I’ve been in crypto long enough to recognize when the market stops pricing fundamentals and starts pricing a word, and right now that word is “AI.” A few years ago it was “Metaverse,” and I remember very clearly how quickly everything with even a loose connection to virtual worlds started flying after Meta changed its name. Tokens like Decentraland and The Sandbox became proxies for a future that sounded inevitable, valuations detached from usage almost overnight, and for a while it genuinely felt irrational not to participate. We all know how that ended, and that memory is probably the biggest reason I’m approaching AI tokens very differently.
To be clear, this is not a cynical “AI is a scam” argument, because I actually believe AI is one of the most transformative technologies of our lifetime and I use it daily. Unlike the metaverse cycle, AI already has tangible adoption, real utility, and measurable impact across industries, including crypto where autonomous agents can trade, rebalance, optimize yield and execute strategies on-chain without human intervention. The technology is not imaginary, the activity is verifiable, and the direction of travel is obvious. What I question is not the relevance of AI, but the automatic assumption that every token associated with it deserves a premium valuation simply because it sits under that umbrella.
Over time I’ve learned to separate narratives from ownership. A powerful narrative attracts attention, liquidity and speculation, but that doesn’t automatically translate into sustainable value capture at the token level. When I look at many AI-related crypto projects today, I often see thin layers built on top of centralized infrastructure such as OpenAI, where the “decentralized AI” pitch starts to look fragile the moment you ask what happens if the API access changes, pricing increases, or regulatory constraints tighten. If a protocol’s core intelligence depends on a centralized provider that can flip a switch, I struggle to treat the token as long-term infrastructure rather than a leveraged bet on external dependencies.
Tokenomics add another layer of hesitation for me, because even when I find the technical direction interesting, the capital structure frequently undermines the investment case. Large venture allocations, aggressive vesting schedules, low circulating supply relative to fully diluted valuation and upcoming unlocks create structural sell pressure that has nothing to do with product success. In those setups, you’re not purely betting on AI adoption; you’re implicitly betting that demand growth will outpace supply expansion at precisely the right time, which is a very different and far more tactical trade than most people admit. I’ve played that game before and I no longer find it attractive as a core strategy.
Regulation is another variable that feels underpriced, especially in Europe where frameworks like the EU AI Act signal that AI will not operate in a vacuum, and when you combine regulatory scrutiny on AI with the already complex and evolving landscape of crypto compliance, you end up stacking two uncertain domains on top of each other. Markets in euphoric phases rarely discount friction adequately, and I prefer to assume that friction will exist rather than betting that it won’t.
That said, my stance is not absolute, and I think it’s important to acknowledge nuance. I still hold a small position in Virtuals Protocol, and I’m comfortable being transparent about that because the reasoning reflects how I differentiate between hype and infrastructure. My exposure is modest relative to my overall portfolio, but I keep it because I see it less as a bet on a single agent and more as a bet on the coordination layer that enables agents to be created, deployed and monetized. If AI agents continue to proliferate on-chain, value is likely to accrue disproportionately to platforms that facilitate that activity rather than to individual agent tokens competing for attention. In other words, I’m more interested in owning a piece of the rail than a piece of one train, and even then I size the position in a way that assumes volatility and narrative compression are inevitable.
Experience has changed how I react to dominant themes, and I’m far less concerned today about missing the first explosive leg of a move than I am about being structurally exposed to something that lacks durable value capture. AI will almost certainly reshape crypto, finance and digital coordination in ways we can’t fully anticipate yet, but that doesn’t obligate me to buy every token that references it. I would rather wait for clearer revenue models, cleaner token design, meaningful decentralization at the infrastructure layer and evidence that fees are flowing sustainably to token holders before increasing exposure. Until then, my approach is selective, sized conservatively, and grounded in the belief that believing in a technology is not the same thing as believing in its current market pricing.
