DePINS In Detail: Understanding The Matchmaking Process At The Heart Of Node Selection

Decentralized Physical Infrastructure Networks are getting a lot of attention thanks to the huge interest in artificial intelligence-powered applications. These days, almost every company believes there are big business advantages to be had from harnessing generative AI and related technologies, giving rise to what some have likened to an “AI arms race”.

Much of the focus is on the underlying AI Models, such as OpenAI’s GPT family, Google’s Gemini and Anthropic’s Claude, but there’s an equally interesting battle shaking up in the infrastructure space, as cloud providers scramble to provide the resources companies need to run those models.

The likes of Amazon Web Services, Google Cloud and Microsoft Azure currently dominate the AI Infrastructure space, but DePin provides businesses with a novel alternative to hosting their models on centralized cloud infrastructure.

As the name suggests, DePins provide access to decentralized computing infrastructure that’s not located in corporate-owned data centers. Instead, the resources are crowdsourced from vast networks of users who are willing to donate their underutilized GPUs to third-parties, in return for financial compensation. The basic idea is that, if you have a laptop with a GPU that you’re not using 24/7, you can outsource your computing power to a DePin network so others can use it when you’re not doing so yourself. It’s a win-win for everyone involved, with GPU owners earning a passive income, and GPU users getting access to a lower-cost alternative to traditional cloud-based GPUs.

How do DePins work?

Numerous DePin networks have emerged in the past few years, and although they operate quite different models, there are certain similarities they all share. For instance, every DePin network relies on blockchain, using the distributed ledger technology as the backbone of their operations. Blockchain does the job of collecting and categorizing all of the information about the available hardware nodes, making it available to potential users. It facilitates the connections between DePin participants, recording all contributions and distributing the financial rewards in the form of tokens.

Smart contracts also play a key role, automating DePin transactions. Smart contracts are self-executing agreements that trigger actions when certain, predetermined conditions are met, such as payments being made or a service being provisioned. The use of smart contracts means there’s no need for an intermediary to handle things in the background. As a result, processes can be streamlined with lower costs.

Cryptocurrency is used to power the DePin economy. Customers who want to access DePin resources are required to pay using digital tokens, which are then distributed to those who provide resources as a financial reward.

The matchmaking process

Most DePin protocols implement an algorithmic matchmaking process in order to connect consumers with suppliers.

Spheron Network operates one of the most comprehensive and open matchmaking models within its DePin network, known as the Decentralized Compute Network. Spheron’s goal is to connect high-performance computing users with the computing power they need to perform tasks such as AI training, machine learning, scientific simulations and CGI rendering, without having to resort to expensive, centralized cloud providers.

The DCN network operates a transparent, on-chain Supply Market which facilitates the transparent trading and allocation of GPU resources, with key elements being the Matchmaking Engine for allocating resources to users, and a crypto-based Payment System to facilitate transactions.

Spheron’s matchmaking engine does the job of processing user’s orders in order to pair their request with the most optimal GPU provider/s, based on the specific requirements of the user. It’s a complex algorithm that evaluates a number of parameters to ensure the best possible match, including the region or availability zone, prioritizing those nodes with the closest proximity to the customer to minimize latency and ensure adherence to any data residency obligations.

The price delta helps to ensure economic efficiency by matching the user’s deployment budget with competitive provider bids, while the uptime/availability parameter gives priority to those nodes that have a track record of reliability. Each node’s reputation, such as its historical performance and standing among network participants, is also factored into the equation, as is resource availability, which considers the node’s capacity to meet the deployment requirements. Slashes, which are the penalties applied to nodes for past contract violations, are another factor that goes into the calculations, and there’s also an element of randomness and unpredictability, which helps to safeguard against any deterministic bias in the node selection process.

Spheron explains that each of these parameters is crunched through a robust algorithm that’s designed to select the most optimal resource provider for each deployment, taking into account the user’s predetermined criteria. Once the best possible node has been selected, the Matchmaker Node executes an on-chain transaction using smart contracts to officially document the allocation, so deployment can begin.

A rival DePin network called Akash Network, uses a somewhat different mechanism that combines a similar set of parameters with a novel, reverse-auction process to match user’s requests with resource providers.

The process begins with the customer (tenant) who defines their application or workload requirements, specifying the amount of GPUs, CPUs, RAM and storage they require, along with network bandwidth and other details. These requests are compared with the available resources of the network.

Once suitable resource providers have been identified, a reverse auction model is employed, where tenants set the price and terms of their deployment, and providers bid to facilitate their work. To ensure that providers are serious, they are required to place a refundable deposit with their bid. The tenant then selects what they consider to be the most advantageous bid, and a lease is created between them and the winning resource provider.

The two matchmaking models are notably quite different, geared towards different advantages. In the case of Spheron, its matching engine is designed to allocate the most optimal resources to each deployment, ensuring the best possible performance for end users. Meanwhile, Akash uses its reverse auction process primarily to ensure the lowest possible costs for end users, while providing optionality and flexibility of choice.

Disrupting Cloud Infrastructure

DePIN networks are emerging as a disruptive force in the high-performance computing infrastructure market, providing companies and individuals alike with a viable alternative to traditional cloud infrastructure.

With its emphasis on decentralization, community engagement and resource sharing, DePIN looks well-placed to meet the growing demand for sustainable and efficient infrastructure solutions.

As the demand for AI computing power increases, the novel match making processes utilized by DePIN networks will play a significant role in catering to user’s infrastructure resource needs, while expanding privacy and the reducing costs of cloud infrastructure resources.