Infrastructure networks are nothing new to the world, with behemoths like T-Mobile, Amazon S3/AWS, Google Drive, and even Uber all providing valuable products and services. Traditionally, these infrastructure networks have been built by businesses through decades and billions of dollars of investment, with ownership models that are highly centralized. Despite the apparent benefits such networks provide in the everyday consumer’s daily life or for corporate applications, society at large has historically never been able to invest or even take part in building out said networks. Decentralized Physical Infrastructure Networks change that.

What is DePIN?

Decentralized Physical Infrastructure Networks (DePINs) lie at the critical intersection of blockchain technology and real-world utility. Aside from financial applications, DePINs may perhaps be one of crypto’s most significant markets in the end state because each market that a DePIN subsector is trying to replace is individually worth tens or even hundreds of billions.

Fundamentally, these infrastructure networks utilize blockchains to coordinate a wide range of resources through economic incentives, whether that be the deployment/supplying of physical hardware, virtual infrastructure, or other goods and services.

These infrastructure networks are decentralized marketplaces, where on one side, you have suppliers supplying a resource, and on the other, you have buyers consuming a resource. The differences to the traditional marketplaces are the lack of an overly rent-seeking centralized intermediary acting as the middleman/facilitator, and the innovation around investment/resource pooling by the supply side. Without the centralized rent-seeking intermediary and the reduced operational overhead by virtue of relying on a protocol instead of legal contracts and KYC to onboard suppliers, suppliers receive more profit assuming the same amount of demand, and buyers benefit from lower prices assuming the same amount of supply. Centralized rent-seeking intermediaries are transformed into value-distributing protocols, owned and operated in a distributed/decentralized fashion, in which network participants are also stakeholders instead of simply being suppliers/buyers on both sides of a marketplace. But it is that difference between a decentralized network and a centralized entity that makes DePINs so different from existing resource networks, whether that be resource deployment/acquisition, operations, GTM, and more.

There are two large categories within DePINs: Commodity Hardware DePINs and Custom Hardware DePINs.

Commodity Hardware DePINs include resources such as storage, bandwidth, and compute. Notable protocols within this category include Filecoin, Render, io.net, Akash, Livepeer, POKT Network, and AIOZ Network. The key differentiators for this category are that there is often latent, idle supply available from existing commodity hardware owned by suppliers (idle off-the-shelf GPUs sitting around), and there is much lower geographic sensitivity (in general, it does not matter where the supply is currently located, barring some additional latency if the resource being supplied is geographically far). However, it is likely that over time, the line between these two slightly distinct categories will continue to blur as more hardware becomes Commodity.

Within this category, there are some DePIN projects that are best known for providing solutions for crypto-native use cases. One example is POKT Network, a decentralized RPC protocol that provides the coordination and incentive layer for RPC requests (the vital communication protocol that enables end user applications to read/write to open data sources, like blockchains).  To date, POKT Network has focused on blockchain RPC, but is currently expanding to incentivized RPC for indexers and oracles, as well as non-crypto use cases like AI inference. As the number of blockchains and rollups continue to grow, and their collective needs get even more demanding, with higher throughput and shorter block times, the need for performant RPCs for every blockchain has become very clear.

This is especially evident during periods of high load, such as NFT mints or large amounts of memecoin trading activity, where most RPCs cannot handle the load of users interacting with the blockchain in the form of reads and writes. As we move deeper into a digital machine-2-machine (m2m) world of onchain agents - who need to communicate, interact, be paid, and make payments - POKT Network provides a Universal RPC Base Layer to connect it all. With a network of over 15K nodes in 22 countries supporting over 60 data sources, the protocol currently facilitates ~500M requests per day and, at its peak, facilitated up to ~2B daily requests and maintains 100% uptime.

POKT Network has pioneered demand-side decentralization via their ‘Gateway’ ecosystem, which can be seen as DePIN’s equivalent to L2s. Gateways are businesses that build on top of POKT Network and resell RPC as part of their offering. Building directly on the protocol removes the need for a complex backend to serve RPC requests, allowing gateways to focus entirely on the user experience by providing customer support and value add features and focusing on generating demand. Gateways borrowing from the robust uptime and scalability of the protocol while optimizing performance and UX provide end users with the best of both worlds, decentralized infrastructure underneath with centralized UX and support. This started with Grove, the original Gateway to POKT Network whose own client base for POKT Network’s underlying service now includes large RPC providers like Infura. Since October 2023, POKT Network has now added two more sovereign Gateways (Nodies Infra and Liquify) with further Gateway announcements expected imminently, each bringing their respective brands, distribution, and network to the ecosystem. In a similar fashion to how one DeFi protocol might want multiple frontends to facilitate interactions with end users, Gateways provide end users with more options, but can also tailor their product towards particular regions and markets, and provide additional customer support and value add features. As POKT Network proves the benefits of this approach we expect to see other DePINs adopting a similar playbook to rapidly accelerate growth and innovation.

The second category is termed Custom Hardware DePINs, which include resources such as mobile data, mapping data, and imaging. Notable protocols within this category include Hivemapper, Helium, and Dimo. The key differentiators for this category are that it often requires specialized custom hardware that is an upfront capital investment by the supplier and that there is high geographic sensitivity (e.g. you want imaging of specific streets in specific cities or you want mobile coverage in a specific area)

There is one last category that does not necessarily fall under DePIN, but which we are equally excited about. We call it “Decentralized Human Capital Networks”, where the primary resource supplied is human resources. The thesis here is threefold. I believe that play-to-earn games were just the beginning of a fundamental reorganization and gamification of the labor force and that blockchain technology enables:

1. Earning power: Workers can seamlessly earn, claim, and stake monetizable rewards received while working from anywhere in the world.
2. Autonomous organizations: Decentralized support infrastructure that turns the current workforce into a permissionless, worker-driven and owned organization.
3. Work tokenization: Turning a piece of work into an easily transferable, potentially appreciating asset, which unlocks the worker’s economic upward mobility. 

