Reading · Crypto and infrastructure
Cell towers, maps, GPUs, storage: until now, big companies paid to build them. DePIN (“dee-pin”) tries to build the same infrastructure by having individuals set up devices or lend computing power, and receive crypto tokens in return. This page aims to help you understand the mechanism and separate the parts that work from the parts that don't yet.
In short
Think of everyday infrastructure. A carrier builds the towers, a mapping company drives cars around, and the cloud lives in a giant company's data center. All of it needs huge upfront capital, so few firms can do it.
DePIN's idea is to split that upfront cost thinly across people worldwide: someone puts a small radio on a windowsill, someone mounts a dashcam and drives, someone lends an idle GPU or spare disk. They supply the hardware and effort, and receive tokens.
The term spread around 2022, reportedly after a push from the crypto research firm Messari to replace earlier labels such as “MachineFi.” The blockchain's role here is a ledger that records who contributed what and pays rewards automatically.
The business model is usually described as a loop, often called the “token flywheel.”
If it works, the network starts on subsidy and becomes self-funding through revenue — a startup's discount-to-acquire-customers playbook, applied to infrastructure. Compared with traditional operators:
| Traditional operator | DePIN | |
|---|---|---|
| Who supplies equipment | The company (staff and contractors) | Many individuals and small operators |
| Where money comes from | Equity, debt, customer fees | Token issuance (and, later, revenue) |
| Record of contribution | Internal company systems | A public blockchain ledger |
| Strengths | Quality control, support, regulatory handling | Fast start, wide placement, low cost |
| Weaknesses | Heavy upfront capital, slow expansion | Uneven quality, spoofing, rewards tied to token price |
DePIN is not one product; it is an umbrella over separate projects, each in its own area. Here are representative ones, grouped by what physical resource they gather. No rankings, no prices.
| Area | Examples | What participants provide |
|---|---|---|
| Wireless | Helium | Small radios (LoRaWAN, later 5G cells for mobile service). Moved to the Solana blockchain in 2023 |
| Mapping | Hivemapper | Dashcam-style cameras that film roads as you drive, used to refresh a map |
| GPU / compute | Render Network, io.net, Akash | Idle GPUs and servers, aimed at 3D rendering and AI training and inference |
| Storage | Filecoin | Spare disk capacity for storing data |
| Vehicle data | DIMO | Your own car's driving and status data |
| Weather / positioning | WeatherXM, GEODNET | Weather stations and reference stations for centimeter-level positioning |
There are many more, and the field turns over quickly. DePINscan and Messari's reports keep running lists.
One pattern shows up: some areas have buyers that are easy to name (GPUs for AI, map data, weather data, positioning), while in others “who uses this, and why?” is still hard to answer. That is the next section.
The most important distinction when judging a DePIN project:
Rewards ≠ revenue
Rewards are tokens handed to participants, usually newly issued. Revenue is money that outside customers actually paid. Rewards are a subsidy; revenue is the result of doing business. They often get blended into “the network is growing,” but in the launch phase large rewards are almost a given.
What to watch is how much outside customers pay relative to the value of tokens going out as rewards. If that ratio grows, the flywheel is starting to turn. If rewards balloon while revenue barely moves, it is a business of distributing tokens, not yet an infrastructure business.
One way to read an example: Helium is known as an early network that spread LoRaWAN radios worldwide. For a long stretch, actual traffic was small compared with the number of devices, and the project then shifted its weight to 5G for mobile service, moving toward being a carrier with subscribers. “More devices” and “more use” are different things, and this is a case of a project steering to close that gap.
Another problem is proving that a device's location or a driving log is real. When rewards are on offer, someone will spoof locations and fake devices. Helium has had location-spoofing problems reported. Projects invest heavily in “proof of contribution,” and it remains an area of ongoing improvement.
If you are thinking of buying equipment or tokens, check at least these:
In Japan, the National Tax Agency treats the market value of crypto rewards at the time of receipt as taxable income (generally miscellaneous income). That can mean a tax bill arrives before you ever sell. Details depend on your situation, so check the agency's guidance or a tax professional. This page is not tax advice. Rules elsewhere differ.
For all those cautions, the idea is coherent. The marginal cost of infrastructure can drop if you pool resources that are sitting unused: a home GPU, spare disk space, a commuter's car. Paying people to bring them out is an extension of the sharing economy.
In an era when AI compute is scarce, gathering GPUs from many small owners lines up with real demand. If it works, part of what only big companies could build becomes something a crowd of small participants can build. But the road there involves a long stretch of reliance on subsidy. Keep both in mind and the news reads differently.
Three questions for reading the news