Summary of Key Points
AI agents (which can be understood as AI programs capable of performing tasks independently) are learning to spend money on their own, but the payment systems designed for them are not yet fully developed. Recently, there has been a surge of activity in the industry: the Linux Foundation has reactivated the HTTP402 protocol, which has been dormant for 35 years and is now being used for internet payments; Stripe has introduced virtual cards dedicated to AI; the transaction volume via XRP, a blockchain platform, involving AI agents, has exceeded one million; and a company called Natural has raised $30 million to develop a native payment system for AI. However, this field still faces many challenges: existing payment systems are not suitable for AI, legal responsibilities are unclear, and trust in AI-driven payments from both businesses and consumers has not yet been established.
The AI Payment Sector Suddenly Heat Up: Recent Developments Indicate a Critical Point is Approaching
In recent weeks, there has been an unusually high level of activity in the AI payment sector, as if everyone is trying to seize this opportunity:
- Reactivation of HTTP402 Protocol: This payment interface, reserved since 1991 and unused for 35 years, has now been activated by the Linux Foundation with the support of 40 major companies including Visa, Mastercard, and AWS, specifically for handling autonomous payments between AI systems.
- Stripe Launches AI Virtual Cards: Stripe has collaborated with banks to issue “one-time use virtual cards” for AI, with each transaction using a new card that is discarded after use, without the need to access users’ real information. Over 1.6 million transactions have already been settled this way.
- XRP Transaction Volume Exceeds One Million: The number of transactions involving AI agents on the XRP blockchain platform has surpassed one million, indicating that AI-based transactions are becoming significant in scale.
- Natural Raises $30 Million: This company, founded just a year ago, has raised a total of $40 million in funding, with Forerunner, a venture that previously invested in consumer brands, as the investor. This suggests that capital is beginning to bet on the infrastructure for AI payments.
These developments collectively indicate that the demand for AI payments has moved from concept to practical action, and the industry is about to enter a period of rapid growth.
Natural: Aiming to Be the “Stripe” of the AI World?
Natural’s goal is not to create AI itself but to help AI make legal and secure financial transactions within the system, essentially acting as an “AI payment manager.”
- Team Background: The three founders have experience in consumer finance, such as managing joint accounts for couples, and understand the complex issues surrounding compliance and payment processing, which are crucial for AI payments. They have also recruited key employees from Stripe and Square.
- Product Features: Six features have already been launched:
- Wallets: AI wallets with FDIC insurance in the United States, ensuring that funds are safe even if the company goes bankrupt.
- Vaults: One-way accounts that allow only deposits but not withdrawals to isolate risks (e.g., setting a budget for AI purchases to prevent unnecessary spending).
- Pay/Request/Transfer: Functions for making payments, receiving payments, and transferring funds.
- Connect: Allows platforms to integrate with this system.
In the future, Natural plans to introduce features such as AI-powered phone payments, debit card issuance, and credit limits. Their ambition is to become a “payment giant” in the AI sector, rather than just a small addition to existing services like Stripe.
- Current Status: The company is not yet generating revenue (pre-revenue) and is valued at $150 million based on expectations for the size of the AI economy. With a team of 17 people, they face significant pressure balancing expenses with product development progress.
Why Existing Payment Systems Are Not Suitable for AI?
Traditional payment systems are designed for humans, making them awkward to use for AI:
- Lack of Identity: AI cannot open bank accounts (banks require KYC, or identity verification), and they lack credit records. Even card transactions require the signature of the cardholder, which AI cannot perform.
- Manual vs. Automated Processes: Traditional payments require human confirmation (e.g., entering passwords for online transfers or approval for business purchases), while AI’s strength lies in automation—from finding suppliers to making comparisons to completing payments. Existing systems can only provide temporary solutions (like Stripe’s one-time virtual cards) that are not optimized for AI and are prone to errors.
- Mismatch in Transaction Speed: AI can initiate dozens of transactions in milliseconds, but traditional payment processes take days to settle, and risk control models are based on human behavior (e.g., large transactions in New York may trigger alerts when you usually buy coffee in Shanghai). These models are ineffective for AI, which operates at a much faster pace.
AI Payments Will First Be Adopted by Businesses, Not Consumers
AI payments will not start with consumers (e.g., AI helping with online shopping) but will first become popular among businesses:
- Practical Business Scenarios: For example, freight companies using AI to compare prices and find suppliers, automatic renewal of SaaS subscriptions, and automated payments in supply chains. These scenarios do not require consumer trust, and businesses can control the risks themselves, making the technical implementation easier.
- Consumer Adoption is Further Away: For consumers to let AI make purchases on their behalf, issues such as accountability (who is responsible if the AI makes a mistake) need to be resolved. McKinsey predicts that mainstream consumer applications for AI payments may not appear until 2027-2028.
- Industry Consensus: Products like Stripe’s virtual cards and Natural’s solutions are initially targeting business customers, as their needs are more urgent, and they are more willing to invest in the necessary infrastructure.
Overcoming Legal and Risk Control Challenges
Two major obstacles prevent AI payments from becoming fully mature:
- Legal Uncertainty: Current laws are based on human decision-making. If an AI makes a payment error (e.g., sending the wrong amount or purchasing counterfeit goods), who is responsible? Users, the AI company, or the payment platform? U.S. law does not provide clear answers, which discourages businesses from using AI payments.
- Risk Control Needs Redesign: Traditional risk control mechanisms are based on human behavior patterns, but AI operates at machine speed and unpredictably. Security frameworks must be re-designed. For example, Stripe’s one-time virtual cards impose restrictions (e.g., each card can only be used for specific purchases with a limit on the amount), which, although conservative, provides peace of mind for businesses.
Companies like Natural may need to offer “semi-autonomous” payment solutions where AI can handle tasks, but critical decisions still require human confirmation rather than promoting the concept of completely autonomous payments.
Conclusion: The Trend Has Begun, but the Foundation Is Yet to Be Laid
The AI agency economy is bound to grow (McKinsey predicts a market worth $3-5 trillion by 2030), and payment infrastructure is at its core. Industry players are making bets, but the legal framework, risk control models, and business models are not yet fully developed. Those who can solve these issues first will become the “Alipay” or “Stripe” of the AI payment sector. However, the answer is still unknown as of 2026. The market window has opened, but the foundation needs to be carefully laid out.