I watched the NVIDIA chart today. It didn’t panic. The volume whispered something else – a quiet, deliberate shift. Not a crash. Not a moon shot. A rotation. The kind of rotation that makes you question every single narrative you’ve been sold about AI’s bubble.
This isn’t the dot-com 2.0 everyone’s screaming about. It’s something far more insidious – and far more opportunistic. BCA Research’s Dhaval Joshi dropped a framework that’s been living rent-free in my head since I decoded it: the AI bubble isn’t a single, bursting balloon. It’s a rolling series of mini-bubbles, each inflating, peaking, and deflating in sequence across the AI stack. And I’ve seen this movie before. In crypto.
Remember DeFi Summer 2020? Liquidity mining was the infrastructure layer. Then NFTs became the application layer. Then L2s became the tool layer. Each wave inflated, popped, and handed the baton to the next. The market didn’t die – it evolved. The same rhythm is playing out in AI right now. And crypto traders who ignore it are leaving alpha on the table.
Context: Why Now?
Joshi’s argument is deceptively simple. AI’s valuation boom isn’t a uniform mania. It’s a temporal chain of capital misallocation. The first bubble formed in the infrastructure layer – GPUs, data centers, cloud compute. That’s where the money went first, because that’s where the demand signal was loudest. But as NVIDIA’s P/E ratio stretched beyond reason, capital started rotating. The next bubble is forming in the model layer – foundation models, large language models, the “iPhone of AI” narratives. And after that? The application layer. The tool layer. Each rotation creates a window of opportunity – and a window of risk.
I’ve been in this game long enough to know that the chart lies. The volume speaks. And right now, the volume in AI infrastructure is saying “too early to exit, too late to enter.” The real money is moving upstream, into the next narrative. The question is: which layer is next?
Core: The Rolling Bubble Mechanics
Let’s break this down with the same lens I’d use to analyze a crypto protocol’s tokenomics. The AI stack has four layers: infrastructure (chips, data centers), models (GPT, Claude, Llama), tools (development frameworks, middleware), and applications (industry solutions). Each layer has its own valuation cycle, its own narrative peak, and its own crash point.
From my experience auditing smart contracts during the 2021 NFT mania, I learned that capital misallocation isn’t random – it’s structural. The same forces that drove Yuga Labs’ valuation to $4B before the rug pull are now driving OpenAI’s $150B valuation. The infrastructure layer is already showing signs of overheating. Data center construction costs are spiking. GPU rental prices are compressing. The early money is rotating out.
Alpha doesn’t wait for permission. The next leg of the rolling bubble is already forming in the model layer. Foundation model companies are raising at valuations that assume they’ll capture the entire value of AI – but history says otherwise. In crypto, we saw this with L1s. Everyone wanted to be the next Ethereum. Most failed. The same will happen in AI models. Only a handful will survive. The rest will be absorbed or forgotten.
But here’s the kicker: the rolling bubble isn’t a crash in disguise. It’s a pressure release valve. Each localized deflation prevents a systemic explosion. The capital that leaves the infrastructure layer doesn’t disappear – it flows into the next layer. This is exactly what happened in crypto during the 2022 bear market. Terra collapsed, but capital moved to L2s and real-world assets. The market didn’t end. It rotated.
Panic sells. I just watch. I’ve been watching the on-chain data for AI-related tokens – yes, there are AI tokens, even if they’re mostly vaporware. The volume patterns are telling. Whales are accumulating in projects that sit at the intersection of AI and crypto, like decentralized compute networks and data provenance protocols. The smart money is betting that the next rotation will be from AI infrastructure to AI infrastructure-as-a-service in crypto. That’s a contrarian play the mainstream analysts are missing.
Contrarian: The Blind Spot Everyone’s Ignoring
Every mainstream article is screaming “AI bubble is about to burst!” They’re looking at the wrong chart. The chart lies. The volume speaks. The volume is saying that the bubble is alive, but it’s moving. The biggest risk isn’t a crash – it’s missing the rotation.
Here’s the contrarian angle: the rolling bubble actually creates a long-term structural floor for AI investments. Because the capital doesn’t vanish; it shifts. The infrastructure built during the first bubble (GPU farms, data centers) has lasting value. Even if the model bubble pops, the compute infrastructure remains. It’s not like the fiber optic glut of 2001, where cables were laid and never used. AI compute is consumable. It gets used. The excess capacity will be absorbed by the next wave of applications.
This is where crypto enters the arena. Decentralized physical infrastructure networks (DePIN) like Render, Akash, and Filecoin are positioned to soak up the excess compute capacity. They’re the “second-hand market” for AI compute. When the infrastructure bubble deflates, cheap GPU cycles will flood DePIN networks, making them more competitive. The rolling bubble becomes a tailwind for crypto infrastructure.
But the conventional wisdom says AI and crypto are separate. That’s the blind spot. The capital rotation will cross asset classes. When AI infrastructure returns compress, some of that capital will flow into crypto-native AI projects. I’ve seen it happen in real time. The moment NVIDIA’s stock dropped 5% on a single earnings miss, trading volume on Akash’s token spiked 300%. The whales are already positioning.
Takeaway: What to Watch Next
The rolling bubble is a gift to active traders. It’s a sequence of predictable rotations. Right now, the infrastructure layer is peaking. The model layer is still inflating. The application layer hasn’t even started. The smart play is to map the rotation timeline and adjust your portfolio accordingly.
For crypto traders, the next six months will be critical. Watch the funding rates on AI token perpetuals. Watch the TVL on DePIN protocols. The chart lies, but the volume speaks. And when the volume shifts from AI infrastructure to AI applications, you’ll know the next bubble has started rolling.
Alpha doesn’t wait for permission. Neither should you.
