What is DePIN? How decentralized physical infrastructure networks work, from Helium to XYO

Building physical infrastructure has always been a capital problem. A wireless carrier spends billions on towers before it signs its first customer. A mapping company sends fleets of camera cars down every road and then has to do it again a year later because the map has gone stale. Decentralized physical infrastructure networks, known as DePIN, attack that problem from the other side. Instead of one company raising the money and owning the hardware, a protocol pays a token to anyone who installs a device and provides a useful service, and customers pay to use the network that results. The idea has produced some of crypto’s most concrete businesses and some of its most expensive disappointments. This guide explains how DePINs work, walks through Helium, XYO and the other networks that define the category, and sets out how to tell a working network from a subsidy.

What DePIN means

The term was popularized by the research firm Messari in late 2022 to group projects that had previously been called “token-incentivized physical networks,” “proof of physical work” or “EdgeFi.” A DePIN has three parts: a supply side of independent operators who own and run hardware, a demand side of customers who pay for what that hardware produces, and a blockchain that measures each contribution and pays for it in a native token.

Messari’s framework splits the category in two. Physical resource networks deploy hardware tied to a specific location, where the value depends on where the device sits: wireless coverage, street-level imagery, weather readings, satellite positioning corrections, energy. Digital resource networks pool fungible resources where location matters much less: compute, GPU time, storage and bandwidth. A Helium hotspot in an empty field is worth little, while a spare GPU is worth roughly the same whether it sits in Texas or Tallinn. That distinction shapes everything from how rewards are designed to how easy the network is to fake.

How a DePIN works: the flywheel

Almost every DePIN follows the same sequence, sometimes called the DePIN flywheel.

1. Bootstrap supply. Early in a network’s life there are no customers, so the protocol pays operators in newly issued tokens for deploying hardware and proving it works. This is the subsidy that replaces a company’s upfront capital expenditure.
2. Reach useful coverage. Once enough devices are in place, the network becomes usable: a map is current, coverage is continuous, enough GPUs are online to run a job.
3. Sell to customers. Businesses or consumers pay for the service, usually in dollars or stablecoins, which the protocol converts into token demand.
4. Shift rewards from emissions to revenue. As paid usage grows, operators earn more from real demand and the network relies less on inflation. In theory, a rising token price attracts more operators, which improves the service, which attracts more customers.

The flywheel also runs in reverse. If customers never arrive, operators are paid only in a token whose supply keeps growing. They sell to cover hardware and electricity costs, the price falls, rewards in dollar terms shrink, operators switch off, and coverage degrades. Many first-generation DePINs went through exactly that cycle in 2022 and 2023.

Proving the work

Because rewards are paid for physical activity that a blockchain cannot see, every DePIN needs a way to verify that a device is real, is where it claims to be and did what it claims. Helium uses radio “proof of coverage” challenges between nearby hotspots and, for its Wi-Fi network, actual data transferred. Mapping networks check imagery against GPS, timing and other contributors’ drives. Compute networks benchmark hardware and verify completed jobs. Spoofing is a constant problem: Helium spent years fighting fake hotspot locations, and every network that pays per device attracts people who try to emulate devices. The quality of the verification layer is one of the least visible and most important parts of any DePIN.

Burn-and-mint and other token designs

The most common way to link customer revenue to the token is burn-and-mint equilibrium. Customers pay a fixed dollar price for the service, and the protocol buys and burns the token, or requires the token to be burned to create usage credits, while separately minting new tokens to reward operators. Helium’s version is the cleanest example: Data Credits cost a fixed $0.00001 each and can only be created by burning HNT, so every dollar of network usage permanently removes HNT from supply. Render uses a similar model. When burns exceed new issuance, supply shrinks. Other networks use buybacks from fiat revenue, as GEODNET does, or route fees to stakers, or simply keep revenue in a company and leave the token as a reward instrument with no direct claim. Which model a network uses determines whether the token benefits from the business at all.

The main DePIN categories

Helium: the network that defined the category

Helium is the oldest and most studied DePIN, and its history contains nearly every lesson the sector has learned. The company was founded in 2013 by Amir Haleem, Shawn Fanning and Sean Carey and launched its token-incentivized network in 2019. The pitch was simple: people would buy a hotspot for a few hundred dollars, plug it in at home and earn HNT for providing long-range, low-power coverage for Internet of Things devices using the LoRaWAN standard.

