It arrived without fanfare, buried in a press release from a crypto-adjacent outlet: TPG, the private equity giant, is in exclusive talks to acquire Netrality, a regional data center operator, for a staggering $3 billion. The numbers are dryly impressive — seven facilities, 24+ megawatts of power, strategic positions in Philadelphia and St. Louis — but the signal beneath the numbers is anything but dry. We are watching the fossilization of a centralized model, a model that thinks it can buy its way into the future while the decentralized physical infrastructure network (DePIN) quietly rewires the planet’s nervous system.
Curating the soul in a world of derivative clones.
Context: The Cathedral of Compute
Data centers are the cathedrals of our digital age. They house the prayers of every cloud service, every AI inference, every streaming binge. The industry is dominated by a handful of giants — Equinix, Digital Realty, CyrusOne — who have perfected the art of building massive, power-hungry fortresses. They are venture-backed, publicly traded, and deeply integrated into the web’s backbone. Their moat? Switching costs so high that customers would rather pay premium rents than endure a single minute of downtime during a move. Their network effect? Local, not global — the more carriers in one meet-me-room, the stickier the ecosystem.
Netrality is a mid-tier player, not a hyperscaler. Its 24MW capacity is a fraction of what AWS or Google self-builds in a single region. Yet TPG is willing to pay $3 billion. Why? Because PE firms see data centers as the new toll roads: stable, inflation-hedged, with predictable cash flows and a monopoly-like grip on their immediate geography. The playbook is simple: buy, optimize, lever up, resell to a bigger sucker (or take public as a REIT). The narrative is one of “AI demand” and “digital transformation.” But beneath that narrative lies a brittle assumption: that centralized control of compute is the only way to serve the coming wave of artificial intelligence.
I have spent years staring at the governance of decentralized systems. During my time analyzing over 500 voting proposals for MakerDAO, I learned that the most efficient systems often hide the deepest inequities. Centralized infrastructure is efficient, yes, but efficient for whom? TPG’s acquisition is a bet that the cathedral will remain standing, that the rent-seekers will continue to collect tithes from every query processed in their halls. It is a bet I believe is wrong.
Core: The DePIN Alternative — Compute as Commons
Let us now examine the technical and economic architecture that TPG is implicitly rejecting. Decentralized Physical Infrastructure Networks — DePIN — represent a radical reimagining of how compute, storage, and bandwidth are provisioned. Instead of a single entity raising billions to build a fortress, a DAO or protocol issues tokens to incentivize thousands of individuals to contribute spare capacity from their homes, offices, or existing data centers. Think Filecoin for storage, Render Network for GPU compute, Helium for wireless coverage, and Akash Network for cloud compute.
The unit economics are fundamentally different.
A centralized data center operator like Netrality must earn a return on its enormous capital expenditure — land, building, power infrastructure, cooling, security, compliance. This fixed cost is passed on to renters in monthly fees, often with a markup that ensures a 20%+ EBITDA margin. The customer is locked into a long-term contract with painful penalties for early exit. The switching cost is intentionally high. That is not a market; it is a tollbooth.
In contrast, a DePIN network distributes the capital expenditure across a global pool of participants. A single GPU owner in Chengdu can contribute to Render and earn RNDR tokens without needing to raise a penny. The protocol coordinates supply and demand through smart contracts, not through a sales team. The result is a market that is far more elastic, far more competitive, and far more resistant to censorship. The network effect is global: every new GPU added increases the likelihood that a job finds a match, which attracts more demand, which attracts more supply. This is a virtuous cycle that the centralized model cannot replicate because it is constrained by geography and balance sheets.
During my work architecting governance for CivicChain, a DAO focused on municipal data sovereignty, I saw first-hand how the presence of decentralized storage can shift power from infrastructure monopolists back to communities. We did not need to buy a data center; we simply pinned our data to IPFS and paid storage providers in DAI. The cost was a fraction of what AWS would have charged. The resilience was higher because the data was replicated across jurisdictions. And most importantly, no single entity could unilaterally delete our records or change the terms of service.
Power is not just about who owns the hardware. It is about who defines the rules of the game.
TPG’s acquisition of Netrality is a bet that the rules will remain written in legal contracts and enforced by property rights. DePIN bets that the rules can be written in code and enforced by cryptographic consensus. The former is a system of permission and identity; the latter is a system of permissionless participation and pseudonymity. One leads to gatekeeping; the other leads to self-sovereignty.
