DePIN Token Economics

by Tom Trowbridge
Co-Founder & CEO of Fluence
15.01.25
DePIN Is Not a Meme!
Since the inception of the crypto ecosystem, crypto investing has been meme driven. In the absence of material revenue, investors have chased narratives to see which projects are ‘winning’. Investors look at sentiment, opinion leaders and ‘leading’ investors for signals as to what projects either will get traction or just to see what other people are likely to think.
To support these narratives, the crypto world has cycled through a number of metrics including transactions per second (TPS), Telegram channel size, Total Value Locked (TVL), GitHub pull requests, active developers, number of partnerships, number of transactions (real or not), and Twitter followers. And around and around we have gone with the top 20 largest crypto projects on Coin Markecap changing substantially every few years.
But this is about to change. DePINs (Decentralized Physical Infrastructure Networks) are on the cusp of generating substantial real revenue which, when combined with clear token economics, will change how we look at crypto. Meme coins will continue to flourish as well, but there will be a clear difference between DePIN and the rest.
DePIN Token economics
The DePIN sector has grown dramatically in the past twelve months with some research groups estimating over 1,000 projects. Across the DePIN ecosystem, over two million providers are powering various DePIN networks by providing Wi-Fi coverage, mapping roads, collecting car data, sharing weather data and providing storage and compute.
Given this early traction, we see DePIN as having the potential to bring crypto mainstream given the tangibility of so many of the DePIN projects and their clear value proposition to end customers. As the revenue and attention grows, investors will focus ever more on the token economics of the projects and of the sector overall which differs in important ways from other crypto sectors such as DeFI.
To prepare for this scrutiny, we wanted to lay out the main aspects of DePIN crypto economics to provide a framework for evaluating projects and as a guide for DePIN founders, investors and participants.
⚠️ Disclaimer: there is no standard format for describing token economics and projects use a variety of channels: blogs, threads, and white papers to explain their tokens in a range of detail. And as the space is constantly evolving, we have seen updates to a number of token models which adds additional complexity to authoring a comprehensive report. We view this paper as an introductory overview to the topic, and we welcome feedback, corrections and addition so that it can be as useful to the DePIN community and the broader crypto and investor communities as possible.
DePIN
We define DePIN as crowdsourced physical infrastructure, connected in a network, providing a useful service that is secured and incented by crypto economics. As with any new sector that is attracting interest and capital, we see a number of more traditional crypto projects attaching themselves to the DePIN name, but if we use this definition, the number of true DePIN projects is substantially smaller than the numbers mentioned above. But even this smaller list leaves us with dozens of high-quality projects that have the potential to disrupt massive sectors and be worth hundreds of billions and even trillions in the aggregate.
Token Models
We have seen several great research pieces on this growing space but none that are focused on token economics. The DePIN sector is new, and a number of projects are still updating and evolving their token models and most projects are yet to launch their economics.
This report serves to highlight the main components of DePIN token economics and to suggest the components that we think will drive value in the long term.
Thinking about long term value to the token is important because without a clear link between adoption and token economics, projects that gain traction run the risk of being eclipsed by later projects that can drive wealth, attention and ultimately higher adoption by well architected token economics.
💡To paraphrase the famous Bruins coach ‘Red’ Sanders, tokens economics aren’t everything, they are the only thing.
Many token models are new, and we see constant evolution of these models. As soon as this report is published, some aspects will likely be out of date, but we think it still relevant to note where we are now with an eye on the likely trajectory of these models.
Despite the evolution we are seeing, we think it is important for projects to launch with as mature a token model as possible because changing economics after launching is time consuming at a minimum and can be challenging depending on the scale of the changes. While it might sound prudent to develop models over time, the challenge with making substantive changes post launch is that constituencies form quickly around token models. These constituencies invest capital based on expectations, and if they are disadvantaged by any change, no matter how beneficial to the network, they can be a powerful, motivated force against change.
Token Economics
Token economies have two primary parts: the token issuance and the token use but within those two broad categories are a number of subcategories. On the issuance side, DePIN projects incorporate the token for rewards and on the use side, for payment, trust, governance and revenue. No project token incorporates all of these uses as DePIN networks have a variety of different requirements depending on the providers, users and security.
