DePIN

What the token is actually paying for

A DePIN pays people to buy and operate equipment the protocol never owns. Radios on rooftops, dashcams on windscreens, GPUs in spare rooms. The operator carries the purchase price, the electricity and the site. In return they receive tokens that did not exist before their hardware came online. What the protocol gets is a supply curve it never had to finance.

The 2022 essay that named the pattern put the arithmetic plainly: labour cost in the Helium network is $0 to the protocol, because the network is built by independent hosts rather than by crews on a payroll. By April 2022, thirty months after launch, Helium had passed 600,000 hotspots. The entire US telecommunications industry runs 417,000 cell towers.¹

Read that comparison carefully, because it says one thing and one thing only, which is that token incentives assemble physical supply faster than corporate capex does. A hotspot is not a cell tower and the two numbers do not measure the same service. Whether anyone wanted the supply is a separate question, and it is the question that decides the whole design.

The four pieces every DePIN economy has

A 2026 scoping review of DePIN tokenomics in Frontiers in Blockchain sorts working designs into four recurring primitives. A verification method that proves the off-chain work happened, such as Filecoin's proof of replication for storage or Helium's proof of coverage for wireless. A fiat-denominated pricing rail, usually prepaid non-transferable credits minted by burning the native token, so customers are quoted in dollars and not in a volatile asset. A burn and mint equilibrium that routes the burned value back into operator rewards. And staked collateral that can be slashed for poor performance or dishonest reporting.⁶

The same paper states the failure mode without decoration: high initial inflation required to bootstrap supply can depress long-term token value if it is not met with sufficient demand-linked value capture, and crypto volatility feeds straight through to provider return on investment.⁶ If your design is missing the pricing rail or the burn path, you have the inflation and none of the capture, and the review is describing you.

Helium, and what a nine year emission schedule looks like from inside

The first HNT was emitted on 29 July 2019 with no premine. Under HIP-20 the network runs a two year halving: 60,000,000 HNT targeted in year one and again in year two, 30,000,000 in years three and four, 15,000,000 in years five and six, 7,500,000 in years seven and eight starting 1 August 2025. The network moved from its own chain to Solana on 18 April 2023.²

Then demand arrived and the reward per unit of work fell anyway. HIP-149, an approved governance proposal dated 2 June 2026, records that network rewardable bytes grew roughly fourfold over the preceding year, from about 24,000 GB per day in June 2025 to about 97,000 GB per day in April 2026, while HNT issuance stayed fixed under the HIP-20 schedule and the HNT price more than halved.³

Work the arithmetic and you can see what the deployer felt. HIP-149 puts the Mobile data deployer baseline at about 13,870 HNT per day against about 91,000 GB per day of rewardable volume, which is 0.152 HNT per GB. At the roughly $0.27 HNT price the proposal cites, that is about four cents per GB, against the $0.50 per GB target rate HIP-53 had set. Four times more traffic, a fifth of the dollar reward per unit. Nothing in the emission schedule noticed that demand had grown, because emission schedules do not look at demand.

What Helium did about it, written down with an end date

HIP-149 does four things at once. It ties Mobile deployer earnings to the per-GB price the network charges payers, with a floor at half that price and a cap at three times it. It retires proof of coverage on both the Mobile and IoT networks, so deployers earn pro rata of rewardable bytes rather than for proving that coverage exists. It adds a seven seat advisory council with power to escalate to a curtailment vote. And it mints an operations and growth supplement of roughly 141,000,000 HNT over 36 months, about 77% of the roughly 182,500,000 HNT on chain, front loaded so about half arrives in the first twelve months. Effective maximum supply rises from roughly 206,000,000 to roughly 347,000,000.³

That is a large number and the proposal does not hide it. Both windows are hardcoded at deploy and self terminate, so no future vote is needed to end them.³ Our view is that this is the honest shape of the trade. A DePIN that needs more subsidy should say how much, for how long, and what stops it, in a document with a proposal number on it. The dangerous version is the one that keeps topping up through discretionary treasury grants nobody has to vote on.

Hivemapper and Render pay operators out of what customers burn

Hivemapper caps HONEY at 10 billion tokens and allocates 40% of that, 4 billion, to contributors, minted as global map progress is made. The first HONEY was minted on 1 November 2022. Separately from that pool, map consumers burn HONEY to buy data, and MIP-15, finalised in April 2024, reissues 25% of the burned amount to contributors as consumption rewards while permanently destroying the other 75%, capped at 500,000 HONEY per week.⁴ Two reward streams, one funded by a schedule and one funded by customers, kept visibly separate in the protocol's own accounting.

Render runs the same split more tightly. Jobs are priced in fiat, paid in RENDER, and the RENDER is burned when the work completes, with node operators rewarded from a declining emission schedule that the community sets by vote. RNP-006 allocated 9,126,804 RENDER for year one and RNP-018 allocated 5,905,580 RENDER for year two, a cut of roughly 35% between the two.⁵ The taper is a published number attached to a proposal ID, which is what makes it a commitment rather than an intention.

The ratio that separates a working network from a subsidised one

Take the tokens paid to operators over a period. Split them into the part funded by tokens customers destroyed and the part funded by scheduled issuance. That fraction, tracked over quarters, is the only honest measure of whether a DePIN is turning into infrastructure or is still buying supply on credit. Node count moves in the wrong direction under this test, because every new node dilutes the same reward pool.

Helium now encodes the ratio directly. The revenue-linked top up in HIP-149 is bounded by a seven epoch moving average of HNT destroyed on chain, so the demand-funded portion of the reward can never mint more HNT than the network has actually burned over the same window.³ The supplement in Decision 2 sits outside that bound, which is exactly why it needed its own vote, its own council and its own expiry.

The design consequence is unglamorous. Publish the split. If usage-funded rewards are 3% of the total, say 3%, and say what has to be true for it to reach 30%. Founders who cannot state that number in a board meeting have not been running a network, they have been running a distribution.

What to settle before the hardware ships

Five things, and the order is the point. What physical work you are paying for and how you verify it happened. How customers pay, in what unit, and whether that unit insulates them from token volatility. What burns, and where the burned value lands. What the emission schedule is and what triggers the taper, expressed in usage rather than in dates. And what an operator's payback period looks like at the reward rate the schedule produces two years from now, not at the launch rate.

Reverse that order and you get the version we see most often: hardware shipped, operators recruited, a reward pool sized to a growth target, and a demand-side pricing rail designed a year later once the token has already been diluted to build supply nobody is paying for. The token is infrastructure financing. The business is whether customers show up and pay.

One boundary. This page is design reference for founders and operators. It is not investment advice and nothing here is a recommendation to buy, sell or hold any token.