Exploring the DePIN-Driven Robot Economy: Why Robots Need Money

Written by: Thejaswini M A

Translated by Chopper, Foresight News

Have you ever imagined that robots need money? Here, we’re not referring to AI agents that call API services, but actual physical machines. Why would such a device need a crypto wallet or an independent identity credential?

Many people might find this idea quite unusual. However, more than ten crypto companies are already launching related products—let’s explore the business logic behind them.

Starship Technologies operates approximately 3,000 six-wheeled delivery robots across eight countries, primarily deployed in urban centers throughout Europe, with over 10 million commercial delivery orders completed to date.

Serve Robotics, listed on Nasdaq, has expanded its operations to 44 cities including Los Angeles, Chicago, Atlanta, Miami, and Dallas, handling food delivery orders for Uber Eats and DoorDash.

These wheeled cargo devices can carry approximately 20 kilograms per trip and travel several miles on a single charge. If you live in one of the cities mentioned, you’ve likely encountered them on the street.

Suppose you purchase a delivery robot to manage a small delivery operation within your community. This robot operates 12 hours per day, serving as a cost-effective labor solution, and once purchased, full ownership is yours.

Based on the specifications of Starship’s robots, this device can operate for up to 18 hours on a full charge. It will eventually need to return to a charging station to replenish its power. Starship’s solution is to have its robots operate outdoors and return to proprietary charging stations built by the company. This model is practical for Starship—given that the company operates thousands of robots and has the financial resources to deploy supporting charging infrastructure across all its operational areas.

But if you own only one robot, it can only use third-party charging stations. Operators can deploy charging infrastructure similar to how electric vehicle charging stations are laid out; street-side shops can offer charging services to the public, or other delivery providers can open up their idle charging ports. The robot’s battery capacity is comparable to that of an electric bicycle, consuming very little power per charge, with charging costs ranging from a few cents to under one dollar—we’ll estimate it at 40 cents for now.

How would this fee be settled under the existing traditional financial system?

The most straightforward solution is to link a payment card to the robot to enable contactless payments. Credit card payments involve two fees: a percentage of the transaction amount, plus a fixed fee of 10–30 cents, which remains unchanged regardless of whether the transaction is $0.40 or $400—the processing cost itself is constant. For a $0.40 order, a 30-cent fixed fee accounts for as much as 75% of the total.

As the business scales, problems will become more severe. Fixed fees are charged per transaction. If, in the future, machines process $1 billion in small-service transactions annually, with an average payment amount of about 32 cents per transaction, the total number of transactions would reach 3 billion. With 3 billion transactions at a fee of 30 cents per transaction, the fee costs alone would amount to $900 million out of $1 billion in revenue.

Bank wire transfers are more expensive. A cross-border SWIFT transfer incurs a fee of $15–$50, with each intermediary clearing bank adding an additional $10–$30, plus currency conversion costs. This payment method is only suitable for transfers above $5,000.

The current fiat payment system was designed for a small number of high-value transactions. In contrast, payments autonomously executed by machines are fundamentally different. If high fees are the first barrier, the second core issue is that machines themselves do not possess independent customer identities.

Even if the fee issue is resolved, the charging station still cannot directly deduct funds from the robot’s account. Payment accounts are owned by individuals or businesses, which come with comprehensive dispute resolution mechanisms. This payment is still fundamentally your responsibility—the robot merely triggers the card swipe action. This is the current implementation method for all robot payments worldwide.

Within just two seconds of authorizing an unknown device to draw power, the charging station must verify multiple pieces of information: Is this device a genuine physical unit, or a scripted fake request designed to steal free electricity? Has this device demonstrated a reliable operational history, or is it likely to abandon charging mid-session? If a device malfunction causes losses, is there verifiable accountability? Granting the robot an independent identity addresses these three verification questions without requiring a signed paper contract.