The most straightforward representation of this would be onchain human capital marketplaces, of which the leader is Braintrust. Fundamentally, blockchains are the best infrastructure for bounties and permissionless incentive systems to get humans to do actual work, and I firmly believe that we are only beginning to see the possibilities. Another interesting example of this would be Bittensor, which attracts human capital to build subnets through its tokenomics, as the protocol relies on well-fine-tuned models, which require AI professionals. The most left-curve view of this thesis would be gamified global permissionless labor arbitrage. For example, one could easily envision mystery shopping companies as a decentralized human capital network built on blockchain rails, accelerated by crypto-economic incentives. However, this only applies to job functions where humans are irreplaceable, e.g., a robot can’t give you a review of whether a museum was a good experience or not (or at least not yet). In the future, many job functions will likely be transformed into AI agents, and these AI agents will be facilitated through a decentralized marketplace.

Inherent to DePINs is the DePIN flywheel, which broadly describes a positive reflexive cycle that allows these networks to be bootstrapped and scale. The flywheel begins with token incentives rewarding the supply side of the deployable resource, resulting in latent supply on the network, available for consumption. After that, as the network's supply side continues to grow, the service offering/quality continues to improve until a certain threshold is reached where there is now a unique selling proposition to the demand side. This leads to increased structural demand, which results in increased earnings to the supply side. This then results in a higher token price, as often there is some positive relationship between token prices and network utilization/resource demand through a value accrual mechanism. Finally, the higher earnings and token prices result in further supply joining the network, with the marginal unit of supply improving the network, and the flywheel continues. However, after the protocol reaches a certain level of supply, more supply is only helpful when utilization is already close to full capacity. It has also been historically hard for protocols to bootstrap the demand side, which is critical to the flywheel continuing. It is prudent to note that this flywheel is fairly reflexive both ways: when network utilization and demand increase, but also when network utilization and demand decrease. This could result in a rapid spiral in demand and supply through a “death loop” as a lack of demand results in a lack of earnings for suppliers and lower token prices, and thus a decrease in supply, and thus a worsening product offering, and so forth.
 

DePINs don’t always need to strive for the same competitive advantages as existing non-crypto counterparts. In fact, it’s often very hard to compete on the same value proposition for reasons stated later in this report, and often, buyers are willing to pay a higher price for better products or service guarantees. Therefore, DePINs don’t need to strictly compete on cost, DePINs can compete on other value propositions. For example, Hivemapper doesn’t need to strictly compete on cost for mapping imagery, it could instead compete on freshness by providing week-old mapping imagery data in comparison to Google Maps’ year-old mapping imagery data. This net new value is something that traditional counterparts cannot offer and should often be a DePIN’s unique selling point rather than simply cost. However, DePINs will still need to be competitive on cost and reliability. Traditional counterparts are often highly reliable, so DePINs first must compete on reliability, at least closely matching the same level of reliability, before considering competing on cost. 

What are the critical implications and opportunities for DePINs?

One often asks why DePIN is needed, which is a valid question, considering that centralized companies provide most of these resources at a fairly reasonable price and sufficient product quality. The primary value of DePINs is incentive alignment, allowing anyone to contribute permissionlessly to a set of shared objectives; in the case of DePINs, building out a resource network that acts as a decentralized and permissionless alternative to existing solutions. Most DePINs can be built in a centralized manner, but such companies are likely to be rent extractive even though they provide a valuable service to society. Creating a marketplace where prices are determined by market dynamics, and the market is open to all participants, allows the unit costs for the digital product/service to settle into its fair equilibrium. This sidesteps the usual moats that large providers enjoy, significantly drives down prices, and facilitates truly global competition at the resource level. The remaining margin, decreased by fair competition, is proportionally captured by every network participant.

One could argue that the world would be a much better place if all infrastructure networks were decentralized and operated such that the general society would benefit from the generated revenue. However, infrastructure networks require a large amount of initial capital, and whoever builds them out often takes on a large amount of risk, so one could argue that the rent-seeking is justified. DePINs can be valuable because individuals both get the service and can also benefit financially. However, one could also argue that in their current state, you wouldn’t want to rely on a DePIN protocol for your internet/GPS/data connection/etc., given the lack of coverage and possibly lack of solid performance/uptime guarantees.

DePINs allow individuals to gain equity-like ownership in a collective manner by building out an economically productive infrastructure network, something that requires the correct coordination and incentive mechanisms. DePINs can exist at a very limited scale at an individual level, but connecting these small individual networks together into more extensive and economically productive networks is where the real value lies. For example, there are multiple examples here and here of individuals starting up their own hyper-localized ISPs. But to expand those networks to have larger geographic coverage is a futile effort without using a blockchain to act as a single source of truth to reduce coordination costs at scale and to leverage token incentives to bootstrap the supply side.

Following the token incentives and equity-like ownership conversation and using a Web2 comparison, Uber drivers may have been much better off today had they been offered UBER shares for being one of the early Uber drivers instead of receiving sign-up dollar bonuses. Dollar bonuses are one-off and do not give upside to the overall network growth or cash flow generated of a $150B company today (with the caveat that UBER shares would not have been sellable until the IPO in 2019, which is a big bonus of liquid crypto tokens in incentive systems). In addition, had drivers been given UBER shares, they would likely have had much more say in the operational decisions and direction of the ride-hailing network. However, this is likely a moot point given that most DAOs are incredibly ineffective today and existing DAOs have shown us that centralization to leaders who know what they are doing can often be a positive (but having a say and being able to express a view through a vote is still powerful). Token incentives may have also meant that Uber, the company and team, would not have had to raise a total of $13.2B in funding to bootstrap the network, which likely significantly diluted the founder’s ownership in doing so (with the caveat that ownership would have been diluted too by reserving portions of the token supply for airdrops, community, treasuries, etc.). Instead, token incentives could have been used to achieve the same scale, perhaps even faster in a less dilutive manner.

