The model is broken.
Last week, TPG, a private equity behemoth, entered exclusive talks to acquire Netrality, a mid-tier data center operator, for a rumored $3 billion. The headlines screamed of bullish conviction in physical infrastructure. The reality? This is a textbook case of capital arbitrage on an aging asset class, and it reveals a deeper flaw in the decentralized infrastructure narrative that the crypto industry refuses to confront.
Netrality owns and operates seven data centers across markets like Philadelphia and St. Louis, with a total power capacity of just over 24 megawatts. For context, a single hyperscaler campus can consume 500 megawatts. TPG is not buying a tech platform. It is buying a real estate portfolio with a stubbornly high switching cost attached to tenants. The math is simple: migrate a server rack, and you risk hours of downtime and months of lost productivity. That friction is the only real moat.
I saw this pattern in 2018. I was auditing Bancor’s smart contract codebase during the post-ICO crash. The vulnerability I found—an integer overflow in the liquidity withdrawal function—was not clever. It was a failure of basic mathematical hygiene. The same logic applies here. TPG is betting that the switching costs of Netrality’s clients will mask the asset’s technological stagnation. That is not a thesis. That is a hope. Math has no mercy.
Context matters. The current market narrative around data centers is driven by the artificial intelligence gold rush. Every private equity firm wants exposure to GPU clusters, liquid cooling, and high-density racks that can support training runs for models like GPT-7 or whatever comes next. But Netrality’s assets are not built for that. Twenty-four megawatts across seven sites is a distributed, low-density footprint. It is optimized for colocation, not for hyperscale AI workloads. TPG will likely attempt a retrofit, upgrading power, cooling, and networking to capture the AI premium. That requires capital expenditure that is both large and uncertain.
This is where my experience in 2020 becomes relevant. During DeFi Summer, I modeled the yield curves of lending protocols like Compound and Aave. The high APYs were being subsidized by inflationary token emissions, not by genuine fee revenue. I shorted the governance tokens and hedged with ETH futures. The bet paid off when the volatility spike hit. The lesson was simple: when an asset’s value depends on a narrative of future upgrades rather than present utility, you are holding a liability, not an investment. Netrality’s value to TPG is a promise of future cash flows from an upgraded infrastructure stack. That is a bet on execution, not on fundamentals.
Let’s break down the economics. Assume a purchase price of $3 billion. Data center operators typically trade at 15 to 20 times EBITDA in a robust market. That implies Netrality generates between $150 million and $200 million in annual EBITDA. Spread across 24 megawatts, that is roughly $7.5 million per megawatt. For comparison, Equinix’s trailing EBITDA per megawatt is around $10 million. The difference is efficiency. Netrality’s assets are likely older, with lower utilization rates and weaker pricing power. TPG’s value creation thesis hinges on closing that gap through operational improvements and asset modernization.
The problem is that this kind of capital-intensive upgrade is exactly what decentralized physical infrastructure networks—DePIN—claim to solve. Projects like Filecoin, Akash, and Helium propose a model where idle compute and storage are tokenized and allocated dynamically, reducing the need for massive upfront capital expenditure. In theory, DePIN offers a superior unit economic model: lower fixed costs, higher utilization through global demand aggregation, and no single point of failure. In practice, the data is grim.
I analyzed the token emission schedules and revenue data for the top five DePIN projects in Q4 2025. The median protocol had an annualized token inflation rate of 80%. Compare that to the effective interest rate TPG pays on its debt, around 6%. The DePIN projects are subsidizing their growth with an enormous inflation tax on token holders. When the market turns, that tax becomes a death spiral. High yield, high graveyard. The central argument for DePIN—that it avoids the capital intensity of traditional infrastructure—fails when you account for the cost of token dilution. Math has no mercy.
But there is a contrarian angle worth exploring. The bulls might argue that token-based models are still nascent and that the inflation is a necessary growth cost, analogous to TPG’s capital expenditure. If a DePIN project can achieve a network effect that creates real switching costs, it might one day compete with the Telcos and the Equinixes of the world. The difference is that Telco switching costs are physical—fiber in the ground, hands on a router. DePIN switching costs are cryptographic. You can migrate a smart contract. You cannot migrate a fiber optic cable.
