I didn’t think I’d ever write an article about a data center IPO on a crypto newsletter. But here we are. Switch, the Nevada-based colossus, is reportedly preparing to go public with an $80 billion valuation. That’s not a typo. Eighty. Billion. For a company that rents out space, power, and cross-connects.
If you’re a DePIN maxi, this number should sting. Because while you’re busy shilling decentralized storage tokens with $50 million in TVL, Switch is about to raise more capital in one IPO than the entire market cap of Filecoin, Arweave, and Akash combined. The blockchain doesn’t care about your proof-of-replication algorithm. It cares about latency, power costs, and SLAs.
Let’s unpack what this IPO actually means for the intersection of physical infrastructure and crypto. No hopium. Just cold, hard P&L logic.
Context: The Machinery Behind the Hype
Switch is a wholesale/retail colocation provider. Their product is a concrete box with redundant power (N+1, sometimes 2N), advanced cooling (air and liquid), and dense fiber connectivity to major cloud providers and internet exchanges. Their customers are hyperscalers, financial institutions, and increasingly AI labs. The unit economics are brutally simple: rent per kilowatt (kW) per month, minus construction cost (CapEx) and electricity cost (OpEx). The magic lies in scale - building a 100-megawatt campus costs far less per megawatt than a 10-megawatt one. Switch’s flagship “The Citadel” campus in Tahoe Reno is rumored to be one of the largest single sites in the world.
Now, contrast this with the typical DePIN pitch: a distributed network of home miners or small data centers, token-incentivized, with a governance token that supposedly aligns participants. Sounds great in a whitepaper. But in practice, the unit economics are abysmal. A residential node with a 1 Gbps uplink and a 2 TB SSD costs ~$50/month in electricity and bandwidth. The token rewards need to exceed that to attract supply. Meanwhile, Switch can buy power at $0.03/kWh wholesale, while a residential user pays $0.12/kWh. The blockchain doesn’t magically erase the 4x power cost disadvantage.
Core: The Moat That Can’t Be Forked
I spent three years analyzing crypto protocols, but my real education came from the MEV incident in 2020. I watched gas wars destroy retail traders. That experience taught me to look at micro-structure, not just price action. So let’s apply that lens to Switch.
Switch’s competitive moat is not tech - it’s asset scale and interconnection density. They own land with secured power contracts, often with access to cheap renewable energy (hydro in Nevada, wind in the Midwest). They have long-term PPAs that lock in low rates. And they operate meet-me rooms where dozens of carriers and cloud providers peer. That creates a classic indirect network effect: more carriers attract more tenants, which attracts more carriers.
Now, can a decentralized network like Akash or Render replicate this? In theory, anyone can offer compute. But in practice, the switching cost for an enterprise to onboard a new provider is enormous. They have to re-negotiate security, compliance (SOC 2, HIPAA), integrate with their existing network, and trust the uptime. One outage can cost millions. Switch’s SLA guarantee is backed by insurance and decades of operational history. A DAO with a multi-sig wallet can’t offer that. Yet.
The contrarian angle here is that most DePIN projects are not competing with Switch. They are competing with each other for the long tail of compute - hobbyists, researchers, small businesses. Switch serves hyperscale demand. The only decentralized project that even comes close is Filecoin, but its deal-making volume is still tiny compared to AWS or even Switch. According to the latest data, Filecoin’s active storage deals represent ~30 PB. Switch’s total deployed capacity is likely in the exabyte range. The blockchain doesn’t scale to enterprise SLAs yet.
Contrarian: The AI Hype Is Helping Both Sides, but Unevenly
Everyone says AI will save crypto. “Decentralized AI compute will disrupt the cloud.” I don’t buy it. Not yet.
Switch’s $80B valuation is directly fueled by AI demand. The AI training clusters require high-density power (30-50 kW per rack), liquid cooling, and low-latency interconnects. Switch has been retrofitting its older facilities for liquid cooling and building new ones with 200 MW+ capacity. Stock analysts are drooling over the backlog of pre-leased capacity from hyperscalers and AI labs.
Meanwhile, AI compute on blockchain networks is a punchline. The latency on Ethereum layer-2s is still seconds. Solana is fast, but its compute units are limited and don’t support GPU workloads. Projects like Gensyn and Akash are promising but currently serve only batch inference and training for niche models. The unit economics don’t work for large-scale AI training because the network design relies on untrusted nodes that need verification, which adds overhead. For an AI lab that wants to train a 100-billion-parameter model, the opportunity cost of waiting for a decentralized network to verify results is too high. They’d rather pay AWS or Switch and move fast.
This doesn’t mean DePIN is dead. It means the low-hanging fruit is not AI compute - it’s tasks like CDN caching (Helium’s shift?), archival storage (Filecoin), or even gaming compute (Render). The real use case for decentralized infrastructure is where centralization introduces censorship risk, not where raw efficiency is king.
Takeaway: What Smart Money Will Do
If Switch IPOs at $80B, traditional investors will scramble to buy. They see a recurring revenue model with high switching costs and a secular tailwind from AI. Crypto investors should look at this as a signal to get realistic about DePIN’s timeline.
I’m not saying sell all your DePIN bags. But I am saying the blockchain doesn’t eliminate the laws of thermodynamics or utility pricing. Any decentralized physical infrastructure network that cannot demonstrate unit economics comparable to a wholesale data center will eventually collapse under token decay. The projects that survive will be those that focus on underserved niches - not head-on competition with Switch.
My personal trade? I’m short the hype, long the infrastructure that actually exists. Watch the power procurement announcements from public data center REITs like Equinix and Digital Realty. If they start buying renewable PPAs at record volumes, that’s your signal that the AI boom is real - and the DePIN narrative may need to pivot from “disruption” to “complementary layer.”
One final thought: airdrops aren’t business models. Sweat equity is. The next bull cycle won’t be won by the best whitepaper, but by the team that can build a network with real SLAs and real power contracts. Until then, I’ll keep my powder dry and my eyes on the 800-pound gorilla called Switch.