Furthermore, in traditional infrastructure networks, there is a large amount of waste and inefficiencies generated due to having multiple competing businesses either compete for the same limited supply, or build out duplicates of the supply, e.g. Uber and Lyft competing for the same drivers, or T-Mobile and AT&T building cell towers next to each other. DePINs allow competing businesses to tap into the same underlying infrastructure network, and this coopetition dynamic is only possible by having a decentralized protocol that anyone can permissionlessly leverage. These end-user facing businesses then compete on brand, customer support, value-added features, etc, creating massive consumer and potentially producer surplus as they will have lower startup costs and can cater to more targeted markets instead of having one company trying to serve every global market.

Another aspect of DePIN where crypto is a strong value add is capital formation and ability to scale. Relative to traditional forms of capital formation, e.g. venture funding through selling equity, DePIN protocols allow networks to be built faster, in a more cost-effective manner, and enable the network to more granularly react to hyper-local market needs. The ability to rapidly scale across geographies, jurisdictions, and different markets in parallel in a permissionless manner gives DePIN a leg up compared to traditional infrastructure networks, as long as the deployment of resources does not require an overwhelming amount of technical expertise. Building a marketplace as a Web3 protocol removes the centralized overheads on scaling and reduces the barriers to entry for new supply participants, allowing the network to potentially encompass every available hardware device in the world. In addition, connecting infrastructure through a single decentralized network provides a level of scalability that is currently impossible to achieve through any existing provider, giving unprecedented on-demand access to the entirety of the world’s supply of hardware for the particular purpose that a DePIN cares about.

Another crucial component of crypto in DePINs is credible neutrality. Traditional infrastructure networks are burdened by regulation, albeit sometimes for the better. For example, you wouldn’t want random individuals hogging up radio channels that are used for air traffic control. In addition, many of the rails that traditional infrastructure networks are built on, or the companies that own the infrastructure networks themselves, are mostly extremely centralized, with significant implications for social welfare. The credible neutrality applies to the underlying blockchain that the DePIN lives on and, in some cases, applies to the DePIN itself. Credible neutrality is defined here as a mechanism that does not discriminate for or against any specific people. For example, you would not want a DePIN to be built on traditional financial rails, where the cost of processing payments could increase by 10x overnight, or payments were censored, and you would have no recourse.

Imagine if the government did not like a certain DePIN, and ordered Visa to block transactions that helped to facilitate payments for this DePIN. This represents a significant censorship risk for a DePIN, and thus, it requires permissionless financial rails. A DePIN protocol could require users to be KYC’d/KYB’d before they are paid, but at its core, the underlying blockchain has to be permissionless and censorship-resistant.  Such guarantees are essential to DePINs given the inherent financial nature of these networks, and one would not want the underlying blockchain to ever favor one DePIN over the other. However, most DePINs themselves are not credibly neutral as evidently, most DePINs have incentives or mechanisms that optimize for specific outcomes, such as reaching a certain threshold on the supply side or heavily incentivizing the demand side, and such rules are often subject to change through DAO governance.

Lastly, the low-cost financial rails of blockchains paired with the near-instant settlement make blockchains well-suited for peer-to-peer micropayments of both fiat equivalents and tokens that traditional payment systems cannot support. DePINs require frequent payments to guarantee responsiveness to incentives, and also changes in incentive distribution. One may not be willing to take the risk of deploying a resource if they were only receiving payments every two weeks.

Supply Side

Suppliers are individuals or stakeholders who are providing a resource to the network. Often, a DePIN requires a certain scale on the supply side in order to reach a minimum viable product quality. For example, a telecom network isn’t that useful without a strong signal strength and wide geographic coverage. A taxi network isn’t helpful if you wait 30 minutes for a ride. Reaching this threshold to kickstart the DePIN flywheel is often a make-or-break requirement for a successful DePIN. There will likely be a large set of customers willing to pay for a certain service or product, but only provided that it meets a specific quality threshold. Without attaining said threshold, it is unlikely for the network to have any sustainable demand, as there is zero willingness to pay below that quality threshold. If a DePIN fails to reach said critical threshold while having no more token incentives left, then its future is relatively bleak, barring any new external funding or a tokenomics revamp (e.g., MATIC to POL). As such, for a large number of DePINs, especially DePINs that require some custom hardware in order to supply a resource, it is natural that these start as hyperlocalized networks, as a strong geographically concentrated network is required to generate the network effects and attain the minimum viable product quality stage.

There are two methods of supplying resources to an infrastructure network. The first is latent commodity hardware, where a user simply plugs in some existing piece of hardware they have into an infrastructure network, usually with some configuration required. Examples include computing with GPUs and storage with HDDs/SSDs. The other method requires users to buy custom hardware built explicitly for the DePIN protocol and then plug it into the network. Examples include mobile, where suppliers need to buy Helium miners, and Hivemapper, where suppliers have to buy custom dashcams. DePINs that require custom hardware instead of latent hardware are generally much harder to bootstrap, given that suppliers must pay a significant amount of upfront capital before they are eligible to receive rewards/token incentives.

One could argue that a single manufacturer producing the custom hardware is a centralizing force and a significant point of censorship for the network. However, one has to be cognizant of the stage at which Custom Hardware DePIN protocols are today and that decentralization, like most things in crypto, is a gradual process. One example of progress here is Helium with HIP/0019, which enables an approval process for third-party manufacturers and thus creates fail-safes against any censorship vectors or limits to growth on the supply side. 

One differentiating factor between existing infrastructure networks and DePINs is how a resource is supplied. Resources are often supplied at a more “micro” level in DePINs than in Web2 infrastructure networks. For example, with Helium, users supply mobile data through micro hotspots, which are less performant than 5G cell towers deployed by centralized telecom companies but can be deployed at a much more rapid pace and in much higher quantities with a much smaller footprint. Similarly, a Hivemapper dashcam may not be as performant as a Google Street View car given that they host multiple high-definition cameras with a 360 view, but yet Hivemapper dashcams can be deployed in much higher quantities given the lower barriers to entry and road map imagery at a much higher frequency.