In designing token models, founders are usually aware of the risk of a low token price, but the risk of a high token price must also be considered. A low token price can reduce interest in the network, impairing the incentive for providers to offer the service, rendering the network less valuable to customers and catalyzing a negative spiral to irrelevance.
A high token price, however, can attract transient providers only interested in a quick buck, increase the incentive for fraud, and reduce interest in doing the actual work of serving real customers for revenue. Without real customer traction, the token price will eventually fall, causing the tourist or fraudulent network providers to vanish, risking the network’s viability.
We have seen several projects including Hivemapper and Xnet update their token economics to address the behaviors posed by a high token price, and the Filecoin ecosystem has suffered both from a high price which attracted providers only interested in FIL rewards and also suffered as a low token price caused providers to shut down as they were unable to pay their fiat costs.
Token as Payment Currency
The most traditional use of a token in a crypto network for payments on that network. The requirement to use the token gives the token utility and, as payments on the network grow, demand for the token should also grow.
The challenges however are that requiring the token to be used can impair adoption as most businesses, especially Web2 businesses, are not set up to buy and use crypto payments and, even more importantly, can not forecast costs in a volatile currency.
A solution to the first problem is to deploy a fiat front end where customers can pay for services using credit cards and wires, and a provider handles the tokens in the back end. This solution sounds obvious, but setting it up in a way compliant with existing financial regulations takes effort.
We also are skeptical that requiring use of one's native token for the network drives long term token value. Whether with a fiat front end or not, customers and providers can purchase and sell tokens pretty much right upon use, driving token volume, but not materially impacting the scarcity or demand for the token. With regard to the volatility of paying in a native token, networks can address this by pricing in fiat but requiring the token be used based on the current exchange rate.
We think that successful networks will reduce barriers to adoption and that pricing in USD with fiat front ends will become the default solution. Networks should be ambivalent about tokens or currencies used to facilitate adoption. Given the importance of a fiat front end which will accelerate the turnover in the token, it isn’t clear to us that requiring the token to be used in transactions will be worth the complexity.
We have seen projects move to this model with IO.net accepting fiat and Helium, DIMO and Geodent all selling some form of data credits which are priced in fiat with the token used only on the backend. Render, founded in 2016, facilitates the use of fiat via the purchase of Render Credits (USD value but in native token) which are used to use the network. Arweave, however, prices storage in USD, but customers pay in AR or in a number of other tokens.
Projects that require their native token to be used for payments include the compute networks of Akash, Aethir and the storage leader Filecoin.
Token as Reward
One of the great benefits of crypto platforms is the ability to bootstrap an ecosystem via token rewards. Platforms employing this token model pay rewards to providers contributing to the network. These payments are in return for providing a service such as storage or providing a weather station, or a wireless hotspot. These payments have typically been offered as a predetermined number of tokens issued daily or weekly that are divided among the providers and are subject to some decline schedule over time. The tokens are of course subject to significant market volatility but the rewards, even if low in value, can be compelling if the network scales and the tokens paid for these rewards appreciate significantly.
Why Early Rewards Really Matter?
The potential to earn early rewards that can be appreciated over time is one of the most compelling aspects of DePIN and helps generate a passionate and aligned community. The ability to reward early providers in a potentially appreciating asset is unique to crypto and one of its most democratizing features. Even as many projects become more institutionalized over time, the early providers remain beneficiaries of this growth and evolution thanks to their early rewards.
The rewards vary by project but usually are based on minimum payment to anyone providing services to the network with some offering higher payments to those who are providing the most useful service or in the most useful location. Most reward schedules decline over time like DIMO which declines 15% annually for 40 years with others like Glow that mint a fixed number of tokens every month in perpetuity. Filecoin, arguably the first DePIN project, incorporates a dual release schedule where 16.5% of supply releases to miners over 30 years subject to a six year halving but 38.5% releases to miners only upon the achieving of ever more aggressive storage capacity peaking at 1,000x the current cloud storage capacity in 20 years. IO.net rewards all GPUs on their network from the rewards pool which is subject to a 20 life with a 12% annual decline. Render rewards are subject to a 25 year decline.
