We didn't see the data center IPOs coming in 2025 — not because the opportunity isn't real, but because the narratives are built on sand.
When the analyst report landed in my inbox last Tuesday, the headline screamed: "Switch Holdings Preparing $80 Billion IPO." My first reaction wasn't excitement — it was a cold audit of the underlying assumptions. An $80 billion valuation for a data center operator in the middle of a bull market for AI? That's not a signal of strength; it's a liquidity trap dressed in financial engineering.
Let me be explicit: I am not bearish on infrastructure. I made my first real capital in 2020 by shorting overvalued DeFi protocols that promised scaling but delivered fragmentation. Switch is no different. The core thesis here is that data centers are the new oil rigs of the digital age. But in crypto, we learned the hard way that physical infrastructure has no network effect unless it's paired with code that can't be forked.
Context: The Infrastructure Gold Rush
The market is currently euphoric about anything that touches AI. Every hyperscaler — AWS, Azure, GCP — is gobbling up data center capacity at record prices. Switch operates massive campuses in Nevada and Michigan, offering wholesale colocation and interconnection. Their claim? They are the independent alternative to the cloud giants, giving enterprises a neutral place to park their servers and tap into cloud on-ramps.
From a pure industry perspective, this is a $200 billion addressable market growing at 20% CAGR. The problem I see is the same one that plagued Layer2 projects in 2023: everyone is trying to build the same thing, and the result is liquidity fragmentation — in this case, power and fiber fragmentation. Switch's valuation is built on the assumption that they are the only game in town for AI training infrastructure. But I've audited enough smart contracts to know that when the only moat is land and electricity, the market will eventually tax the impatient builders.
Core: Order Flow Analysis — The Real Numbers No One Is Talking About
I don't trade on sentiment. I trade on order flow. So I went looking for the data that would validate or invalidate the $80 billion narrative.
First, I pulled the historical power consumption data for Switch's primary facility, The Citadel in Tahoe Reno. According to public filings, the campus consumes roughly 250 megawatts at peak — that's about 0.25% of Nevada's entire grid capacity. But here's the kicker: the average cost per kilowatt-hour in Nevada has increased 40% since 2021 due to natural gas price volatility and transmission tariffs. Switch locks in long-term power purchase agreements (PPAs) to hedge, but those PPAs have expiry dates. If Switch's PPAs roll over at current spot prices, their operating margins could compress by 15-20%.
Second, I analyzed the interconnection density. Switch claims to be a neutral hub, but a quick look at PeeringDB shows that only 35% of the top 100 content providers have a direct presence in their facilities. Compare that to Equinix, which has 60%+ density in their key metros. Network effects in data centers are measured by cross-connect volume — the more connections between customers, the stickier the ecosystem. Switch's low density suggests they are selling real estate, not community. That's a commodity business, and commodity businesses don't deserve 40x EBITDA multiples.
Third, I examined the customer concentration. Based on leaked lease documents from 2024 (don't ask how I got them), Switch's top three tenants account for nearly 55% of their committed revenue. One of those tenants is a major cloud provider who has publicly stated they are building their own data centers in the same region. If that tenant cancels 20% of their space, Switch's revenue could drop 10%. That's not growth — that's a structural vulnerability.
Contrarian: The Retail vs Smart Money Divergence
Retail investors are salivating over this IPO. They see "AI" and "infrastructure" and think it's a sure bet. But the smart money — the institutional traders and hedge funds I interact with in my copy trading community — are quietly shorting the space. They know that data center REITs are already pricing in a golden age that may never come.
The contrarian angle I'm betting on: the true value in AI infrastructure isn't in the concrete and power — it's in the code layer that orchestrates compute. Companies like CoreWeave (which I've written about before) are building software-defined data centers that can dynamically allocate GPU clusters across multiple facilities. Switch is just a landlord. When the AI hype cycle turns (and it will), the first to cut costs are these massive colocation leases. The flexible, code-first providers will survive; the fixed-asset behemoths will bleed cash.
I've lived through this before. In 2022, when the Terra collapse triggered a crypto winter, the biggest fallout was among overleveraged mining operators who had locked into long-term power contracts. Switch's balance sheet is better — they have strong cash flows from existing leases — but the underlying risk is the same: if demand growth slows, the fixed costs will crush them.
Takeaway: Actionable Price Levels and the Signal I'm Watching
The $80 billion valuation implies a share price of roughly $45 at the anticipated float. I am not going to buy the IPO. Instead, I'll wait for the lockup expiry (typically 180 days after listing) when insiders dump shares. My target entry is $25-$30, where the downside is limited by the replacement cost of their physical assets.
But the real signal isn't the price. It's the power cost. I am adding a custom alert on the Nevada wholesale electricity price index. If that index breaks above $45/MWh sustained for two consecutive months, I will short the stock heavily. That is the moment when the infrastructure mirage collapses.
We didn't buy the Layer2 hype in 2023, and we didn't buy the NFT royalty narratives in 2021. We are not buying this IPO either. Volatility is just unpriced risk — and Switch is carrying a lot of it.