The crux of growing the supply side is token incentives. In most cases, users have no organic incentives to supply a resource to a network as there is no demand from day one, meaning they would receive zero revenue/income. Thus, the only way to get supply to join the network is through ongoing token incentives to compensate for that lack of revenue and to ensure that the network has a minimum amount of latent supply. As mentioned earlier, this minimum amount of latent supply is critical to providing a minimum viable service/product quality and can also help suppliers cover their operating expenses. Furthermore, the same suppliers gain equity-like ownership in the network, which should incentivize them to help grow the network, spread the word, and increase demand, as it should theoretically result in a higher token price.

Such token incentives are often much more important for DePIN protocols with custom physical hardware, such as Helium miners or Hivemapper dashcams, as suppliers would not be operating said hardware without the explicit goal of supplying it to the network. Conversely, for resources that require hardware that most people own already, such as if I have a gaming PC that already has a GPU, the opportunity cost of supplying that GPU to a compute DePIN is relatively low, and there is no upfront fixed cost.

Supplying the first unit of resource to an infrastructure network should yield the supplier significantly higher ownership of the network (token emissions) than the 100,000th unit of resource, as the first supplier took an outsized amount of risk, perhaps when the token price was much lower (or the token did not even exist yet), and it was unclear if there would ever be sufficient demand for the infrastructure network for it to be self-sustaining (in terms of the network and suppliers being profitable with zero token incentives). This is not that different from VCs requiring less and less ownership of a startup in later investment rounds as various components are derisked over time.

However, there are still structural barriers to entry for most DePINs on the supply side. For example, we are not sufficiently far along on the adoption curve of crypto in general and wallet UX still leaves much to be desired, meaning that for a typical consumer, receiving payments and token incentives through a wallet still represents a barrier to entry. However, there are new technologies, such as zkLogin and account abstraction, which should significantly reduce these barriers over time. In the end state, supplying a resource to a DePIN and receiving income from the protocol shouldn’t look that different from receiving interest revenue in your bank account!

One interesting question to ask is whether DePINs can ever have too much supply. For Commodity Hardware DePINs, excess supply can be defined as when the network’s average utilization is below a certain threshold (e.g. 80%) for a prolonged period of time (e.g. 3 months), which could mean that the protocol is overpaying its suppliers, assuming token rewards are being emitted. However, said networks, such as decentralized compute networks, also want to have sufficient capacity during periods of peak demand, similar to how electricity grids in the UK need to be able to handle the peak electricity load when everyone turns on their kettle at halftime during a football game. Balancing those needs will be critical to maintaining a sustainable network. Too much, and the protocol is likely paying too much through incentives and suppliers will have their resources constantly under-utilized and thus likely be underpaid, with rational suppliers leaving the network in the long-term. Too little, and the network will never have sufficient excess capacity for users to consume, and this would likely result in a deteriorated user experience, e.g. having to wait a long time before there is additional storage capacity in a decentralized storage protocol. Note that this argument mainly applies to the long-term state of a Commodity Hardware DePIN protocol, as in the early bootstrapping stages, it is likely that most Commodity Hardware DePIN protocols will face a lack of structural demand, as large consumers likely won’t come in until a significant amount of resource capacity is available and the network is proven viable.

On the other hand, for Custom Hardware DePINs, the definition of excess supply can be slightly more blurred. For example, for Hivemapper, would that be considered too frequent if the average section of a roadway was mapped every 3 days instead of every 2 weeks? For some use cases, perhaps, but for others, maybe not. Or, for example, if Helium had 10 mobile hotspots in the same building, would the marginal mobile hotspot in the same building provide any noticeable marginal benefit to the consumers? This question has to be answered as it gives us a view of what is the end state for many of these Custom Hardware DePIN protocols. In that end state, there can certainly be a scenario where there is excess supply, and in that case, the protocol has to grapple with the fact that due to that excess supply, the individual resource provider may perhaps not be profitable. Most Custom Hardware DePIN protocols distribute revenue to the specific unit of supply that is utilized/consumed, which means that the market should naturally adjust based on demand, and rational suppliers should leave the network given that they are operating at a low or negative profit margin. Eventually, the market should find a stable equilibrium between supply and demand, however, it will be interesting to see how that pans out in a decentralized world.

One could argue that in Custom Hardware DePINs, for example, where you are attaching a dashcam to your car as you drive to work, or Commodity Hardware DePINs, for example, where you are supplying your excess compute from your computer, both of these groups are incentivized to continue supplying the resource as their operating cost is near zero. This is particularly true if those operating costs are incurred regardless of the resource provision. If you are driving to work, operating a dashcam does not make a difference, and similarly, if you are working from home using a computer, you still require your computer even if you can’t provide hard drive storage space to a decentralized storage protocol. As such, there is a potential end state where such networks are consistently underutilized and have excess supply. This could be considered the end goal—to enable individuals to earn additional income on latent/idle resources!

The above is also one of the larger benefits of DePINs, the structural cost arbitrage. For example, Verizon has to lease space to put their mobile towers on top of. For Helium on the other hand, the real estate leasing cost is free as users are placing mobile hotspots on their own premises. In addition, Verizon has to pay for operational costs such as maintenance, whereas for Helium, those costs are borne by the suppliers. Without turning your Helium hotspot back online, you are the one losing out on potential revenue, not Helium. Another example of this is Google paying individuals to drive around for Google Maps imagery data. In contrast, Hivemapper suppliers don’t need to be paid to drive around with a Hivemapper dashcam as they often drive around for other purposes. Or, for example, Nvidia has to pay for all the maintenance/operational/electrical costs associated with its data centers, whereas for decentralized compute protocols, those costs are borne by the suppliers in the course of their daily external use.

