Traditional computers are composed of five parts: the computer, memory, controller, bus, and I/O. From the perspective of blockchain development, the progress of the computer and memory components is relatively mature. If we compare the entire distributed system to a human, then the brain and memory systems are already well-developed, but the sensory and perceptual systems remain in a very primitive state. At this stage, DePIN is undoubtedly the most popular buzzword, but how can it be realized? It undoubtedly starts with “trustworthy touch,” and as we know, “sensation” relies on the spine and nervous system for processing.

If blockchain systems represent the consciousness built on an iceberg, then the sensor networks represented by DePIN are the subconscious beneath the iceberg. Now, the challenge arises: who is the spine and nerves of the distributed system? How do we construct the spine and nerves? In this article, we will start with small lessons from the development of the Internet of Things (IoT) to construct the development ideas of DePIN and help builders better implement them.

TL;DR

1. Depin should not be based on devices as units because devices lack horizontal scaling capabilities. Instead, it should focus on modules. The core of depin lies in the Pin, and the core of the Pin is the authorization code. We regard a device as a collection of sensor modules, and each sensor module’s pin code is the permission for data to join the network and also the PoPW authentication permission. Only devices with network access permissions and those whose contributions are recognized can be called mining machines. Therefore, the core of the entire depin sector is how to make edge devices contribute measurably and how to ensure consistent metrics for contributions from different devices with the same sensors.
2. According to traditional computer data transmission, buses (Bus) can be divided into three categories: the Data Bus for transmitting various data, the Address Bus for transmitting various address information, and the Control Bus for transmitting various control signals. Similarly, the DePin bus will have the following components: as the identity credential for devices joining the network (Address Bus); as the PoPW credential for data verification (Data Bus); as the means for device management (Control Bus).

a. Address BUS: Device DID (Dephy)
b. Data BUS: Virtual Communication Layer + Sensor Network
c. Control BUS: Cellular Management Module

1. Due to its partial RWA attributes and connection to the physical world, the Depin project is relevant to real economic life. Therefore, more real-time management methods are needed for autonomous risk control. There are two main implementation channels: Firstly, through governance of cellular operator traffic, once a device violates regulations, it can lose PoPW mining rights from the traffic end, which is a more real-time management method compared to slashing. Secondly, by buying out upstream resources through a miner + resource pool model. For example, if a distributor owns 100 number segment resources and 30 of them are at risk, they could face penalties or warnings about license revocation. Today, we mix these 30 resources with those of other distributors, buy out real-world resources (RWR) through miners, and use a mixed number segment approach for resource risk control. This ensures maximum resource acquisition under the premise of safeguarding upstream distributor risks. The Liquity model is replicated across various types of RW resources.

1. A Review Of The History Of The Internet Of Things

Looking back on the history of IoT development since 2015, there were two main challenges that year: firstly, hardware devices had limited input-output capabilities; secondly, after devices joined the network, their product features did not enhance, lacking scalability.

During this period, the key question was: what changes would occur when hardware devices’ microcontrollers joined the network? Initially, connectivity enabled hardware devices to upload and download data. The subsequent question was: why do hardware devices need to upload and download? Can these actions enhance product competitiveness? At that time, we saw a wave of products like smart curtains, smart air conditioners, etc. However, due to the relatively fixed I/O architecture in hardware design and limited space for software development, the addition of network connectivity mainly offered features like mobile app control, such as “remote air conditioning activation” and “remote curtain closing”. These functionalities were primarily remote extensions of traditional controllers, which were somewhat underwhelming for end users.

Another crucial issue was whether IoT devices had the ability to scale after connecting to the network. As mentioned earlier, network connectivity enabled data upload and download. While downloads represented functional upgrades and expansions, uploads facilitated data aggregation and integration. However, during the early IoT era, the value of data lakes was cumbersome due to exponentially rising storage costs and challenges in tapping into data sales opportunities.

In summary, IoT devices in both download and upload modes struggled to enhance product capabilities and service dimensions. Looking ahead to the Depin era, can these challenges be overcome?

What Changes Has AI Brought?

