On a date that SpaceX has not disclosed, Ukrainian military units reported that their Starlink service had been restricted. No advance notice reached the affected commands. No published reason accompanied the degradation. The terminals that coordinate artillery fire, control reconnaissance drones, and link forward command posts to rear headquarters simply lost capacity. Polish Foreign Minister Radosław Sikorski responded in public, questioning whether Elon Musk was exercising political control over a communications layer that a sovereign state depends on to conduct a war. The event produced a predictable secondary reaction. Advocates of decentralized physical infrastructure networks, the blockchain sector known as DePIN, cited it as vindication. A single commercial entity, they argued, had demonstrated the precise failure mode that token-incentivized infrastructure was designed to eliminate.
The data does not support that conclusion. Data does not negotiate; it only reveals. And what the DePIN coverage metrics reveal is a gap between the decentralization claim and the decentralization reality that is wider than the Starlink event itself. I have spent the better part of eighteen years auditing the distance between what infrastructure protocols claim and what their on-chain records show. The distance, in this case, is measurable.
Context: Why a Satellite Restriction Became a Blockchain Argument
Starlink operates a low-Earth-orbit constellation of roughly seven thousand satellites. The network delivers broadband through consumer terminals. In the Ukrainian theater, those terminals became the backbone of tactical communication after terrestrial networks were destroyed. Command-and-control, drone telemetry, and artillery correction all route through commercial low-Earth-orbit links. This is not a marketing claim. It is an operational dependency, documented in procurement records and confirmed by the paying parties.
The dependency has a structural property that matters for this analysis. The service provider is a private company. The paying party, in part, is Poland. The using party is Ukraine. Three separate entities, three separate interests, and one switch that only one of them controls. When the provider adjusts availability, the user absorbs the consequence and the payer has no contractual remedy that executes in real time. The Foreign Minister's public challenge was an attempt to convert a private decision into a matter of public accountability. It failed, because the architecture does not contain an accountability mechanism. It contains a vendor relationship.
The blockchain industry read this event as a product demonstration. If battlefield communication ran on a permissionless network with no single operator, the argument went, no single operator could throttle it. DePIN protocols market exactly this property. Helium sells decentralized wireless coverage. Render sells decentralized compute. Filecoin sells decentralized storage. Each pitches the same thesis: replace the corporate chokepoint with a token-coordinated swarm of independent operators, and the chokepoint disappears.
The thesis is coherent. The execution is not. I audited the arithmetic behind it, and the arithmetic is where the claim breaks.
Core: The Four Premises of DePIN, Tested Against On-Chain Records
A decentralization claim is a testable proposition. It decomposes into four premises. First, that no single actor can unilaterally alter service availability. Second, that the operator set is genuinely distributed. Third, that the token incentive aligns operator behavior with network reliability. Fourth, that the physical layer cannot be captured by the same corporate structures it claims to replace. Each premise can be checked against public data. I checked all four.
Premise One: unilateral control.
Helium migrated its entire network state to Solana in 2023. The migration was presented as a scaling decision. It was also a consolidation of dependency. Every hotspot, every coverage proof, every reward calculation now settles on a single Layer 1 whose validator set, block production, and fee market are governed by a foundation and a validator cohort that most hotspot operators have never interacted with. The network removed its dependence on a corporate telecommunications provider and replaced it with dependence on a corporate-adjacent blockchain foundation. The switch moved. It did not disappear.
This is the first forensic finding, and it recurs across the sector. Decentralization claims are almost always assertions about the operator layer, while the control layer remains centralized. The operator layer is visible. It is the thing the marketing describes. The control layer is the upgrade key, the treasury multisig, the foundation board, the RPC provider list. It is less visible, and it is where unilateral action lives.
I encountered this pattern in 2020, when I analyzed the Compound governance distribution. The protocol presented itself as governed by token holders. The distribution algorithm concentrated effective control in a small cohort through a logic flaw that the market celebrated as organic growth. Fifty percent probability, I estimated, of governance capture. Three security firms cited the memo later that year. The mechanism is identical here. The operators are many. The deciders are few.