Another point of consideration for Commodity Hardware DePINs is that switching costs are often zero or near zero. For example, in the long term, individuals should be able to supply storage or compute with one-click configurations, even if that is not the case today. At this point, you may be thinking, what stops individuals from constantly switching between various networks to supply their resources? While valid, and while nothing is stopping individuals from doing so, historically, we have mostly seen such vampire attacks of a similar nature fail in DeFi, and similar dynamics will be at play for Commodity Hardware DePINs. Even if resources switch over to a newer protocol providing more incentives, more often than not, those incentives are not sustainable. Inherent to this is the network effects that Commodity Hardware DePINs have. When more people use the network, its value increases, both on the supply side as income increases as there is more consumption of the resource and, to a certain extent, on the demand side, as the product quality increases as there is more of the said resource being supplied. As such, the long-term Commodity Hardware DePIN market structure likely trends towards a more monopolistic nature across geographies compared to centralized counterparts.

Extending the conversation to Custom Hardware DePINs, given the custom hardware required to supply resources to these networks, the switching costs are much higher than Commodity Hardware DePINs. However, one possible scenario where this case does not stand is a future where a company builds a “modular” enclosure that allows individuals to connect different hardware components in a plug-and-play fashion into different slots (think Rasberry Pi), or where when hardware specifications are eventually open sourced, a company builds a miner that supports multiple networks offering the same product simultaneously. Similar to the network effects of Commodity Hardware DePINs, the monopolistic structure should be even stronger given the higher switching costs. 

Demand Side

Historically, some DePIN protocols have incentivized both the supply and demand side. However, incentivizing the demand side means that protocols will never be able to effectively evaluate whether they have true product market fit. It’s easy to temporarily switch to using a DePIN if you are getting paid to do so, similar to how users would often choose Uber or Lyft depending on which was giving out a larger discount code. Thus, one could argue that DePIN protocols should never incentivize the demand side, and focus on building out the supply side using existing token incentives. Having said that, that doesn’t mean that DePIN protocols should not put any effort into building out brand awareness to educate potential customers about their unique selling proposition. It is also likely that protocols will incentivize demand as it’s another avenue for competition, and achieving the first meaningful user base increases the protocol’s likelihood of continuing as the market leader in the long run. In addition, DePIN protocols should be cognizant of what customers actually want from a resource network; it is not always simply a matter of lower cost.

One clear example of this is decentralized compute. Despite the clear cost advantages of decentralized compute, most decentralized compute protocols are nowhere close to 100% utilization. A similar case can be made for decentralized storage/streaming/transcoding/etc. DePIN protocols have historically not given the demand side a significant amount of attention, simply hoping that users become aware of the unique selling proposition and eventually convert to paying customers. That lack of marketing and brand awareness through a centralized entity is a significant problem, especially compared to existing Web2 incumbents. Furthermore, many of these DePIN protocols have historically tried to target consumers, e.g. Helium is cheaper than T-Mobile, so switch your family plan over to Helium, but reaching such consumers through traditional advertising channels is likely something that most protocols aren’t prepared to execute yet.  

This brings us to the other type of customers that DePIN protocols should target: businesses. This is especially the case for Commodity Hardware DePINs, as B2B or B2B2C focused strategies give Commodity Hardware DePIN protocols a much more targeted user base, which can be approached through a centralized entity instead of purely hoping that the market eventually finds out about its lower-cost solution. Furthermore, such a business model comes with more predictable and long-term revenue. One example of this on the Custom Hardware DePIN side is Helium signing carrier offload deals, acting as a carrier offloading avenue for telecom companies. However, there is a significant barrier to business demand, that being the lack of SLA/insurance/centralized points of interaction. If I’m a business using a storage/cloud/compute/etc. solution, I want to be guaranteed a certain level of service quality. For example, I might want a 99.999% uptime guarantee, specific ISO qualifications, a 24/7 account manager and customer support, and special pricing on a tiered basis. 

In traditional infrastructure networks, these are merely contractual guarantees, and all faith is placed upon a centralized entity to maintain such guarantees. For DePIN protocols where the counterparty is a decentralized network of resource providers, they will instead demonstrably show the uptime history over months and years, and with SLAs being enforced at the protocol level through crypto-economics. It is important to note that we are still in the early days of DePIN protocols, and as such, it will be interesting to see what type of crypto-economic mechanisms (staking, slashing, withdrawal queues, emissions, etc) will be implemented to achieve “SLA” like guarantees. If anything, DePIN protocols will provide scalable and resilient alternatives to both compete with, but also supplement, existing non-crypto products. One example of this is Infura and other other RPC providers now using POKT Network for backup and failover traffic. 

This brings us to an interesting conversation about how DePINs will scale and the centralization/decentralization dynamics at different levels of the protocol. In the end state, most customers will get their infrastructure via end-user-facing gateways/frontends (that effectively act like L2s to the underlying protocols). As mentioned earlier in the report, this is exactly the strategy that POKT Network is taking, pioneering demand-side decentralization via their ‘Gateway’ ecosystem. For DePIN protocols, these centralized gateways/frontends can provide ISO qualifications, 24/7 support, value-added features, enterprise or consumption-based pricing, etc, with the base layer DePIN protocol and its supply network powering said gateway. At scale, this is a symbiotic relationship where DePINs can provide immense cost and scale benefits that centralized entities cannot match, while the centralized gateways can provide the value-added products/services that only a centralized entity can support. As always, decentralization is a spectrum, and not all layers of a DePIN product have to be decentralized.  

So then, if targeting consumers is too hard and fragmented, but targeting businesses is also not feasible given the lack of certain product/service guarantees and additional services offered, how do DePIN protocols move past that barrier? The potential solution here, at least for Commodity Hardware DePINs, is to also take a partial B2B route on the supply side, and thus, the protocol can provide service-level guarantees to buyers by also requiring service-level agreements on the supply side. Commodity Hardware DePINs should be able to guarantee a certain acceptable baseline level of capacity, which is then supplemented by consumers supplying resources.