Currently, these issues are resolved manually. They will invoice you monthly, and if the robot causes damage, they will come after you and sue you. This system might work if it were just you and the robot. But in a world with two thousand operators and forty thousand machines interacting on the streets, the system needs records so strangers can quickly check and decide whether to service this machine.

This is the core concept of the entire solution.

Before diving deeper, let’s first clarify a key fact: the vast majority of bots will never need this autonomous payment system.

A robot requires a separate asset account only if it meets all four of the following conditions:

• Devices belong to different owners.
• The order volume is relatively small, and the trading scenarios are random and fragmented.
• There is no centralized platform to facilitate settlement between the two parties.
• Service authorization must be completed on-site within seconds.

Amazon has deployed over a million proprietary robots across more than 300 warehouses, all controlled by a proprietary scheduling system, thus failing to meet the first constraint; platforms like Uber Eats fail to meet the third constraint; and devices signed into prior agreements with merchants fail to meet the second and fourth constraints. This means that this segment represents a niche market, not a universal requirement across the entire robotics industry.

Approximately 1,000 Optimus humanoid robots have been deployed inside Tesla's factories; during the earnings call, the company stated that the devices are still in the learning and data collection phase and are not yet available for commercial use.

FANUC, a leading global manufacturer of industrial robots, sells hardware products along with the FIELD equipment monitoring platform for fault prediction; this system does not involve any fund transfers.

Chinese company Unitree has reduced the procurement cost of humanoid robots to an affordable range for the general public. The company completed its IPO in Shanghai in August 2026, raising approximately $619 million, with over 5,500 humanoid robots shipped in the prior year. According to the company, buyers are using the robots in a wide variety of real-world applications. In May 2026, Unitree launched UniStore, an app store designed for humanoid robots.

The above industry giants are replicating Apple’s business model, keeping coordination, collaboration, and payment settlement entirely within their own proprietary systems, with profits retained on the platform.

Open protocol ecosystems can only thrive in the gaps between major closed systems, such as cross-brand collaboration among robots from different providers or public charging services. The DePIN sector market, as defined by CoinGecko, is significantly smaller than the overall robotics industry; all subsequent analysis must be viewed with this premise in mind.

For a robot to complete autonomous payments, it must first determine its precise location. Standard GPS positioning has an error margin of several meters. Starship’s CEO has publicly stated that standard GPS accuracy is insufficient for business needs, and the company’s robots require navigation precision at the inch level.

To reduce positioning errors from meters to inches, signal corrections are required. Ground reference stations obtain their precise coordinates, calculate positioning deviations, and broadcast correction data to nearby devices—this technology is called Real-Time Kinematic GPS (RTK-GPS). The effective coverage radius of correction signals is approximately 30 kilometers, requiring a dense network of reference stations.

GEODNET incentivizes users to install positioning base stations on their rooftops through a token reward model. Over 21,000 devices have been deployed globally, covering more than 150 countries, with an annual recurring revenue of approximately $11 million. Multicoin Capital led an $8 million token acquisition round.

Project revenue comes from providing centimeter-level RTK positioning services to independent devices such as delivery robots, drones, and agricultural machinery. At the consumer level, this business model does not heavily rely on cryptographic tokens—traditional subscription-based payment models can also be implemented. However, at the infrastructure development level, GEODNET has demonstrated that token incentives can rapidly establish a global physical network; building such a network through traditional corporate self-investment would cost billions of dollars and take decades.

GEODNET Token Supply

Currently, robots produced by different manufacturers run on separate software systems and cannot communicate with each other. Enterprises can only select a single brand when purchasing hardware. Deploying a mixed environment of robots from multiple brands requires custom-developing an integration program that does not currently exist.

OpenMind has secured a $20 million funding round led by Pantera, with founder Jan Liphardt, a professor at Stanford University. The team is developing OM1, an open-source general-purpose operating system, and FABRIC, a coordination layer. Just as Android enables the same apps to run across different smartphones, OpenMind is building a unified software foundation for all robots.