Supply arrived faster than anyone expected. Hundreds of thousands of hotspots went up worldwide, making it one of the largest LoRaWAN networks ever built. Demand did not follow at the same pace. Sensor customers were few, most hotspot income came from token emissions, and when HNT fell sharply in 2022 the economics of the IoT network collapsed for many operators. Helium also faced a credibility problem when it emerged that some of the brand-name customers it had cited were not active users of the network.

The pivot to mobile and carrier offload

Helium’s operator, Nova Labs, rebuilt the network around a much larger market: mobile data. In April 2023 the network moved from its own blockchain to Solana. It launched Helium Mobile, a phone plan that combined T-Mobile’s network with Helium’s own hotspots, and in 2024 shifted from CBRS radios to Wi-Fi hotspots that carriers could use to offload traffic. In January 2025 a governance vote returned all rewards to HNT, retiring the separate IOT and MOBILE tokens that had been introduced in 2022.

The decisive step was selling capacity to the incumbents. In April 2025 AT&T announced a partnership under which its customers automatically connect to Helium hotspots, with Helium reporting connection quality back to AT&T every few minutes so the carrier can decide whether to keep a phone on Wi-Fi or move it to cellular. At that time Helium had more than 90,000 hotspots across the US and Mexico serving more than 600,000 users a day. By mid-2026 the network was offloading about 274 terabytes a month at roughly 2,485 restaurant locations across 34 US chains alone, including KFC, Taco Bell and Starbucks.

Selling the phone business

On June 2, 2026, Helium Mobile was acquired by Noble Mobile, the carrier founded by Andrew Yang, for undisclosed terms. Subscribers kept their numbers and access to T-Mobile’s 5G network. The Helium network and HNT were not part of the sale: Nova Labs said it would concentrate on the network business, which it called the larger opportunity, rather than run a consumer phone service. Around the same time, Haleem stepped down as chief executive and became chairman, and Mario Di Dio, who had led the network business and the carrier partnerships, took over as CEO. At the time of the sale the network had nearly 139,000 mobile hotspots and was generating more than $47,000 a day in revenue.

Helium’s numbers show both the promise and the gap. Messari found Helium’s on-chain revenue rose about eightfold during 2025 while HNT fell 77 percent over the same period. Usage, in other words, was compounding while the token was not. HNT’s thin market has made it volatile: in late August 2026 it rose about 170 percent in a single weekend, a move driven mainly by short liquidations rather than new buying. On the regulatory side, the SEC sued Nova Labs in January 2025; in April 2025 it dropped the claims that HNT and the hotspots were unregistered securities, and Nova Labs settled the remaining claims about statements on its partnerships for $200,000 without admitting or denying them.

XYO: location data and proof of origin

XYO Network was founded in 2018 by Arie Trouw and Markus Levin to answer a question that matters more every year: how can a smart contract, or now an AI system, trust information about the physical world? Its original design was a location oracle. Devices called sentinels observe and record data, bridges relay it, archivists store it and diviners answer queries, with each interaction recorded through “bound witness” exchanges in which two devices cryptographically sign that they observed each other at a given time and place. Chaining those records produces a “proof of origin”: evidence of where a piece of data came from, in what order it was created and that it has not been altered.

COIN and a consumer-scale network

XYO’s distribution came through the COIN app, which turns a smartphone into a network node that contributes location data and earns rewards, a process XYO calls geomining. The app has passed 10 million installs worldwide and has paid out more than 5 billion XYO in rewards. XYO says about 80 percent of its users come from outside crypto, which makes it unusual in a sector where most operators are token holders first. The company reported $8.8 million in revenue for 2024. Because a phone is already in everyone’s pocket, XYO avoided the hardware cost that slows most physical networks; the trade-off is that phone-based location data is easier to fake than data from a dedicated device, which is why verification sits at the centre of its design.

XYO Layer One and the AI pivot

On September 16, 2025, XYO launched XYO Layer One (XL1), a blockchain built specifically for data-heavy applications rather than for trading or general smart contracts. It uses a dual-token model. XYO, with a fixed supply, is used for staking, governance, DePIN rewards and security. XL1 is the chain’s gas and transaction token and is earned by staking XYO. The chain introduces mechanisms XYO calls Lookback Window, Step Hash and Proof of Perfect, designed to keep performance steady as data volume grows. In 2026 XYO reported a two- to fivefold throughput increase, both tokens were listed on Crypto.com in July 2026, and in September 2026 XYO connected XL1 to Autonomys’ Auto Drive storage so that large datasets live on Autonomys while a compact, signed record on XYO Layer One proves where they came from.