From the characteristics of AI, we see many possibilities:

1. Anthropomorphism of Everything: Independent upload and download requirements. If edge-side inference cannot handle large models, then endpoint devices will need independent networking. This will shift the past structure where mobile endpoints were stars and devices were satellites to a communication structure where devices independently connect to networks.
2. Device Sovereignty: Moving from simple product sales to a dual-wheel drive of user purchases and data sales. Devices are accountable to users as a whole and responsible to data merchants as sensor collections.
3. “Data Trustworthiness, Reliable Privacy”: These are prerequisites for ordinary devices to transform into mining machines. If data is untrustworthy, logically, opening multiple virtual machines could hack the entire incentive system. If privacy is unreliable, long-term user interaction intentions will be inhibited.

In conjunction with AI development, we see several potential differences for Depin:

1. The emergence of AI increases the necessity for AI hardware to autonomously connect to networks. The cost of device networking may rapidly decrease in the next three years, combined with reductions in storage and computing costs, significantly lowering the cost of edge computing/sensor deployment. Once many devices are deployed, converting them into mining machines to collect sensor data may reach a tipping point.
2. Once the issue of independent connections between devices and the cloud is resolved, there will be more scenarios for interconnection between devices. Exploring interactive uses with various low-cost hardware like NFC could become potential innovation points.
3. The commodification of various collected perceptual data is a core bottleneck for device mining. Establishing standards for abstract information commodities is a major challenge.

2. Investment Themes and Perspectives on Depin:

Based on the past 5 years of IoT development experience and the changing landscape of AI features, we believe there are three major investment themes:

• Cellular modules as the core hardware infrastructure.
• Abstract communication layer services centered around communication information commodities.
• Broad mining as a form of distributor service.

Investment Theme One: Depin Infrastructure Centered Around Address Bus Modules

What is a module?

A module integrates baseband chips, memory, power amplifiers, and other components onto a single circuit board, providing standardized interfaces. Various terminals utilize wireless modules to enable communication functions. As the entire computing network evolves, the definition of modules continues to expand, forming an ecosystem of cellular connectivity, computing power, and edge applications:

• Traditional cellular IoT modules: Basic connectivity modules designed primarily for cellular communication. These modules include chipsets that support this type of connection without additional functionalities.
• Smart cellular IoT modules: In addition to providing connectivity like traditional modules, these incorporate additional computing hardware in the form of central processing units (CPU) and graphics processing units (GPU).
• AI cellular IoT modules: These modules offer functionalities similar to smart cellular IoT modules but also include specialized chipsets for AI acceleration, such as neural, tensor, or parallel processing units (NPU, TPU, or PPU).

Looking at the entire industry chain, upstream chip makers and downstream device manufacturers capture the majority of the value chain. The intermediate module layer is characterized by high market concentration and low-profit margins. Traditional service devices mainly include PCs, smartphones, and POS terminals. Due to their significant concentration, deploying widely accepted module intermediaries essentially transforms various existing devices into mining machines. If traditional Web3 users are considered on a per-person basis, the intermediate layer represented by modules will enable a large number of smart devices to enter Web3, generating a substantial on-chain demand through transactions between these devices.

Reflecting on the early competition between Nvidia and Intel, we gain valuable historical insights: in the early years, the computer chip market was dominated by Intel’s x86 CPU architecture. In niche markets like graphics acceleration, there was competition between Intel’s dominant ecosystem of accelerator cards and Nvidia’s GPUs. In broader markets (areas with uncertain demands), Intel CPUs and Nvidia GPUs cooperated and coexisted for a period. The turning point came with Crypto and AI, where large-scale computing tasks characterized by small tasks executed in parallel favored the computational capabilities of GPUs. When the wave arrived, Nvidia prepared on several dimensions:

1. CUDA parallel computing instruction set: Facilitated better utilization of GPU hardware by developers.
2. Rapid iteration capability: Surpassed Moore’s Law in iteration speed, securing its survival space.
3. Co-opetition with CPUs: Effectively leveraged and utilized Intel’s existing resources, swiftly seizing market opportunities in sensitive decision-making areas.

Returning to the module market, there are several similarities with the competition between GPUs and CPUs in the past:

1. High industry concentration, with leading groups possessing significant pricing power over the entire industry.
2. Development dependent on new scenarios: Communication modules, smart chips, and standard protocols are likely to establish strong barriers at the device end.
3. Opportunities for rapid iteration to seize new opportunities: Traditional players have long decision cycles, making them vulnerable to the rapid changes in emerging scenarios conducive to the birth of new species.