Another solution is to dogfood your own demand. For example, AIOZ Network is a DePIN protocol that offers various solutions for Web3 storage, decentralized AI computation, live streaming, and video-on-demand (VOD). Their own standalone blockchain utilizes a dBFT consensus mechanism and is connected to EVM/Ethereum through the Gravity Bridge and Cosmos-based chains through IBC.

AIOZ Network has a few key products. W3S is a decentralized storage solution that provides an efficient and affordable data storage experience. It has S3 (Amazon’s Simple Storage Service) compatibility that allows users to switch over seamlessly and features over 70K+ edge nodes, enabling an average download speed of 151MB/s. W3IPFS is a similar solution, but primarily for NFT storage. W3Stream allows the protocol to offer on-demand live video streaming by leveraging a decentralized content delivery network (dCDN). Potential customers include media and entertainment companies, video streaming platforms, and e-learning platforms that need to deal with large volumes of video, audio, and image content and require streaming bandwidth. Lastly, W3AI has two components: inference and training. Inference enables users to conduct decentralized model inference, while training allows users to access decentralized model training & finetuning. Furthermore, a W3AI marketplace empowers users and organizations to contribute AI datasets and models, monetize their work, and build and deploy AI dApps.

In AIOZ Network’s case, they successfully dogfood demand for their own product. For example, when a customer utilizes W3AI for model training, they might require storage to store various datasets or encoded data required for computing tasks. Or, for example, when a customer utilizes W3Stream for on-demand live video streaming through AIOZ Network’s dCDN, they will utilize AIOZ Network’s GPU resources that are supplied to their W3AI solution, while also using W3S to store any media that needs to be streamed. It is entirely possible that eventually, the market will see Commodity Hardware DePIN protocols that operate in multiple distinct verticals, yet have synergies in terms of cross-product resource demand.

Regarding the question of marketing to a broader audience and building brand awareness in the real world, that’s likely a question for much further down the road, as DAO governance is very much in the early stages, and even the most developed DAO structures in DeFi are still iterating on the long-term ideal state of governance. Realistically, we will likely replicate the Foundation + DAO structure that many L1/L2s have today, with electable and replaceable committees that manage key business and operational functions.

On the demand side, users should not be aware that they are using a product/service facilitated by crypto rails. If anything, the experience should be no different than how they would interact with a centralized Web2 resource provider, with all of the functions of a typical company, such as marketing, customer support, etc. The demand-side user complexities are easier to abstract away, especially so if there are no token incentives involved. Users should be able to pay with fiat payment rails, and then receive the product/service near instantly. However, in the long term, it is possible to have this be achieved in a relatively decentralized manner, similar to how one DeFi protocol can have multiple front ends.

Tokens

Token incentives are a crucial component of DePINs. I’ve briefly covered why tokens are so important, especially in that they kickstart the flywheel in building out the network and give early users ownership and incentive alignment of building out the network. If you’ve been in crypto for a while, you must have heard stories about individuals collecting six or seven figures worth of HNT, Helium’s token, as they bought more and more miners, eventually watching it all collapse in spectacular fashion. New DePIN protocols have learned from the mistakes of the past and designed more sustainable and incentive-compatible token emissions.

For example, POKT Network, like many other DePIN protocols, fell victim to extremely high inflation during the bull market. However, thanks to 3 critical DAO proposals (PUP-11, PUP-13, PUP-32),  inflation today stands at <5%, with a target of <2% by end of year, and deflationary by end of 2025. Such forward-thinking and, importantly, proactive discussion and implementation by DePIN protocol DAOs with respect to token emissions is perhaps one of the most important factors for a protocol’s long-term success. It represents the network’s ability to actively respond to changes in market dynamics, cycles, and structure.  

As mentioned earlier, token incentives allow DePINs to build up the supply side of the network to attain a minimum threshold, whether that be geographic coverage, amount of resources supplied, or some other soft capacity/performance guarantee. Token emissions are a way for suppliers to derisk the initial capital outlay and to have sufficient incentive to connect supply to the network, as well as to reward users who actually contribute to building out the network. Importantly, Web2 scale is nearly impossible to break through for new entrants in infrastructure networks, given the regulatory moat and high capital outlay. Utilizing token incentives allows new infrastructure networks to be built using a bottoms-up go-to-market strategy.

Early DePIN protocols often had excessively high token emissions, which worked outstandingly well in rapidly growing the network when prices were going up but also acted in an equally reflexive manner when token prices started spiraling downward. Newer DePIN protocols have partially learned from the lessons of the past and as a result, have much lower token emissions from the get-go. However, there is still much work to be done in finetuning token emissions. This would be a good point to note that token emissions are very network specific. There are general guidelines and principles that one should adhere to, however, token emissions ultimately depend on the type of DePIN being built, different demand and supply dynamics, and more.

Regardless of the type of DePIN, token emissions should be granular and customizable. Granular in the sense that token emissions shouldn’t be distributed pro-rata to every single unit of supply connected to the network. Instead, token emissions should be distributed to incentivize specific types of behavior or a specific type of supply in specific types of locations, especially for Custom Hardware DePINs in a more granular fashion. As mentioned earlier in the report, DePINs have specific product/service thresholds they would like to meet. That could be anything from a certain level of compute power, a certain level of geographic coverage, or a certain level of network capacity. For Commodity Hardware DePINs, such networks can afford not to be so granular, as ultimately, the resource being supplied is relatively similar across users. For example, someone supplying storage to Arweave through a desktop computer in the USA is providing almost the exact same resource to someone supplying storage to Arweave through a data center in Japan. However, for Commodity Hardware DePINs, there are often latency considerations, and as such, geographical distribution is still a goal to work towards. In that scenario, a Commodity Hardware DePIN protocol should for example, overweight token emissions in a geographic region where they are lacking suppliers and underweight token emissions in a geographic region where they have an excess proportion of suppliers. Or, when it comes to, e.g., decentralized compute protocols, they may want to overweight incentives for specific types of powerful GPUs, such as H100s, that can run large model training and inference, that other GPUs perhaps may not be able to support unless provided in large quantities.