Developers write upper-layer business logic that enables the program to simultaneously support Unitree humanoid robots, quadruped robots, and wheeled delivery devices. The ultimate goal is to achieve cross-brand device recognition, collaborative operations, and automated payment settlement using this universal interaction language.

The project builds a machine identity system based on the ERC-7777 standard, defining the behavioral boundaries of devices. If a robot is designated as an "auxiliary service" device, the program automatically rejects tasks that violate its designated role. Robots can also cross-verify sensor data with each other to prevent safety incidents caused by anomalies in a single device's perception.

OpenMind has partnered with Circle to enable gas-free USDC micropayments based on the x402 standard, addressing the pain point of disproportionately high fixed fees for small transactions. In an official demo video, a robot successfully autonomously paid its electricity bill. The demonstration ran on a test network and currently has no real on-chain transactions. However, this experiment proves that physical devices can independently host wallets, identify purchasable physical resources, and fully automate the entire transaction process without requiring trust between parties.

Since 2017, IoTeX has been deeply focused on blockchain solutions for physical devices, launching two core products to address identity issues.

The hardware identity ioID embeds a cryptographic fingerprint into physical devices, enabling them to independently sign and attest to operational actions; the real-world proof-of-work W3bstream transforms physical-world execution behaviors into verifiable digital credentials on the blockchain.

While IoTeX addresses identity and attestation, it overlooks credit and financing systems—this is where peaq comes in.

If robots need to make payments, verify credentials, and maintain transaction records with unknown third parties, they require a trust and settlement infrastructure equivalent to that of human society—similar to business registration systems or the SWIFT cross-border payment network.

peaq has built a comprehensive ecosystem, primarily divided into four core modules. peaqID serves as the registration credential. Machine NFTs act as ownership records and can be fractionalized using the ERC-3643 standard. ERC-3643 is a token standard that permits transfers only between approved holders. In August 2026, peaq introduced support for P256 chip signatures, moving the verification process to a hardware security chip.

The peaq Machine Credit Score system assigns a rating from 0–100 based on machine revenue data, activity levels, and履约 reliability, using a Moody’s-like rating scale ranging from AAA to unrated.

A robot pays a 40-cent charging fee, and the charging station provider can specify the payment network as Solana or Ethereum. peaq’s core value lies in enabling robots to connect to any settlement channel specified by the provider.

In May 2026, during a demonstration of peaq’s Serve delivery robot making autonomous payments, the funds were ultimately settled on the Solana blockchain, not on peaq’s own blockchain.

Throughout 2026, peaq will primarily act as an ecosystem integrator. From January to the end of August, the project completed 49 development milestones and implemented 20 ecosystem integrations: connecting with GEODNET location services, NAVER map navigation, and World ID for human-machine identity separation; integrating compute resources from Akash, Acurast, and Arcium; and incorporating Unitree humanoid robots and LG CLOi commercial service robots as hardware endpoints.

Enterprise registration is endorsed by national regulatory authorities, courts have the authority to verify it, and banks use SWIFT because the entire industry has standardized transaction formats. For peaq to break free from traditional compliance systems, it must proactively engage with Dubai’s regulatory authorities and apply for an official license. Until it obtains regulatory compliance certification, financial institutions will not recognize this machine-based credit scoring system as legally valid.

peaq, in collaboration with CoinList, is launching Initial Machine Offerings (IMOs), allowing users to purchase shares of machine-generated revenue, with asset structures designed by DualMintRWA. The flagship project is a tokenized vertical farm in Hong Kong, with 80% of operations automated. Twenty tokenized claw machines will follow. To date, the farm has distributed approximately $3,600 in earnings to token holders.

Even if robots have positioning systems, operating systems, identity credentials, and credit ratings, the hardware itself still requires funding for procurement. One current market possibility is that the buyers could be AI agents. Virtuals has already connected approximately 17,000 on-chain AI agents to Solana’s BitRobot network, where AI agents pay to hire physical robots to perform offline tasks, with funds held in smart contracts and automatically settled upon task confirmation.

