Last week, as Nvidia’s market capitalisation crossed the $3 trillion mark for the first time, a quieter but more telling signal emerged from the semiconductor landscape. Funds that had been pouring into capital equipment stocks—ASML, Applied Materials, Tokyo Electron—suddenly reversed course. The narrative shifted from ‘building the AI factory’ to ‘who actually profits from the chips inside it?’ This rotation wasn’t panic. It was precision. And for those of us who have spent years inside blockchain markets, the pattern feels uncomfortably familiar.
In crypto, we’ve lived through the same cycle. The 2017 ICO boom fed a frenzy around infrastructure—miners, validators, and base-layer protocols. But by 2020, the market had rotated sharply toward applications: DeFi lending protocols, NFT marketplaces, and yield aggregators. The asset that captured the most value wasn’t the one that built the roads, but the one that tolled the traffic. Today, as I observe the AI semiconductor landscape mirroring our own history, I’m convinced that blockchain is about to experience its own version of this rotation—and most investors are still looking at the wrong part of the stack.
Context: The Infrastructure Mirage
Blockchain’s first decade was dominated by the ‘picks and shovels’ thesis. Bitcoin miners, Ethereum validators, and Layer-1 tokens were the safe bets. The logic was simple: if the ecosystem grew, the underlying asset would appreciate. And for a while, it worked. Bitcoin’s hash rate climbed, Ethereum’s staking pool expanded, and every new Layer-1 launch promised a superior consensus mechanism. But after auditing 42 failed ICO whitepapers back in 2017, I realised that 85% of them lacked a sustainable value proposition beyond speculation. They were infrastructure projects with no clear path to user adoption. The market eventually agreed. From 2018 to 2020, infrastructure tokens underperformed while protocols that actually generated revenue—Uniswap, Aave, Chainlink—saw outsized returns.
The parallel today is stark. The AI boom has driven massive capital expenditure on semiconductor fabrication equipment. ASML’s extreme ultraviolet lithography machines are the equivalent of Bitcoin mining rigs: essential, expensive, and increasingly commoditised. But just as mining profitability became a race to the bottom, so too will equipment margins once the initial build-out saturates. The real value in AI, as in blockchain, lies not in the tools that produce the chips, but in the chips that produce the intelligence—and the applications that monetise that intelligence.
Core: The Blockchain Analog for Value Capture
At the heart of this rotation is a fundamental question: where does value accrue in a decentralised ecosystem? In blockchain, the answer has evolved. During the 2021 bull run, we saw the rise of ‘fat protocol’ theory—the idea that Layer-1 tokens capture more value than application tokens. But recent data contradicts this. Ethereum’s fee revenue, while substantial, has been eclipsed by the combined fees of Layer-2 solutions and high-velocity DeFi protocols. The value has shifted upward: from the consensus layer to the execution layer.
Based on my experience building the ‘Ethical Node’ newsletter, I documented how DeFi protocols like Uniswap captured disproportionate value relative to the Ethereum base layer. In 2022, I isolated the trading fees of the top five Ethereum applications and compared them to the fees paid to validators. The applications generated 3.2x more value per transaction than the base layer captured. This is not an accident. It is a structural trend: as the ecosystem matures, value flows to the interfaces that users touch, not the infrastructure that powers them.
The same dynamic is now playing out in AI. Nvidia’s GPUs are the infrastructure—the ‘Layer 1’ of AI compute. But the applications built on top—ChatGPT, Midjourney, Copilot—are capturing the lion’s share of end-user spending. And as AI moves from training to inference, the need for specialised chips (like groq or custom ASICs) will create a new layer of value capture, much like Layer-2 solutions in blockchain that optimise for specific use cases.
Contrarian: The Danger of Confusing Liquidity with Loyalty
The market’s rotation from equipment to AI chips feels rational, but it carries a hidden risk: it equates liquidity with loyalty. Just because capital is flowing into one area does not mean the underlying technology is ready for primetime. In blockchain, we saw this in 2021 when billions poured into NFT projects that had no secondary market—China’s digital collectibles being a prime example. Without a secondary market, NFTs are one-off sales. Even speculators don’t hold them. The same principle applies here: without the equipment to manufacture chips at scale, AI applications will hit a supply bottleneck. The rotation may be premature.
I recall a conversation with a traditional finance academic during my work on the ‘Values-Based Investment Framework’ in 2024. He pointed out that institutional allocators often mistake early adoption for enduring value. In blockchain, they poured money into infrastructure tokens without understanding the cultural ethos of decentralisation. They were drawn to the narrative of ‘digital gold’ or ‘world computer’ but ignored the messy reality of governance and community. Today, similar money is pouring into AI chip stocks, assuming that hardware scarcity will persist indefinitely. But history suggests otherwise: Moore’s Law, while slowing, hasn’t stopped, and new entrants (including blockchain-native compute networks like Render or Akash) are already competing for AI workloads.
Takeaway: The Architecture of Value in the Next Cycle
The question every blockchain investor should ask is not ‘which Layer 1 will win?’, but ‘which application will capture the most value per unit of infrastructure?’ The rotation from equipment to chips is a microcosm of a larger shift: from capital-intensive commodity production to value-dense intellectual property. In blockchain, that means focusing on protocols with strong network effects, recurring revenue, and community alignment. In AI, it means looking beyond the chipmakers to the model developers and the deployment platforms.
I’ve seen this pattern before—during the 2020 DeFi summer, when the market finally realised that value accrues to applications, not infrastructure. The next bull run in blockchain will be driven by applications that solve real-world coordination problems: supply chain tracking, identity verification, and cross-border payments. The tokens that will survive are those that reward loyalty, not liquidity. As I wrote in my 2017 manifesto, ‘The Soul of the Chain’: true decentralisation is an ethical imperative, not a technical feature. The market’s rotation today is a reminder that we must look beyond the hardware and ask where the deepest trust is built.
Silence is the loudest vote in a DAO. Right now, the market is silent on equipment stocks and noisy on AI chips. But I suspect that silence will be short-lived. The next wave of value will come not from the picks and shovels, but from the hands that wield them.