Over the past decade, AI has been defined as a form of "intelligent productivity"—helping humans make decisions, optimize processes, and save costs. Now, a more fundamental question looms: when AI moves beyond mere thinking and expression and becomes able to autonomously "spend money," how will the world's economic order be rewritten?
The emergence of AI payments marks the first time machines have participated in value exchange. They have evolved from cognitive systems into economic entities. They require identity, accounts, credit, contracts, settlement, and even accountability. When discussing the next steps in AI, the focus is no longer on algorithms, computing power, or large-scale models, but rather a more fundamental question: when AI truly participates in economic activities, how will it "pay"? Who will grant it accounts, credit, settlement, and accountability?
The redesign of the payment system will determine the boundaries of the intelligent economy. We have found that three distinct paths have emerged around the question of "how to pay for AI."
One is an institutional path centered on identity and regulation , attempting to allow AI to operate legally within the existing financial system; one is an encryption-native path based on blockchain , allowing AI to become an autonomous economic node; and another is a vertical integration path centered on the corporate ecosystem , which allows AI payment to be implemented first through productization and packaging.
These three paths are like three civilizations: independent yet interactive. Rather than one eroding the other, they explore the same question under different logics of trust: how should trust be rebuilt in a world where machines can also fulfill contracts?
Google: Let AI “spend money on behalf of people” in the system
Google's approach most closely resembles imperial thinking. Its AP2 (Authorized Payment Protocol) is an "AI-authorized payment protocol" that attempts to embed machine economic behavior into the existing financial system . In other words, it's not about giving AI accounts, but about enabling AI to execute payments on behalf of humans within its authorized scope.
Its core mechanism includes three parts: identity authentication, authorization credentials, and payment execution.
Users first define their AI's permissions within the system—for example, they can pay subscription fees, place ads, and settle computing power bills on their behalf, but they cannot transfer funds or invest. Whenever the AI initiates a payment, the system generates a short-term cryptographic authorization (similar to a one-time digital signature). Google's network verifies the identity and executes the transaction. Once the payment is completed, the authorization automatically expires.
Under this mechanism, AI is merely an executor, not the principal of the account. The money it spends still comes from human wallets, and settlements still go through payment channels like banks, card schemes, or Google Pay. AI is merely an "agent," and its authority and boundaries are defined by both users and regulators.
The problem Google wants to solve is how AI can safely help people spend money without exceeding its authority or violating the law. It's not about freedom, but about controllability and compliance . This is "imperial trust": trust comes from identity and regulation, not algorithms.
Coinbase: Let AI “Spend Its Own Money”
Coinbase's approach is the opposite. Rather than having AI represent others, it allows AI itself to become an economic entity .
In this crypto-native system, each AI can generate its own crypto wallet (equivalent to a bank account), and its behavioral rules are defined by smart contracts.
For example, you could pre-inject 1 ETH into the AI wallet and write rules: "Each task expenditure must not exceed 0.05 ETH; payments must be recorded by an on-chain contract; if the balance falls below 0.1 ETH, automatically request additional funds." From that moment on, the AI becomes an independent "economy." It can autonomously transact with other intelligent agents, pay API fees, purchase computing power, and distribute rewards— all without any human authorization.
Coinbase's x402 protocol is designed for this type of autonomous economy, allowing direct communication and settlement between different AIs. Transactions are completed through the blockchain, with funds transferred directly from the AI wallet without the need for a bank or payment company intermediary.
The core logic of this system is that trust no longer comes from identity, but from code. There's no authorization from higher authorities, no human review—the contractual rules are law. It addresses another problem: how can AI independently participate in economic activities without human backing? This is precisely why it worries regulators the most: in such a world, who owns the assets? Who bears the risks? Who can regulate?
Stripe: Implementation is the most important
If Google wants AI to spend money within institutional constraints, and Coinbase wants AI to break free and spend its own money, Stripe represents a third approach: empowering AI to spend money first. It doesn't pursue grand institutional revolutions or the freedom of the crypto world, but rather returns to the fundamental principle of business: settlement .
Stripe's starting point is practical. In today's internet world, almost all economic activities rely on automated payments—advertising, subscription billing, API calls, cloud computing settlements—all of which cannot be accomplished through manual instructions. If AI is to truly enter the business cycle, it must possess the ability to conduct autonomous settlements. However, traditional payment systems don't support this, and encryption systems are not recognized by regulators. Therefore, Stripe chose a third path—building a usable bridge between the existing and future systems.