Software agents paying robots and robots transacting with each other represent two distinct models, with the former demonstrating stronger commercial sustainability. AI agents possess digital capital, business objectives, and computational power, yet cannot interact with the physical world; physical robots have mobile hardware capabilities but lack native funding or business needs. When robots pay each other, the devices are typically owned by the same company, making internal bookkeeping far simpler and more efficient than actual on-chain transactions.

Across the entire market, the total market capitalization of robot sector tokens on CoinGecko is approximately $100 million for GEODNET and $55 million for peaq.

It’s clear that this sector remains a niche within a niche.

Industry statistics show that, according to the International Federation of Robotics' 2025 World Robotics Report, 542,000 new industrial robots were installed globally in 2024, with approximately 4.66 million units in operation, over 2 million of which were deployed within China.

In an external research report released in July 2026, J.P. Morgan forecasted that global robotics market sales will reach approximately $100 billion in 2025; under the base case scenario, annual market sales are expected to reach $2.5 trillion by 2035, with a pessimistic estimate of $500 billion and an optimistic estimate of $8 trillion. The same report estimated that the humanoid robotics market size will grow from $2 billion in 2025 to $300 billion by 2035 under the base case scenario.

This is the target market that the cryptocurrency industry aims to enter and build payment and identity infrastructure for. The crypto industry has never lacked grand visions, and this time is no different.

The concept of machine-to-machine interaction is not new; similar ideas were proposed in the early days of the Internet of Things (IoT) sector. As early as 2015, IBM and Samsung demonstrated a washing machine capable of autonomously purchasing laundry detergent via Ethereum, under the project name ADEPT. Months later, IBM invested $3 billion into its IoT business. However, the IoT sector eventually shifted its focus toward data monitoring. Billions of devices simply uploaded operational data to manufacturer backends. Although the entire technology was fully implemented and devices obtained communication certificates from manufacturer servers, automated machine-to-machine transactions never achieved market scale, and the traditional payment system remained unchanged.

Robots differ fundamentally from ordinary IoT devices. A temperature control device, costing only $200 to purchase, can only operate in a fixed manner. In contrast, robots have a high acquisition cost but possess the ability to generate revenue. Once a device can generate income, credit and insurance become essential, and third parties must verify its ability to fulfill payment obligations. IOTA has long aimed to build a public blockchain for the machine economy, but its business focus has since shifted; its implementation cases are primarily concentrated in government record-keeping scenarios such as Kenya’s customs documents, UK port trade credentials, and Argentina’s organ donation registries. Government agencies purchase blockchain identity tools for authentic data recording, yet automated machine-to-machine payment services have yet to gain market traction.

The growth of the robotics industry itself does not rely on cryptographic technology. The value of crypto assets lies in filling the gap in cross-party collaboration: providing devices with publicly verifiable identities, self-managed crypto wallets, and low-cost micropayment channels.

Let’s envision an open robotic labor market where anyone can rent a robot they don’t own to complete tasks. This market requires comprehensive infrastructure: verifiable device identities, low-cost micropayment solutions, and履约 credit profiles to avoid renting devices with battery or performance issues; meanwhile, it offers value-added rental services such as high-precision positioning, cloud computing power, remote human operation, and idle charging stations—all of which DePIN is uniquely suited to support.

Of course, if Tesla, Amazon, and leading domestic hardware manufacturers continue to keep their entire operations closed within their own ecosystems, an open market may never truly emerge. Once an open market is established, new business challenges will follow: Who provides financing for robots that can generate revenue on-chain? Who insures the devices? How can robot fleets be used as collateral? Who builds the matching platform to receive orders from warehouses and accept bids from robots? Only after a robust identity system is in place can asset securitization become the next priority. Tokenization only unlocks true value when assets can flow freely.

DePIN is merely a lightweight underlying channel serving niche scenarios in the robotics industry that are not closed-loop. Industry giants have no obligation to direct business to this ecosystem.