The strategic shift is toward AI. As autonomous agents and robots act on behalf of people and companies, XYO’s argument is that they need an audit trail: verifiable evidence of what data they used and what they did. Trouw and Levin have described this on Crypto Coin Show as moving from proving data to proving action. Whether that becomes a paying market is the key question for XYO. Like most DePIN tokens, XL1 has traded well below its launch levels, and the value of the chain will depend on developers building data provenance into products that customers pay for.

Other DePINs worth knowing

Hivemapper and Bee Maps: crowdsourced street maps

Hivemapper, now operating as Bee Maps, pays drivers in HONEY tokens for capturing street-level imagery with dedicated dashcams. Its AI processes the footage into map features such as speed limits, road signs and construction zones, and because contributors drive the same roads every day the map refreshes far faster than a traditional survey fleet can manage. Within two years of launching in November 2022 the network had mapped more than 17 million unique kilometres, about 29 percent of the world’s roads. In October 2025 Bee Maps raised $32 million led by Pantera Capital and introduced a $19-a-month plan for its Bee camera in place of an upfront price of about $600. Named customers include Lyft and Volkswagen’s robotaxi program, and the network’s annualized revenue was estimated at roughly $18 million in early 2026.

GEODNET: centimetre-level positioning

Standard GPS is accurate to a few metres. Drones, farm machinery, survey equipment, robots and autonomous vehicles need centimetre accuracy, which requires a dense network of fixed reference stations that broadcast real-time kinematic (RTK) corrections. GEODNET pays operators to install those stations. As of September 30, 2026 it had more than 22,000 base stations in 174 countries, more than 150 enterprise customers and $2.8 million of revenue in the third quarter of 2026. It directs about 80 percent of revenue to buying back and burning its GEOD token, one of the clearest links between business revenue and token value in the sector.

Render, Akash, io.net and Aethir: compute for AI

Compute is the largest DePIN category by value because it sells into the AI boom. Render Network, built by the graphics company OTOY, began by matching 3D artists with idle GPUs and moved its token from Ethereum to Solana in 2023; it uses burn-and-mint so that rendering jobs burn RENDER. Akash runs an open marketplace on its own Cosmos-based chain where anyone can rent out or rent servers and GPUs, and its utilization has run above 80 percent with usage growing more than fourfold year on year. io.net and Aethir aggregate GPUs from data centres, crypto miners and individuals and sell them as cloud capacity for AI training and inference, with Aethir focusing on enterprise and gaming customers. The pitch is price: these networks rent GPUs well below the major clouds by tapping capacity that would otherwise sit idle. The challenge is reliability, since a training run that spans many machines needs fast interconnects that scattered consumer hardware cannot provide, so most real demand is for inference, rendering and smaller jobs.

Filecoin and Autonomys: decentralized storage

Filecoin, built by Protocol Labs and live since October 2020, pays storage providers to keep data and uses cryptographic proofs to check they are still storing it over time. It built one of the largest pools of decentralized storage capacity, but only around a third of it is used, a reminder that supply is easy to subsidize and demand is not. Newer networks such as Autonomys focus on permanent storage for AI and machine data, which is where XYO has chosen to put the bulk of its datasets.

Grass: bandwidth for AI data

Grass pays users for sharing unused home internet bandwidth through a browser extension or app. That bandwidth is used to collect public web data, which is sold to AI companies for model training. It became one of the highest-earning DePINs in 2025, with about $33 million of revenue paid in USDC, most of it in the fourth quarter. Token holders did not receive any of that revenue, and in July 2026 the project put revenue capture mechanisms to a governance vote, a good illustration of the gap that can exist between a DePIN business and its token.

DIMO: vehicle data

DIMO lets drivers connect their cars, through an app or a plug-in device, and share telemetry such as mileage, battery health and driving behaviour in exchange for tokens. Developers and businesses, including insurers, fleet managers and EV services, pay to access that data with the driver’s permission. It shows a DePIN model where the hardware already exists and the network’s job is to give owners control over the data it produces.

peaq: infrastructure for the machine economy

peaq is a layer 1 blockchain built specifically for DePIN and robotics rather than a single network. It provides machine identities, payments and data verification that other DePINs build on, and says more than 60 DePINs across 22 industries use it, including the mobility and mapping networks NATIX and MapMetrics and the wireless network Roam. Its mainnet launched in November 2024, it signed a memorandum with Dubai’s virtual asset regulator in October 2025 to develop a “machine economy free zone,” and in 2026 it brought World ID verification to robots so machines can prove they are machines. Crypto Coin Show has covered peaq extensively, including its Purple Paper on robot money.

