017 to the MakerDAO governance battles of 2020. I have seen capital flow like a river in flood, then vanish into a dry riverbed. The rolling bubble narrative resonates with me because it mirrors the very cycles of belief and disillusionment that have shaped Web3. Today, I want to trace this concept back to its ethical and structural roots, and ask: what does it mean for those of us who believe in decentralization as a practice of radical empathy? Let me begin with the core insight. Joshi argues that AI valuation is not a single overvalued asset class, but a sequence of overheated sub-sectors — infrastructure (GPUs, data centers), foundational models (LLMs), tooling (development frameworks), and applications (industry solutions). Capital migrates from one layer to the next as each new narrative captures the imagination of investors. The risk is not a sudden crash, but a persistent misallocation of resources: too much money chasing GPU clusters before the demand for AI inference is proven, too much hype around model providers before they show sustainable revenue, too many startups burning cash on chatbots that nobody will use six months later. This is capital misallocation in slow motion, disguised as innovation. I have seen this pattern before. In 2017, the ICO bubble was itself a rolling phenomenon. First came Bitcoin, then Ethereum, then ERC-20 tokens, then exchange tokens, then infrastructure projects like storage and compute. Each wave promised a new paradigm, and each wave left behind a graveyard of failed projects — but also a few enduring survivors. The difference is that the ICO bubble was a pure expression of decentralized, unregulated capital. The AI bubble, by contrast, is driven by the same institutional forces that have always controlled the levers of production: Microsoft, Google, Amazon, Meta. They are the ones writing the checks for the GPU clusters. They are the ones funding the model labs. They are the ones who will decide when the music stops. And here is where the rolling bubble thesis becomes a crypto story. If AI capital is a series of waves, then the next wave may well spill over into the crypto space. Not because crypto is a better investment, but because the narrative of AI will eventually need a new frontier — and decentralized inference, tokenized compute, and AI-agent economies are already being pitched as the next big thing. The question is whether this capital will bring genuine value or simply inflate another set of valuation balloons. Based on my experience auditing smart contracts and observing the MakerDAO governance process, I am deeply skeptical. Capital that flows from a bubble in one sector to a bubble in another without underlying value creation is not innovation; it is a Ponzi scheme with a longer timeline. Let me illustrate with a specific example from the infrastructure layer. The current AI capital expenditure by the four largest cloud providers exceeded two trillion dollars in cumulative CAPEX over the past two years. Much of this has gone into Nvidia's H100 and B200 GPUs. Yet the actual utilization rates for these chips — especially for inference workloads — remain below 50% in many data centers, according to industry estimates I have gathered from operator conversations. This is the classic sign of capital misallocation: money spent on capacity that will not be fully used for years, if ever. The rolling bubble allows this because the narrative shifts from infrastructure to models before the infrastructure's ROI is even measured. The capital that went into GPUs is now chasing foundation models, leaving the GPU owners holding the bag when demand fails to materialize. Now, consider the crypto parallel. In 2022, after the collapse of Terra and FTX, the narrative of 'decentralization' was corrupted by centralized exchanges and opaque funds. I retreated to a quiet apartment in Hanoi and wrote the Ho Chi Minh Trust Manifesto, arguing that true decentralization requires psychological resilience and community verification over algorithmic guarantees. The same principle applies here. The rolling bubble in AI is a test of our collective patience and our ability to see through hype. If we, as a community, start chasing AI tokens merely because the capital is flowing, we will repeat the same mistakes. But there is a contrarian angle worth exploring. Joshi's rolling bubble thesis may actually underestimate the self-correcting power of decentralized markets. In a permissionless system, capital can flow to the most efficient use cases without the approval of a central allocator. The crypto market has survived multiple rolling bubbles — DeFi Summer, NFT mania, Layer-2 wars — because each cycle left behind infrastructure that was genuinely useful. Uniswap, Aave, and Ethereum itself are still here. The same could be true for AI: even if the current wave of capital misallocation is wasteful, the leftover GPU clusters and open-source models could become the foundation for a more decentralized AI ecosystem. The protocol must serve the human spirit, not the other way around. Let me be clear about the risks. The rolling bubble does not eliminate the possibility of a systemic crash; it delays it, and in doing so, allows the misallocation to compound. If the macroeconomic environment shifts — rising interest rates, geopolitical conflict, or a sudden liquidity crunch — the multiple layers of the bubble could deflate simultaneously, causing a cascade that sweeps through both AI and crypto. This is the tail risk that Joshi's framework does not fully address. As someone who has witnessed the 2022 crash firsthand, I know that narrative can reverse in a matter of hours. We must build our protocols and our communities to withstand such shocks. Listening to the silence between the blocks, I hear a different truth. The rolling bubble is not a prediction; it is a warning. It tells us that capital is not a rational allocator but a herd animal, moving from one feeding ground to the next. The only way to avoid being trampled is to build something that does not depend on the herd's direction. Decentralized infrastructure, community-governed treasuries, and sovereign identities — these are the assets that will survive when the rolling bubble finally stops rolling. Tracing the code back to the conscience, I remember the 2017 Parity audit. I found a reentrancy vulnerability that could have drained $300 million. I reported it privately, and the developers fixed it. That moment taught me that code alone cannot ensure trust; it requires human vigilance. The same is true for the AI bubble. No amount of algorithmic efficiency can replace the ethical judgment of a community that decides to allocate capital not for maximum short-term gain, but for long-term sustainability. Governance is not a vote; it is a vigil. We must stay awake as the AI bubble rolls through each layer, and we must be ready to redirect the flow when it threatens to drown the very values we hold dear. The crypto market is not a passive recipient of this capital; it is an active participant in shaping the future of human coordination. If we accept the rolling bubble as inevitable, we lose the opportunity to build a better alternative. We build bridges from the ashes of belief. The belief that AI will save us, or that crypto will save us, is a form of technological messianism that blinds us to the real work of creating resilient systems. The rolling bubble is a call to focus on fundamentals: revenue, retention, and real-world impact. It is a call to reject the VC-driven narrative that liquidity fragmentation is a problem to be solved with more products, when in fact it is a symptom of an industry that has lost its way. Holding space for the digital soul requires us to ask difficult questions. Where will the next wave of AI capital land? If it lands in crypto, will it lift all boats, or will it create a new layer of misallocation? I have no easy answers, but I know that the answer will not come from a spreadsheet or a whitepaper. It will come from the collective wisdom of a community that dares to think differently. Truth is the only immutable asset. In a world of rolling bubbles, the only thing that lasts is the integrity of the systems we build. Let us build them with care, with humility, and with a deep respect for the human spirit that technology is meant to serve. The AI bubble is not bursting. It is rolling. And if we are not careful, it will roll right over us. But if we are vigilant, we can catch the wave and ride it to a place where capital serves community, not the other way around.