Similarly, for Custom Hardware DePINs, protocols should refine their token emissions to be even more granular than Commodity Hardware DePINs, as often, a good product/service requires a certain level of geographic density or coverage in the resource being supplied. For example, Helium should focus token emissions for bootstrapping supply in specific geographic regions, rather than global coverage. Instead of token emissions being distributed pro-rata to suppliers regardless of location, token emissions should be distributed to hotspots being connected in certain cities, then certain states/provinces, and then certain countries. No user in the USA will switch mobile carriers if the competitor only provides coverage in, for example, New York, or only on the East Coast. The DAO or Foundation should think carefully about what KPIs they are optimizing for, specifically, what minimum product/service quality threshold would be sufficiently attractive for users to switch over from traditional counterparts. Then, token emissions should be utilized to grow supply until said product/service quality thresholds are met before being redirected to growing the product/service in other geographic areas.

However, one must also consider that suppliers also want predictability in their revenue, whether that be from token incentives or from paying customers. Given that most DePIN protocols are currently demand-constrained and realistically, do not have any predictable revenue, it would be wise for protocols to bake in demand-focused elasticity into their token emissions rate so that suppliers are earning some “baseline” level of revenue, ensuring that supply stays connected to the network. Note that this baseline revenue does not need to be particularly high, as from the supplier’s point of view, any hardware purchased is already a sunk cost, and operating costs are likely minimal (e.g. electricity for a Helium hotspot doesn’t cost much, and likely no one is driving around with a Hivemapper dashcam solely for the purpose of mapping).

If demand and thus revenue is high, token emissions should decrease, and vice versa, to ensure that the supply stays online in the network. Token emissions should be responsive to both supply and demand, and tailored to some KPI. A simple implementation that takes into account both supply and demand would be network utilization for Commodity Hardware DePINs. For example, a DAO could target a certain network utilization percentage. If there is excess supply in the face of low demand, then network utilization is low, and thus, token emissions to the supply side should be reduced. However, there are certainly major caveats here, such as what if demand is low because there is insufficient supply on the network? If there is a lack of supply in the face of high demand, then network utilization is high, and thus, token emissions to the supply side should be increased. However, many of these arguments largely apply in the stages where the network is still growing. In the end state where the network is built out and “stable,” token emissions should either be zero or there should be some terminal emission rate utilized for overall network inflation, perhaps counteracted by some deflationary mechanisms.

Ultimately, a token economy cannot be sustainable without fiat inflow. Said fiat inflow stems from users purchasing some product/service that the DePIN is providing. This can manifest in a few different ways. For example, users can buy the token and then use it to purchase resources. Or, for example, users can pay in US dollars, and that capital is then used to buy the token, which is either distributed to suppliers or burned. There are different methods through which a token can accrue value, but in most cases, a percentage take rate is implemented, e.g. 20%. The accrued revenue to the protocol is then held in a treasury, used to buy back tokens, or a combination of both. 

A typical token model used by protocols such as Render and Hivemapper is the burn mint equilibrium (BME) model. In a BME model, funds used to purchase a resource are used to purchase a DePIN’s native token on the open market. Said tokens are then burned, and an equivalent amount is minted to suppliers whose resource was consumed.

Evaluating DePIN Protocols: The Good and the Bad

There isn’t an established broad framework for evaluating how good/investable a DePIN protocol is. The below paragraphs are some of my thoughts on how one should think about investing in this specific sector. It is likely that there will be future reports diving into specific DePIN protocols and their investability.

The criteria below are split into three key categories, those being market, product, and tokenomics. The criteria below are by no means an exhaustive list, and there are many nuances depending on the sector the DePIN protocol is competing in.

A few things to look out for when evaluating a market that a DePIN protocol competes in include:

1. Market size: What is the total addressable market? What is the total revenue opportunity when looking at non-crypto counterparts? What is the serviceable available/obtainable market? 
2. Market growth: What does the market CAGR look like? Are there any secular trends or tailwinds that would accelerate market growth?
3. Non-crypto competitive landscape: What does the existing competitive landscape look like? If non-crypto counterparts have a monopolistic market structure, it may be harder for a DePIN to compete, given the monopoly power compared to a market with a perfect competition market structure.
4. Crypto competitive landscape: Does the DePIN protocol have a first-mover advantage? Have they managed to establish significant network effects? 

Following the market analysis, a few things to evaluate on the product side include:

1. Value proposition: What is the unique selling proposition of the DePIN? Is the DePIN only competing on cost (this is not necessarily a bad thing, but in an ideal world, there is another key differentiating factor, especially for Custom Hardware DePINs)? 
2. Buyers: Does the DePIN protocol take a B2C, B2B, or B2B2C GTM strategy (the latter two are more ideal, but this is situational)? Can the DePIN clearly describe its target customer? Is there genuine long-term demand (not just one-off trials by users) for the product? Is there strong absolute growth in demand and revenue generated? 
3. Supplier set: How diversified is the supplier set? What does the average supplier behavior look like? Is supplier retention high? How much organic supply growth is there, or is it primarily driven by token incentives?
4. User experience: How abstracted and seamless is the user experience, both on the supply and demand sides? Are switching costs low or high? Is onboarding friction low or high (e.g. how complicated is it to supply a resource to the network, how complicated is it to switch from an existing product to this DePIN solution)?

Lastly, on the tokenomics front, the following questions should be asked:

1. Token emissions: Does the protocol have well-thought-out token emissions? Is the protocol overpaying for supply incentives? Is there a targeted granular emissions policy that aims to achieve various KPIs, or is it more of a spray-and-pray approach? 
2. DAO: Does the protocol/DAO actively react to new market dynamics and adjust emissions/inflation accordingly? Are foundation/treasury tokens used effectively to increase brand awareness and drive forward critical, useful initiatives? Is DAO governance active, and is there an established and functional organizational structure?
3. Value accrual: How does the token accrue value? Is there some buyback and burn mechanism or fees distributed to tokenholders? How is value passed on to suppliers? Is there any utility to the token, e.g. the ability to use it to vote on emissions changes?
4. Tokenomics: Does the token have sensible vesting schedules? Is token emissions entirely programmatic or do various stakeholders have the power to change it? What is the long-term token inflation, or is there a finite long-term supply?