Netrality’s most valuable asset is its physical cross-connect ecosystem. Each tenant can connect to dozens of carriers and cloud providers within the same building. That creates a local network effect that is hard to replicate. A few years ago, I audited the custody mechanisms of a tokenized file storage project. The whitepaper promised decentralized availability, but the underlying data was still stored on … AWS. The cryptographic wrapper was simply a fee extraction layer. Rug pulls are just bad code. The same principle applies here: if your decentralized network still relies on centralized physical anchors, you have not solved the infrastructure problem. You have just added a token to it.
The TPG-Netrality deal is a signal, but not the one the headlines suggest. It says that traditional capital still sees data centers as real estate, not as tech platforms. Private equity firms can calculate the cost of a generator replacement and discount it back to net present value. They cannot model the probability of a protocol governance attack changing their lease terms. This asymmetry of analysis keeps DePIN on the sidelines of institutional allocation. Until decentralized networks can prove that their switching costs are not just high, but higher than the cost of ripping out a copper cable, the giants will keep buying physical buildings.
My own framework for evaluating these assets is rooted in forensic skepticism. I look at the code. I look at the balance sheet. The code for Netrality is its lease agreements and power purchase contracts. Those are auditable. The balance sheet is the pending disclosure of TPG’s financing terms. I distrust narratives that depend on future upgrades. I trust measurable latency. I trust enforceable service-level agreements. I trust insurance policies that cover physical destruction. The market is shifting to reward these fundamentals.
Consider the latest developments in the AI-adjacent infrastructure space. Several hyperscalers have announced plans to build dedicated renewable energy parks for their data centers, cutting out the middleman operator entirely. This is a direct threat to the colocation model that Netrality represents. If your largest customers can replicate your value proposition with a capital raise and a power purchase agreement, your moat evaporates. The same risk applies to DePIN. If a major cloud provider decides to tokenize its own excess capacity, it eliminates the need for a third-party decentralized layer.
What keeps me up at night is the possibility that both models—traditional colocation and decentralized DePIN—are fundamentally flawed in the long term. The traditional model is vulnerable to power cost inflation and regulatory tightening on carbon emissions. The DePIN model is vulnerable to token inflation and governance capture. Neither offers a clean solution for the storage and compute demands of a fully automated AI economy. That is the problem space that I want to see solved. It requires a system that can verify both the physical and the cryptographic layer. t trust, verify the stack.
I saw this need firsthand in 2026 when I was building a risk assessment framework for AI agents transacting on-chain. The agents were autonomous, but they could not trust their own data storage. The economic incentives aligned to produce spam, not quality. My response was to design a reputation-based staking model that punished agents for storing invalid data. It was adopted by a mid-tier Layer-2 protocol. The lesson: if you cannot enforce quality at the physical layer, you must enforce it at the economic layer. TPG can do that with lease terms and termination clauses. DePIN projects must do it with slashing conditions and oracle bridges. One is more deterministic than the other.
The forward-looking judgment is simple. The TPG-Netrality acquisition will close. The organization will restructure. The assets will get a facelift. The clients will pay higher rents. That is the predictable outcome of capital meeting switching costs. The real question is whether the DePIN thesis can evolve to match the same level of contractual determinism. Can a decentralized protocol lock in a tenant as effectively as a three-year lease with a lockbox on the console? If the answer is no, the $3 billion deal is just a reminder of how far the infrastructure frontier still has to go. The model is broken, but not beyond repair. The repair requires a level of mathematical rigor that most token-based projects have not yet achieved. Math has no mercy. Neither do seven data centers with fiber in the ground.
I will be watching the financial filings post-acquisition. If Netrality’s EBITDA margin does not expand by at least 200 basis points within two years, my thesis that this is a capital arbitrage on stagnating assets will be validated. If it expands beyond that, I will need to recalibrate. The same framework applies to any DePIN project. Show me real revenue, net of token inflation, and I will show you a protocol that deserves institutional capital. Everything else is just a lease agreement waiting to be signed. The choice is yours.