Helium rewards decline on a two-year halving schedule and the rewards are paid to hotspots based on a number of criteria including uptime, contiguousness with other hot spots and location desirability. Nodle pays a base reward to all users and then additional rewards based on the activity of each device. The Hivemapper network pays rewards to drivers based on their activity and location and also pays providers who help train their AI models.
Geodnet rewards half every year on June 30th and rewards are two to four times higher in areas that lack coverage and are a priority.
Market by Market Rewards
The video mapping network Natix pays token rewards only to participants in markets that are active and Natix increases the token reward pool as the user base grows and in line with revenue, reducing the impact of additional users diluting the reward pool. Given that mapping is largely a local market, a market must have a critical mass of participants to generate sufficient data to be relevant to customers, and rewards are not paid in markets that have not yet reached that critical mass. Delaying rewards until that scale has been reached reduces inflation until, in theory, a market can generate paying customers. Natix also only rewards only the top 60% of participants in each market with tokens. This monthly activity competition between users helps ensure rewards are paid only to the above average participants.
Orienting rewards to participants helps drive participation which is a benefit, but it comes at the expense of rewarding tokenholders overall. The risk with a model that rewards participants more than token holders is that a token buyer will rationally prefer to hold the tokens of a project where the rewards flow to token holders, not to the relevant service participants. When you reward mainly the participants you end up with a kind of co-op which can provide an effective service but will likely not attract the same level of capital interest as a more general token model. And this lack of capital attraction puts much more pressure on the revenue of the network to be sufficient to reward all participants adequately. There is a spectrum between the two models, and no project can be successful without rewarding participants, but when alignment isn’t as clear with tokenholders, we would expect to see valuations impacted.
USD Denominated Rewards
Fluence rewards every CPU on the network with a monthly payment, but unlike most other networks, the payment is calibrated to track $10 per core to reduce the impact of token volatility on rewards which helps data centers have more certainty with regard to their revenue. The rewards also vest over 6 months, requiring the provider to continue providing capacity for the six month period to receive the full reward.
Arbitrary Reward Pool
Rather than set a reward schedule in advance as is market standard, or price in USD to protect from token volatility like Fluence, several projects have decided to determine the reward pool annually. Aethir offers one reward pool for gaming and one for AI, and the foundation determines the size of the pool annually as does Silencio. This architecture makes sense for consumer focused projects, but it seems less suited to institutional infrastructure providers who are more likely to value certainty regarding future reward pools.
Lottery Rewards
One issue that a number of consumer based DePIN projects face is the low dollar value of rewards. Sensor based networks that use smartphones like Silencio, Natix and Nodle require hundreds of thousands of active users. At that scale, each individual user's value to the network is very small but the network must find a way to keep users engaged despite not being able to sustainably offer a high reward.
Silencio solves this problem by offering a base reward to all users, a premium reward in high priority markets and a monthly lottery that gives all users, weighted by activity, the chance to earn any of a number rewards. This lottery approach turns participation in the DePIN network into a monthly lottery ticket which has the potential to increase user engagement far more than if the value of the winning was equally distributed among all users.
Token for Stake and Trust
DePIN networks that provide digital services like storage and compute require a trust model. Some like Akash serve as marketplaces where customers can evaluate the provider who might even have ratings. Other networks like Fluence, Filecoin and Livepeer require tokens to be staked to prove trust and also demand for the token. This stake is at risk, and should the provider behave badly the stake can be slashed.
This stake provides alignment and helps assure customers that the provider can be trusted. The projects take different approaches in determining the stake required, however, and that choice has important implications for the token economics. Filecoin for example, requires 30% of circulating supply to be staked to storage providers. If the providers go off line, this stake is slashed and returned to the network as revenue.
The challenge of a model that requires only a set percentage of supply to be staked is that as supply increases, only 30% must be staked which means the circulating supply of tokens is increasing irrespective of network growth. The other challenge related to a fixed amount of tokens required for stake is that if the token price is low, the trust model is jeopardized as the value of the stake could be below the value of the data stored, while if the price is very high, the cost to acquire and stake the tokens could be prohibitive.
The GPU marketplace IO.net requires providers to stake 200 IO tokens per card and each GPU is subject to a number of multipliers based on its specifications. Stake enables the GPU to join the network and earn rewards and is subject to slashing for malicious behavior or inadequate servicer.