Premise Two: distributed operators.
Helium's active hotspot count peaked above nine hundred thousand and has since contracted to a range that public dashboards place far below that peak. The contraction itself is not disqualifying. Networks reprice. What matters is the geographic distribution of the survivors. Coverage concentrates in metropolitan areas where density produces the highest reward-per-hotspot. Rural and maritime coverage, the segments where centralized carriers are weakest and where decentralized claims would deliver the most marginal value, remains sparse. The network reproduces the exact coverage map of the incumbent carriers it claims to disrupt.
This is not a failure of effort. It is a failure of incentive geometry. Token rewards scale with proof-of-coverage density, and density is cheapest to manufacture where population is highest. The protocol rewards the operator for going where the return is already highest. It does not reward the operator for going where the network is needed most. A token cannot purchase a behavior that its own reward curve penalizes.
I learned the limits of incentive design the hard way in 2021, when I audited a generative art project and missed a minting exploit that drained two million dollars within hours of launch. I spent the following three months reconstructing the attacker's transaction history. The lesson was not that audits fail. The lesson was that incentive models produce the behavior the math rewards, and the math does not read the marketing copy. Helium's coverage map is the marketing copy's refutation.
Premise Three: token incentive aligns reliability.
DePIN tokens serve two functions simultaneously. They reward operators, and they represent claims on future network value. When the token price falls, operator revenue falls, and marginal operators exit. When operators exit, coverage degrades. When coverage degrades, the network's utility falls, and the token price falls further. The loop is reflexive, and it runs in both directions.
Starlink's network does not contain this loop. Its revenue comes from subscriptions denominated in fiat, priced against a cost structure the operator controls. When the market sours, the operator cuts capital expenditure and continues service. The service is not collateralized by a speculative asset. The DePIN network is. In a sustained drawdown, the decentralized network's reliability is a function of a price chart. The centralized network's reliability is a function of a balance sheet. The balance sheet is more stable than the chart. This is not a philosophical point. It is an arithmetic one.
The reflexive loop is not hypothetical. It is the mechanism by which every token-incentivized network I have examined either sustained itself through subsidy or contracted toward the regions where subsidy was cheapest. The subsidy is the product. When the subsidy ends, the product is what remains. For most DePIN networks, what remains is a thin layer of high-density operators in three or four cities.
Premise Four: the physical layer resists capture.
This is where the Starlink comparison becomes genuinely instructive, and where the DePIN thesis encounters its hardest constraint. Starlink's restriction was possible because the operator controlled the space segment, the ground segment, and the terminal firmware simultaneously. The restriction executed at the terminal. The terminal is a piece of hardware with a vendor-controlled software stack that authenticates against a vendor-controlled gateway. There is no configuration a user can apply to override the vendor's decision. The switch is physical, and the physics is owned.
A decentralized network that depends on a physical layer it does not own inherits the same vulnerability one layer down. Helium's coverage depends on hotspots connected to the public internet through incumbent internet service providers. If an ISP withdraws service, the hotspot goes dark. The decentralized network's last mile is the centralized network's first mile. The dependency is not eliminated. It is relocated and obscured.
This is the second forensic finding. The physical layer is the un-decentralizable residue. Tokens can coordinate software. They cannot coordinate radio spectrum, orbital slots, or the fiber that connects a hotspot to a backbone. Those assets are allocated by states and owned by corporations. A network that routes value through token incentives while routing packets through state-allocated spectrum has decentralized its accounting and centralized its physics.
The Layer 2 sector offers the cleanest parallel. Post-Dencun blob data pricing gave rollups a temporary cost advantage. The advantage is already eroding as blob space approaches saturation. When it saturates, rollup fees reprice upward, and the rollups that marketed themselves as cheap become expensive again. The lesson is structural, and it generalizes. A cost or capability advantage inherited from a subsidized or underutilized resource is not a durable advantage. It is a deferred cost. DePIN's decentralization advantage is similarly deferred. It holds only as long as the underlying physical layer remains available on favorable terms, and the terms are set by parties the network does not control.