It has established an enterprise payment network called Tempo . Unlike Google, which relies on the authorization system of financial institutions, nor is it as completely decentralized as Coinbase, Tempo is hosted and audited by Stripe itself . While closed, it is compatible with blockchain-based smart contracts and can also connect to fiat currency accounts. Developers simply need to access Stripe's SDK, and AI can complete payment, settlement, tax filing, and other operations in the background. For example, it can automatically pay advertising budgets, purchase computing power, and settle data fees, all of which are executed by Stripe's system in the background.
In this model, AI neither owns the wallet nor signs authorizations; it hands over payment authority to the platform. Trust no longer comes from regulation or algorithms, but from the business itself—from Stripe's credit, compliance, and risk management. This is a classic example of "business trust": trust stems not from a perfect world but from someone willing to take responsibility.
The difference becomes even clearer when put into context. Consider an AI operating an advertising account: In Google's system, it must obtain authorization before making payments, and the amount and purpose of payments are within regulatory guidelines. In Coinbase's system, it maintains its own wallet and can settle advertising fees directly on-chain, with transactions public but irreversible. In Stripe's system, it doesn't care about wallets or signatures; it simply issues instructions—the Tempo network automatically handles all clearing, tax, and compliance reporting. AI payments become like calling a function: clean, fast, and visible.
The problem Stripe seeks to solve isn't "Can AI spend money?" but "How can AI spend money safely within the real financial system?" It chooses to leverage enterprise credit to escrow machine trust, encapsulating complex financial relationships in a productized manner. However, the cost is clear. Tempo is Stripe's private network, and all settlement paths are controlled by the enterprise. A single platform error could bring the entire system to a halt.
If Google represents the extension of the system and Coinbase represents its challenge, then Stripe represents its integration—replacing institutional conflict with commercial efficiency. Its revolution isn't about overthrowing the old world, but rather about enabling new technologies to be implemented within it. The competition among these three companies is not only about technology but also about institutions. Who will be responsible for the future AI world? Will it be regulation, code, or the platform itself?
This isn't an either-or decision. Payment systems never have a single winner. Whether it's VISA, SWIFT, PayPal, or the digital yuan, they've all coexisted for decades. The emergence of new payment systems doesn't eliminate the old ones; rather, they coexist and overlap with them until both the market and regulators accept them.
Legal challenges
In this transformation, the real difficulty lies not in technology but in law.
Can AI become an independent economic entity? Can it legally hold assets and sign contracts? If its payment behavior goes awry, who bears the responsibility? This is the fundamental problem facing all current AI payment solutions. In the on-chain world, smart contracts are irrevocable, which strengthens trust but also makes errors more fatal. Once AI misjudges and funds are mistakenly transferred, there is no "undo" button. In traditional legal systems, transaction accountability is the bottom line—every payment must have a clear responsible party. If AI's autonomous payments cannot be incorporated into the legal liability framework, its "freedom" will not be recognized by the system.
This is why I believe AI won't possess true economic personality in the near term. It can execute payments, but it can't bear the consequences. Every step of AI payments still requires human signatures, platform custody, and institutional endorsement. This is like the early stages of autonomous driving: technically, it's possible, but legally, it's not. The same is true for the future of AI payments—we must first design a complete "intelligent agent liability system," including authorization rules, loss compensation, risk insurance, and regulatory interfaces. Otherwise, the system will quickly collapse.
In the short term, AI-powered payments will first appear in existing payment systems as "intelligent authorization," such as the automatic settlement features in Google Pay, Apple Pay, or WeChat Pay. In the medium term, enterprise-level scenarios (SaaS calls, advertising settlement, API billing) will first form an automated payment ecosystem, with the Stripe model being the most likely to be commercialized. In the long term, a decentralized system like Coinbase, while imposing the greatest compliance pressure, is also the most likely to foster true institutional innovation. This is because it raises fundamental questions: who owns assets, who defines trust, and who bears responsibility.
Technology will ultimately force the law to find new answers. Perhaps future contract laws will include an "intelligent agent liability clause," or perhaps anti-money laundering regulations will include a chapter on "AI customer identification." The evolution of AI payments will ultimately force institutional evolution.
