BlackRock, in its report *The Machine-Native Economy*, officially discussed the role stablecoins might play when AI agents purchase data, call APIs, and pay for computing power. My first reaction was: if AI can use USDC to buy services, then how many of those AI tokens we’ve been hyping are actually necessary? During the last AI boom, from AI memes like GOAT to chatty, tweet-posting agents, most of what people traded was attention and future imagination. But now, if AI really starts spending money on behalf of humans, the questions become more concrete: What is it buying? From whom? And why would it choose the same provider next time? For example, say you give a research assistant a $5 budget to analyze a project. It needs accurate data, suitable models, and affordable, stable services. Whether the supplier has issued a token might not necessarily help it complete the report. BlackRock mentioned x402, which offers a pay-per-use model: the agent requests a service, receives a payment request, completes the payment, and gets the result. Developers can directly sell useful APIs without needing to issue a token just to collect payments. For developers, this removes a barrier; for token holders, it raises a question: can this business still operate smoothly without my token? Of course, not all old AI projects will be eliminated because of this. Those that can provide data, computing power, or complete tasks still have a chance to secure orders. Tokens might still play roles like staking. But whether a service has users and whether the token benefits from it are two separate things. Moreover, while it seems like AI is choosing services, the real decision-making power largely lies with platforms and developers. Which tools are integrated by default, which wallets are allowed, and which suppliers are prioritized are often pre-determined. AI makes choices within these rules and the user’s budget, meaning the platforms controlling the entry points can influence the flow of orders. Of course, BlackRock’s report reflects forward-looking predictions and its own business stance, but the questions it raises are worth retail investors seriously considering: After AI-driven economic growth, who’s actually paying? If orders increase, what gives tokens the right to benefit from it?