In this competition, the Crypto Stack undoubtedly represents the pinnacle technology stack for building protocols and ecosystems. The migration of existing devices into cash flow mining machines will create opportunities at a beta level. Dephy stands out as a key player in this context, leveraging integrated modules, ledgers, and identity layers to manage the allocation responsibilities across the entire Depin network.

Investment Theme Two: Data Bus - Sensor-Represented Data Collection Mining Machines

What exactly constitutes a mining machine? We believe that hardware/software capable of generating specific information resources and intending to acquire token resources can be termed as mining machines. Under this understanding, mining machines are evaluated based on several criteria:

1. Do they generate specific information resources?
2. Can they settle tokens?

Therefore, in this entire process, the reliability of devices in generating specific information resources, known as Proof of Physical Work (PoPW), becomes crucial. We assert that every sensor producing PoPW requires a Trusted Execution Environment (TEE/SE) to ensure the credibility of edge-side data collection. In the field of sensors, those capable of generating horizontally scalable networks can unify various devices’ video resources, for example, collected by different cameras into a single network for standardized measurement. Compared to independent collection by different devices, horizontally scalable sensors combined with trusted modules can build a larger PoPW resource market. Video materials collected can be better priced according to unified metrics, facilitating the formation of a bulk market for information resources, which is not achievable with Device-Focus alone.

Investment Theme Three: Control Bus - Communication Infrastructure of Generalized Bus

Due to the physical presence of some Depin devices in the real world and their relevance to traditional business society, while the Crypto world features Permissionless characteristics, managing various participating entities in a real-time manner without KYC becomes crucial. We believe that the entire Web3 world needs a communication abstraction layer that integrates cellular networks and public IP networks, where users/devices can access corresponding network services by paying in cryptocurrency. Specific avenues include:

1. Integrating Traffic: Connecting global operator traffic resources, treating traffic as a bulk commodity for trading and pricing with tokens.
2. Integrating Number Ranges: Connecting global number range resources, treating numbers as an identity layer for trading and pricing with tokens, governed by Blockchain.
3. Integrating IP Resources: Connecting public IP resources, integrating public IP pools as a resource for flexible access routing, trading and pricing with tokens, governed by Blockchain.

3. Conclusion

1. Depin should not be based on devices as units since devices lack horizontal scaling capabilities. The core of Depin lies in Pins, and the core of Pins lies in authorization codes. We view devices as collections of sensor modules, where the pin code of each sensor module serves as both the permission for data access and the PoPW authentication permit. Only devices with permission to access the network and contribute recognized data can be termed mining machines. Therefore, the essence of the entire Depin track lies in enabling edge devices to contribute in a measurable way, ensuring consistent metrics across different devices with the same sensors.
2. Different from traditional computer data transmission, which can be categorized into three types: data buses for transmitting various data information, address buses for transmitting various address information, and control buses for transmitting various control signals, the DePin bus will also have similar functions: serving as identity credentials for device access (Address Bus), as PoPW certificates for data verification (Data Bus), and as a means of device management (Control Bus).
3. Due to its Partial Real World Assets (RWA) attributes and its connection to the physical world and real economic activities, the Depin project requires more proactive management tools to achieve autonomous risk control. There are two main implementation channels: firstly, governance through cellular operator traffic, where devices violating rules can lose PoPW mining rights from the traffic end, providing a more real-time management method compared to slashing. Secondly, buying out upstream resources through a miner + resource pool approach. For example, if a dealer possesses 100 number resources and 30 are at risk, warnings of license revocation might follow. Today, we are blending these 30 resources with those of other dealers, applying miner-led real-world resource (RWR) buyouts and segment blending for risk control. This approach aims to maximize resource acquisition while safeguarding upstream dealer risks, replicating the liquidity model across various types of RW resources.

Statement:

1. This article is reproduced from [Foresight Research], the original title is “Foresight Ventures: How to Be Trustworthy—How Do We View the DePIN Track?” 》, the copyright belongs to the original author [Yolo Shen@Foresight Ventures], if you have any objection to the reprint, please contact Gate Learn Team, the team will handle it as soon as possible according to relevant procedures.

2. Disclaimer: The views and opinions expressed in this article represent only the author’s personal views and do not constitute any investment advice.

3. Other language versions of the article are translated by the Gate Learn team and are not mentioned in Gate.io, the translated article may not be reproduced, distributed or plagiarized.