How to evaluate a DePIN

Device counts and coverage maps are the numbers DePIN projects promote most, and they are the least informative. A network can add hardware for as long as its token pays well. The questions that separate durable networks from subsidies are about demand and value capture.

• Revenue versus emissions. Compare what customers pay each month with the dollar value of tokens issued to operators. A network paying out $10 in rewards for every $1 of revenue is still buying its supply. The direction of that ratio over time matters more than its level.
• Who the customers are. Named enterprise customers with recurring contracts, such as AT&T for Helium or survey and agriculture firms for GEODNET, are worth more than activity generated by token holders using the network to earn rewards.
• How value reaches the token. Check whether revenue is used to burn or buy back the token, is paid to stakers, or stays with an operating company. If none of these apply, the token may not benefit from the business growing.
• Operator economics. Work out the hardware cost, running costs and payback period at current token prices, not at the prices in the marketing. Networks whose operators lose money will shrink when prices fall.
• Supply and unlocks. Many DePIN tokens have large allocations to investors and teams that unlock over several years. Scheduled unlocks add selling pressure regardless of how the network performs.
• Verification quality. Ask how the network detects fake devices and fake work. Spoofing dilutes honest operators and makes the service less reliable for customers.
• Centralization. Many DePINs depend on a single company for hardware, software updates, customer contracts and reward rules. That is often necessary early on, but it concentrates risk.

By these measures the sector has improved. Messari’s index of leading DePINs only includes projects with at least $500,000 of annual recurring revenue, and in 2025 it noted that leaders traded at 10 to 25 times revenue, a valuation framework that barely applied a few years earlier, when nearly all income came from emissions.

Regulation

DePIN has had a quieter path through US regulation than much of crypto, largely because operators earn tokens for providing a service rather than buying them as an investment. The SEC’s 2025 decision to drop its unregistered securities claims against Helium was the first signal. In September 2025 the SEC’s Division of Corporation Finance issued a no-action letter to DoubleZero, a network that pays operators for high-performance fibre links between validators, saying staff would not recommend enforcement over its token rewards. A second letter followed in November 2025 for Fuse, a Solana-based energy network. No-action letters cover only the specific facts described and do not bind courts, but they give other networks a template for structuring rewards. Hardware networks still face sector-specific rules, including radio spectrum licensing for wireless networks, privacy law for location, imagery and vehicle data, and energy market rules for grid devices.

Risks and open questions

The largest risk is the one built into the model: a network funded by its own token is exposed to that token’s price. Most DePIN tokens fell between 90 and 99 percent from their highs, and when that happens operators unplug and coverage shrinks just when the network needs to prove itself to customers. A second risk is concentration of demand. Helium’s revenue depends heavily on carrier offload deals, and compute networks depend on the AI spending cycle. A third is competition from incumbents, which can cut prices or build the same capability once a DePIN proves a market exists. Fourth, many tokens are thinly traded, so prices can move sharply on leverage and liquidations rather than on network news, as HNT showed in August 2026. Finally, sector-wide revenue remains small relative to valuations: roughly $72 million of on-chain revenue in 2025 against a sector valued in the billions, with a handful of networks earning most of it.

Why it matters

DePIN is one of the few areas of crypto where the product is something people outside crypto already pay for: mobile data, maps, GPS accuracy, GPU time, storage. The early networks proved that tokens can bootstrap physical supply at a scale and speed no startup could fund. The current generation is proving something harder, that the supply can be sold. Helium routing AT&T traffic, Bee Maps selling to robotaxi programs, GEODNET serving survey and agriculture customers and XYO positioning its chain as an audit trail for AI agents are all versions of the same transition, from paying people to build a network to having customers pay to use it. The networks that complete that transition will look less like crypto projects and more like infrastructure companies whose ownership happens to be distributed. The ones that do not will remain subsidies with a token attached.

Disclosure: XYO Network is a sponsor of Crypto Coin Show. Ashton Addison, founder of Crypto Coin Show, has been an investor in peaq since 2021. Neither had any input into this guide. This guide is for education only and is not investment advice.