The Complexity Tax and the Compliance Hedge
Two additional variables determine whether the DePIN sector can close its verification gap, and neither appears in the coverage dashboards.
The first is complexity. A decentralized wireless network requires operators to configure hardware, manage wallets, stake tokens, monitor reward curves, and respond to protocol upgrades. The operational burden is not incidental. It is the barrier that determines how many independent operators can realistically participate. Uniswap V4's hook architecture made the DEX programmable and, in doing so, raised the technical floor for anyone building on it. The same dynamic applies here. Every increment of programmability raises the expertise required to operate, and expertise concentrates. A network whose operators must be specialists is a network whose operators are few. The complexity tax is paid in centralization.
The second is regulatory positioning. PayPal launched PYUSD not to escape regulation but to occupy a seat at the table before the table was set. The strategic logic was accommodation, not evasion. DePIN protocols that touch spectrum, satellite links, and cross-border data flows will encounter the same regulatory perimeter that the Starlink operator navigated. The protocols that survive will be the ones that accept the perimeter and structure themselves around it. The ones that market decentralization as a regulatory escape hatch will discover that the physical layer is jurisdictionally visible in a way that a smart contract is not. A hotspot has a location. A location has a regulator.
I documented this pattern in 2025, when I analyzed the custodial solutions behind the approved spot ETFs. Eighty percent of custody providers relied on legacy banking infrastructure with security postures that had not kept pace with the assets they held. The marketing described decentralization. The custody described a bank vault with an outdated patch level. Twelve specific compliance vulnerabilities, documented, and the industry ignored the report. It became required reading for institutional risk officers anyway. The gap between the narrative and the infrastructure is the sector's defining characteristic, and it does not close because a narrative is repeated.
Contrarian: What the DePIN Bulls Got Right
The forensic case above is a case against the decentralization claim. It is not a case against the demand.
The Starlink event proved something the DePIN bulls have been asserting correctly for years. Critical infrastructure concentration is a real strategic vulnerability, and the vulnerability is not theoretical. A sovereign state's military capability was degraded by a private operator's unilateral decision, executed without notice, and the affected state had no remedy. That is a verified failure mode. The demand for alternatives is not manufactured. It is demonstrated.
What the bulls get wrong is the conclusion they draw from the demand. They conclude that the alternative must be decentralized, and that a token is the mechanism. The demand does not specify the mechanism. The demand specifies the property: no single actor can unilaterally degrade service. That property can be delivered by a consortium of states, by a multilateral procurement framework, by a treaty-bound operator, or by a public utility structure. None of these require a token. Several of them are more likely to deliver the property than a token network, because they can compel the physical layer that a token network can only rent.
The bulls also get the timeline right and the difficulty wrong. Rebuilding critical communications infrastructure is a decade-scale undertaking. European sovereign constellation programs are underway and will take years to reach operational capacity. In the interim, the dependency persists, and the chokepoint remains. A decentralized network that launched tomorrow would inherit the same dependency on the same physical layer, at a smaller scale, with a less reliable revenue base. The bulls are correct that the problem is urgent. They are incorrect that the token is the answer. The urgency is the argument for institutional commitment, not for speculative coordination.
Takeaway
The Starlink restriction was a demonstration of centralized infrastructure risk. The DePIN sector read it as a mandate. The mandate is real. The mechanism is unverified.
Four premises, four findings. Unilateral control relocated rather than eliminated. Operator distribution concentrated by incentive geometry. Reliability collateralized by a price chart. The physical layer un-decentralizable and jurisdictionally visible. None of these findings is fatal in isolation. Together they describe a sector whose decentralization claim is an assertion about the layer the customer sees, while the layer that decides remains as centralized as the incumbent it critiques.
The forward question is not whether decentralized infrastructure can work. The forward question is whether the sector can survive an audit of its own claims, conducted against its own records, without the audit being reframed as bearish propaganda. I published a forty-billion-dollar analysis in 2022 and was told it was noise. Regulators used it as evidence. The pattern repeats because the arithmetic does not change. Verify the control layer. Count the deciders. Read the reward curve. The dashboard will tell you what the marketing will not. Data does not negotiate; it only reveals.