Like other digital or cloud DePINs, Fluence has a stake model to provide trust, but like its reward model, the stake is priced in fiat but deployed in FLT. Stake is $12,000 per CPU for the duration of the stake commitment which varies from one to 12 months. If FLT is $0.25, then 48,000 FLT are required and if FLT is $1, 12,000 FLT is required. This model has the benefit of providing a clear security model to the network while also equipping the community to forecast the USD demand for FLT to stake based on the number of CPUs forecast to join the network. If we expect 10,000 CPUs to join the network, we can know that $120 million of FLT is required to activate those CPUs.
Fluence thinks this model is more secure than models that require an arbitrary number of tokens to be staked which is problematic both if the tokens have a very low value, reducing security or have a very high value, increasing costs for providers.
Livepeer requires its video rendering providers to provide stake and the revenue each provider receives is linked to the amount of stake. If a provider has 10% of the network staked, they can earn 10% of the network’s ETH revenue, assuming they perform at least 10% of the work. Livepeer also pays rewards but only to staked tokens and, if the overall network stake is less than 50%, the reward increases, if it is more than 50% the reward decreases.
Token Linked to Revenue – Buy and Burn
Buy and burn is the most typical and direct mechanism to tie network scale with token demand. Buy and burn is when the revenue generated by the network is used to purchase and burn the network token. If the amount burned every year is in excess of the tokens generated to pay rewards, the network is deflationary which should be positive for the token price over time. Buy and burn is a relatively simple concept, but it is an untested aspect of the token models because DePIN networks have yet to generate material revenue. As revenue scales, however, the buy burn dynamics have the potential to be very powerful.
The Role of Network Revenue
The obvious precondition to using this model is that it is the network generating revenue, not the individual participants. In the cloud related DePIN networks like Filecoin, Akash, IO.net and Fluence, the providers receive the revenue, with little to none going to the network, rendering the buy and burn model largely irrelevant. For non-cloud or physical service projects, however, like Helium, Hivemapper, Geodnet, Render, DIMO, Spexi, Xnet, Glow and many others, it is the network which generates revenue and so the network can buy and burn the token.
Cloud-related DePIN networks (e.g., Filecoin, Akash, IO.net, Fluence): Revenue flows to individual providers, making buy and burn impractical.
Physical service projects (e.g., Helium, Hivemapper, Geodnet, Render, DIMO): The network generates revenue, enabling buy and burn to function effectively.
For these projects, the value to customers usually comes from the aggregation of network participants which makes any individual harder to reward. There are exceptions, however and Noodel, for example, sends most of its revenue to the users whose devices were used to serve a particular customer. Even with Nodle, however, which tracks locations, there could be hundreds of devices serving a customer so the reward is divided considerably.
Buy Burn Range
The rates at which projects use revenue to burn tokens varies significantly. Glow uses 100% of the revenue generated from the sale of carbon credits to burn its tokens, Render uses 95%, Geodenet and Xnet each use 80%, Hivemapper 50%, and Nodle uses just 5% with the rest returned to the relevant devices as mentioned above.
The buy and burn model is critical to linking DePIN project traction and revenue with token demand, and it is the most widely adopted mechanism to accomplish this link. Some projects that didn’t originally include this feature like DIMO have since adopted it and others are in the process of doing so. Eventually we see buy and burn being a core feature of DePIN token models with the burn amount standardizing in the 80% range. As revenue traction increases, the buy/burn demand should ramp up considerably. Some projects have worried about ‘too many tokens’ being purchased, forgetting that as tokens are purchased, the scarcity should drive up the token price, resulting in the need for ever more revenue to purchase the same number of tokens. We expect token markets to adjust to the demand driven by this revenue and reset the token price to a level that discounts future purchases from revenue. This increase will result in a higher token price, and fewer tokens being purchased but will create a higher equilibrium which should settle at a price that allows the purchases to offset issuance or be projected to do so in the foreseeable future.
Fiat Rewards
Projects should also note that if rewards are calibrated in fiat as opposed to an arbitrary issuance amount, it is far easier for the network to achieve deflation. With buy-burn driving demand and a higher token price, in this model, rewards require fewer and fewer tokens as the price increases, leading to less issuance and then a higher price which fuels a virtuous cycle of appreciation and reduced issuance. This is of course predicated on traction and revenue, because without traction, a falling token price will result in additional token issuance driving dilution and vicious cycle down.