Risks

There are several key risks to DePIN. The first and most obvious is regulatory risk. Given that many DePIN protocols compete in largely monopolistic/oligopolistic industries, there is a strong incentive for existing incumbents to establish a moat through regulatory capture. This has occurred time and time again when large existing corporations are faced with exciting new startups who sometimes have a far superior product but ultimately fail to go to market due to regulatory action or pushback. Furthermore, there are certain resources where the government may want oversight or force certain laws. For example, recent executive orders that would compel US cloud companies that provide computing power for foreign AI training to report they are doing so would be hard to implement in a decentralized network. Or, telecommunications regulations that require mobile service providers to save sufficient bandwidth for emergency communications would also be equally hard to implement in a decentralized network. In addition, there are still many unanswered questions about data privacy for DePIN protocols. For example, managing personal or sensitive user data that one may submit (e.g. complying with GDPR), requirements to monitor user activity within specific use cases, complying with reporting requirements or law enforcement requests, or user liability issues that may arise.    

Another key risk for DePINs, especially Commodity Hardware DePINs, is if these infrastructure networks never reach product quality parity with existing incumbents and always have a long-term lack of demand for their product. Structurally, this could mean that Commodity Hardware DePINs such as decentralized storage, compute, and more never find product-market fit from a B2B perspective, as businesses are happy to pay more if they receive a better service. Similarly, on the consumer side, there is an insufficient unique value proposition for some DePINs. For example, Google Drive works perfectly fine for most of its users, and given that a Google One subscription is already relatively cheap in the grand scheme of things, a lower-cost alternative may not be sufficiently attractive to make the switch. In addition, switching costs for some of these solutions may be prohibitively high from a practical perspective, e.g. how would I migrate all of my files to a decentralized storage solution?

Similarly, tokenomics is excruciatingly hard to implement well from the get-go for many DePIN protocols. Given the wide range of factors that one has to consider and the many nuances, including responding to supply and demand dynamics, real-world competitor responses, onchain liquidity, geographical considerations, crypto market cyclicality, and more, it is entirely possible that consecutive fumbles in tokenomics design ultimately result in an unsustainable emissions model that is hard to come back from.

Lastly, in lower interest rate environments, it is possible that DePINs cannot match cheap debt financing. As traditional infrastructure networks grow, they will be able to borrow more against their balance sheet assets or raise increasing amounts from private funding and proceed to use that capital to build out their performant infrastructure networks. Despite this not making a lot of sense, it is entirely possible for existing non-crypto infrastructure networks to raise capital and then use that cash to incentivize consumer suppliers to supply a resource to a network, in the same manner as how token incentives are used to incentivize suppliers. However, this is unlikely without creating a DePIN since coordination becomes much harder. Imagine a centralized telecom company trying to coordinate mobile hotspot deployments across the entire USA in a permissioned fashion. The key difference here is that in the non-crypto version, suppliers receive fiat and no ownership in the network (which is entirely possible if people just don’t care that much yet), and in the crypto version, suppliers receive tokens and ownership in the network.

Final Thoughts

Where does that land us today? The simple fact is that most DePIN protocols trade much more on narratives, with category leaders receiving a notable market premium, and those two facts are orders of magnitude more important than cash flow or token value accrual. All one has to do is head to sites such as depin.ninja, depinscan.io, or depinhub.io, look at 30D revenue and compare that to the token’s market cap to realize there isn’t any token trading on fundamental valuations. Which makes sense! Nobody invests in a startup by doing valuation models on existing cash flows because they are generally exceedingly negative and it’s impossible to reliably incorporate the potential future parabolic growth into a model. Most assumptions are based on market size, the growth of that market, and how much of that market one thinks they can capture (assuming it isn’t a non-existent market that the protocol intends to create for itself, which mostly isn’t the case for DePIN protocols). Based on income, the largest DePIN protocol is Filecoin, generating $714K in 30D revenue, yet the total global cloud storage market size was $108.69B in 2023, just to offer a sense of comparison of how early most DePIN protocols are.

Investing in a DePIN protocol is no different from having the opportunity to buy AWS/Google Cloud/Nvidia Data Centers/insert any large tens or hundreds of billion dollar infrastructure network in the early days. The first thing to do is be honest with yourself. Are you trading these DePIN protocol tokens or do you see yourself investing in these for the long term? If you are trading these, I would recommend you go read this fantastic list of texts. If you are “investing” in these tokens for the medium-long term, the best way to invest is in a combination of whichever you think will have the strongest narrative (at this stage of crypto’s maturity, as you will still need retail to get excited about the sector), and whichever protocol is trading at a cheap valuation relative to how much they have derisked (have a sticky and large distributed supplier set, genuine long-term sustainable demand, a clear path to scaling and taking over revenue from non-crypto counterparts, and a clear market penetration strategy). DePIN is likely the future of infrastructure networks, and there will be multiple hundred billion dollar projects built in this category, but for now, long live narrative domination.

This research report has been funded by AIOZ Network and POKT Network Foundation. By providing this disclosure, we aim to ensure that the research reported in this document is conducted with objectivity and transparency.  Blockworks Research makes the following disclosures: 1) Research Funding: The research reported in this document has been funded by AIOZ Network and POKT Network Foundation. The sponsor may have input on the content of the report, but Blockworks Research maintains editorial control over the final report to retain data accuracy and objectivity. All published reports by Blockworks Research are reviewed by internal independent parties to prevent bias. 2) Researchers submit financial conflict of interest (FCOI) disclosures on a quarterly basis that are reviewed by appropriate internal parties. Readers are advised to conduct their own independent research and seek the advice of a qualified financial advisor before making any investment decisions.