Let us be quantitative. Netrality’s 24MW of power can support roughly 20,000 high-end GPU servers (assuming 1.2kW per GPU server). That translates to a potential AI compute capacity of, say, 200 petaflops. The Render Network today, with a fraction of that total power, already aggregates GPUs from thousands of individual users. The difference is that Netrality’s capacity is idle unless fully rented, while a DePIN network can dynamically adjust utilization. The centralized model is a bucket; the decentralized model is a river.
The core insight: centralized data centers achieve scale by concentrating resources, while DePIN achieves scale by distributing incentives.
In a world where AI inference is becoming ubiquitous — every smartphone, every IoT sensor, every autonomous vehicle will eventually run local models with occasional cloud calls — the demand for compute will become far more granular and far more variable. A centralized data center is like a medieval fortress built for a siege that never ends. A DePIN network is like a bazaar that grows organically to meet the needs of the day. The fortress looks imposing, but the bazaar can reconfigure itself overnight.
Contrarian: The Pragmatic Test — Why TPG Might Actually Win (And Why That Scares Me)
Let me pause and acknowledge the counterargument. I have never feared the weakness of my own position. In my 2017 essay “Tokenized Equity as Digital Citizenship,” I argued that blockchain was not merely a ledger but a tool for economic empathy. I was laughed at by my Polymath colleagues, who preferred to talk about liquidity and compliance. But I held my ground because I believed in the moral arc of technology. Today, I must apply the same honesty to my own evangelism. Could TPG’s $3 billion bet be the smart move?
The answer is a qualified yes — if the world remains centralized. If regulatory frameworks continue to favor institutional players with deep pockets. If latency-sensitive applications (like autonomous vehicle coordination or high-frequency trading) continue to demand physical proximity to fiber backbones. If enterprise compliance requires SOC 2 audits that small DePIN providers cannot afford. Then TPG will earn a healthy return on their data center assets. They will upgrade the cooling systems, cut the ribbon on a few high-density AI pods, and sell to a bigger fish at a 4x multiple. The playbook works.
But here is the blind spot: the same forces that drive centralization also drive fragility. A single data center can be knocked offline by a squirrel chewing the wrong cable. More importantly, a single government can compel a data center operator to cease serving a particular customer. The Tornado Cash sanctions proved that code can be criminalized, but they also proved that code is harder to suppress than people. A DePIN network cannot be easily shut down because there is no single operator to serve with a subpoena. The network is the operator.
During the bear market of 2022, I interviewed 50 builders who stayed despite their losses. The ones who survived were not the ones who hoarded cash; they were the ones who had built systems that could not be captured.
TPG’s acquisition is an attempt to capture. They see the data center as a castle with a moat. What they do not see is that the drawbridge is being raised not by them, but by the very technology they are buying. The same AI workloads that fill their racks today will soon be executable on a globally distributed mesh of nodes, each one incentivized by a token. When that day comes, the 24MW castle will become a museum — a beautiful artifact of a bygone era when compute was hoarded rather than shared.
Takeaway: The Future Does Not Need Permission
The TPG-Netrality deal will likely close. TPG will make its money. Netrality’s employees will get payouts. The press will call it a victory for infrastructure investors. But the story that matters is not the one in the press release. It is the one unfolding in the open-source repositories and Discord servers where a new kind of infrastructure is being built — infrastructure that does not ask for permission, does not demand a contract, and does not require a middleman to connect supply with demand.
I write this not as a detached observer, but as someone who has felt the weight of centralization. In 2021, I curated The Ethereal Archive, a DAO of 120 members, to preserve digital artifacts without relying on a single custodian. When the market crashed, our archive survived because it was pinned to distributed storage. The centralized marketplaces deleted listings; our records remained. That experience taught me that resilience is not a feature of code alone — it is a feature of architecture. A centralized architecture can be resilient within stable conditions, but only a decentralized architecture can adapt to upheaval.
The ultimate measure of any infrastructure is not its efficiency when everything goes right, but its survivability when everything goes wrong.
TPG’s $3 billion is a vote for efficiency. Mine is a vote for survivability. The future of compute will require both, but the balance is shifting. Every new DePIN node, every new IPFS pin, every new GPU added to a decentralized network tilts the scale a little further away from the cathedrals. The cathedrals might grow bigger, but the number of people inside them will shrink. The bazaar, by contrast, will never close.
Curating the soul in a world of derivative clones.
And so, as I watch TPG attempt to lock down a piece of the digital world, I do not feel fear. I feel a quiet urgency. The urgency to build more, to write more, to teach more about the alternative. Because the future is not a fortress to be bought. It is a network to be grown. And it will not be grown by private equity. It will be grown by communities aligned by incentives and united by a shared belief that the power to compute should not be hoarded, but liberated.