Token as Governance
Decentralized projects use tokens for governance and the DePIN sector is no different. Governance deserves a paper on its own, but it is worth highlighting a few aspects here.
Filcoin has had the longest standing token governance model dating from FIP-0001 on April 2, 2019 with over 70 passed proposals.
Akash has had a very active on chain governance community with hundreds of proposals since launch of Mainnet 2 on March 8th, 2021
For projects planning on token model evolution, one cautionary example is FIL-0056, the Filecoin proposal supported by management to increase the amount required to stake from 30% to 50% which was voted down by the miner-heavy community in March 2023.
Helium has a very active on-chain governance community that has passed over 130 HIP including the significant move to the Solan network and most recently HIP138 which consolidated the number of network tokens from three back to just original core token HNT.
Token voting is important to achieve decentralized governance but it is not without risk. Decentralized systems are harder to govern and decisions can be driven by any number of small constituencies including vocal members who have small economic interests or, conversely, large token holders interested in short term token performance. These issues are not unique to DePIN, but as the revenue creates ever more valuable treasuries, we can expect scrutiny and potentially even governance attacks. When architecting the DAO voting process, projects should consider how to address a rogue vote where low participation allows a large holder to drive through a self-serving proposal. To protect against this potential, the Fluence governing council has a week to veto any DAO vote, but the council is elected by the DAO every two years to ensure that the council can not diverge from the interests of the community.
Natix and Aethir have implemented a staking mechanism for voting where only staked tokens can vote, and Natix gives higher voting weights to tokens that are locked for longer. Besides aligning voting with longer term holders, Natix also pays 40% of revenue to locked tokens, rewarding those holders and providing a mechanism for holders to benefit from network revenue without operating phone hardware.
What Are Medallions?
A recent innovation in token models is the medallion concept that essentially securitizes an aspect of a network.
Daylight is contemplating auctioning geographic specific medallions that permit the holder to purchase the energy from the geography.
Dawn will also sell geographic medallions that enable the holder to share in the revenue from that area.
TheDawn mechanism is especially interesting because it crowdsources the assessment of where the network is most likely to generate revenue because the higher the perceived revenue potential of a particular area, the more expensive the medallion. We expect to see medallions implemented by other projects in similar ways, but are wary of overly complex medallion models.
The best token economics are the simplest ones, and while medallions can generate valuable signals, some projects run the risk of getting caught up in the design and overcomplicating the concept. We think medallions will play a role, but only where really necessary and where they serve as an economic signal as they do for Dawn, and not just a value creation mechanism.
Token Model Evolution
While there is no one token model that works for all projects, we see most DePIN projects evolving to accept fiat, denominate rewards in fiat and link revenue to the token with a buy and burn mechanism. For cloud services DePIN projects, stake denominated in fiat seems the simplest way to ensure network security, and allowing anyone to stake, not just hardware providers, should both ensure the widest community participation and the lowest provider cost.
The coming year will bring the launch of dozens of new projects with a number of new models, and we look forward to seeing the space continue to evolve. A final thought is that token models should be easy to understand, and if they can’t be explained simply, we should understand why. Is the complexity actually adding value or, more likely, is the complexity obscuring the real value drivers or lack thereof.
We ask founders to make the token economics easy to find and easy to understand - put them on one page! And for the community and for investors, make it clear to founders what you would like to see both in terms of token economics transparency and also with regard to the actual mechanics.
Feedback works! And open, clear and simple is the way forward!
Two Paths for DePIN: Trust vs. Simplicity
And even within the small DePIN sector that have trading tokens, the token models fall into two broad categories based on the need or lack thereof to provide trust in the underlying infrastructure.
Digital Service DePIN Projects like cloud replacement or ‘digital service’ DePIN projects that focus on storage and compute, proving trust is critical and token economies must provide a mechanism to generate that trust.
Physical Service DePIN Projects that use hardware to provide more tangible service like video, Wi-Fi or location, trust is less important given the lower likelihood of fraud.
The DePIN sector highlights an essential truth: one size does not fit all. The success of token models lies in their ability to adapt to the specific needs of the network, balancing trust, utility, and simplicity.