Hook
A single data center company—Switch—is preparing an IPO at an $80 billion valuation. That’s not a number from a crypto bull run; it’s the price tag Wall Street is slapping on centralized AI compute. The news broke last week, carried by Bloomberg, with the usual fanfare: top-tier underwriters, a clear timeline, and whispers of ‘AI infrastructure’ gold. But for anyone who has spent years in the trenches of blockchain—auditing smart contracts, participating in DAO votes, writing post-mortems on protocol collapses—this story smells familiar. It’s the same narrative inflation that fueled ICOs, DeFi summer, and the NFT mania. Only this time, the asset is not a token—it’s a building. And the code that runs inside it belongs to a handful of CEOs, not a community.
Context
Switch operates data centers that house the servers powering AI training, cloud computing, and enterprise workloads. Its $80 billion valuation places it alongside Equinix and Digital Realty as a first-tier player. The pitch? That AI’s insatiable hunger for compute makes high-density, liquid-cooled facilities the new oil fields. Switch’s unique “SuperNAP” campuses combine bare-metal leasing with a community of co-located tenants—creating switching costs that supposedly build a moat. But as a crypto editor who has watched narrative cycles since 2017, I see a pattern: every boom redefines what counts as ‘infrastructure,’ and every bust reveals how fragile those definitions are.
Consider the ICO era. Projects raised billions on whitepapers promising decentralized computing. Six months of auditing 17 of those whitepapers in 2017 taught me that most were castles in the air. Fast forward to DeFi Summer 2020: I spent three weeks participating in Compound governance, voting on proposals, and writing about the human layer of yield. The code was open, but the risks were hidden in governance tokens and liquidation cascades. Then came NFTs: I retreated to Big Sur in 2021 to build a project linking art to carbon offsets, only to watch the market drown in speculative PFP trading. Each cycle, the narrative gets bigger. Each cycle, the underlying asset gets more centralized.
Switch’s IPO is no different. It’s a bet that AI compute will grow exponentially—and that a single provider can capture that growth. The same bet was made on Bitcoin mining farms, on Ethereum staking pools, on Layer-2 sequencers. And every time, the promise of efficiency collided with the reality of concentration.
Core
The question isn’t whether Switch is a good business. It is. The question is whether $80 billion is a sustainable narrative—or another example of what I call ‘narrative decay’: when market enthusiasm outpaces technical and economic reality.
Let me bring in data from my own work. In 2022, after Terra/Luna collapsed, I produced a 40-page post-mortem titled ‘Narrative Decay.’ The core finding: broken promises erode trust faster than broken code. The same is true for centralized infrastructure. Switch’s valuation assumes that its clients—mostly hyperscale cloud providers and AI labs—will remain loyal for decades. But history shows that when a single provider controls too much compute, the incentive is to squeeze margins or build in-house. AWS, Google, and Microsoft are already building their own data centers. If they leave Switch, $80 billion becomes $10 billion overnight.
I see a parallel to the BRC-20 and Runes debate on Bitcoin. Using Bitcoin’s base layer for token swaps is like using a Rolls-Royce to haul cargo—it insults the car and doesn’t carry much. Similarly, using a centralized data center for AI compute that could be distributed across thousands of nodes is efficient but brittle. The code doesn’t care about geography; it cares about fault tolerance. Decentralized networks like Akash, Filecoin, and Render offer compute and storage with no single point of failure. Yet they trade at fractions of Switch’s implied value—because their narrative hasn’t peaked yet.
During my work on the Veritas Protocol in 2026, I learned that trust requires human skin in the game. Centralized infrastructure removes that skin. Switch’s SuperNAP community might create switching costs, but it doesn’t create digital sovereignty. When AI models rely on a single data hub, they inherit its single point of governance.
Here’s the counterintuitive part: the AI boom may actually accelerate decentralization. Not because centralized data centers are bad—but because they’re too valuable. Governments and enterprises will demand verifiable, uncensorable compute for sensitive workloads. That’s where blockchains with proof-of-replication, zero-knowledge proofs, and on-chain verification come in. Centralized providers can’t offer that without becoming ‘federated’—which is just a polite word for partially decentralized.
Contrarian
The blind spot in the Switch narrative is its assumption that ‘community’ equals resilience. In crypto, community means distributed governance, shared risk, and open participation. In Switch’s model, community means a landlord-tenant relationship with good marketing. The real value in AI compute isn’t the physical building—it’s the digital provenance of the computation. Did the model train on verified data? Was the inference run on tamper-proof hardware? Will the results be auditable?
Soulless finance is just empty pixels. The same applies to soulless compute. Investors piling into Switch are buying the story of a new gold rush without realizing that the gold itself—trust—is being mined by decentralized networks every day.
Takeaway
The next narrative shift will be from ‘data center as fortress’ to ‘compute as a public good.’ Centralized hubs will always exist for latency-sensitive tasks, but the high-margin, high-trust workloads will migrate to decentralized alternatives. Just as Bitcoin proved that money can exist without banks, decentralized compute will prove that intelligence can exist without a landlord. The code doesn’t care about your building. It cares about your hash.
