Bitget App
Trade smarter
Buy cryptoMarketsTradeFuturesEarnAISquareMore
What would happen if the buyers of computing power become intelligent agents?

What would happen if the buyers of computing power become intelligent agents?

404k404k2026/09/24 00:19
Show original
BlackRock's new AI paper is essentially about inference. It presents the following framework: 1. AI is machine-native intelligence. 2. Digital assets are machine-native currencies. 3. Agents are the link connecting the two. One signal it cites is Stripe’s recent agreement to acquire Openrouter. Openrouter acts as a gateway, connecting over 80 providers including NBIS with more than 400 models, routing based on cost and compatibility. Stripe handles the payments. Combining the two, it’s easy to imagine that purchasing tokens could become part of the billing infrastructure.Capital expenditure remains in the news spotlight. Some of the estimates they reference suggest that, by 2030, investment in AI construction will surpass $5 trillion. They also raise another point: focusing on operating expenses.According to market consensus expectations, hyperscale cloud providers are projected to reach $1.1 trillion in revenue by 2030. Currently, training demand is concentrated among a few buyers, with long contract durations and human participants on both sides of the transaction.Inference is the real mass market. A McKinsey chart they reprint summarizes the core of the entire article: by 2030, inference is poised to account for the largest share of AI data center electricity consumption, estimated at about 43% according to their definition.This involves not only chatbots but also millions of buyers demanding a single unit of effective work: tokens, a task, reserved capacity, a certain region, a specific latency, or a particular price.This presents a power market logic worth paying attention to.When buyers are agents, ordinary bank card payments start to look unsuitable — a single transaction amount could be extremely small, for example.BlackRock points out that always-on payment channels, such as stablecoins, as well as x402-style “pay before API response” mechanisms, could serve as the adaptable underlying infrastructure.Once this mechanism is feasible, inference becomes more like a service that can be purchased and settled without human involvement.Machine-native intelligence needs machine-native currency. Is the implied third element machine-native inference?
0
0

Disclaimer: The content of this article solely reflects the author's opinion and does not represent the platform in any capacity. This article is not intended to serve as a reference for making investment decisions.

Understand the market, then trade.
Bitget offers one-stop trading for cryptocurrencies, stocks, and gold.
Trade now!

You may also like

The United States is planning to promote dollar stablecoins overseas in an effort to strengthen its position as a reserve currency.

According to sources, the Trump administration is considering an initiative to promote US dollar-pegged stablecoins abroad in order to strengthen the dollar’s position as a global reserve asset.

智通财经2026/09/24 00:11
The United States is planning to promote dollar stablecoins overseas in an effort to strengthen its position as a reserve currency.

Warning of AI credit risk, Moody's: The off-balance sheet commitments of the five major US giants have increased eightfold in three years, soaring to $2.8 trillion.

According to a report by Moody’s Ratings, the five companies Amazon, Microsoft, Google, Meta, and Oracle collectively bear about $2.8 trillion in off-balance-sheet obligations related to AI. Among these, lease commitments exceed $1 trillion, purchase commitments total $1.57 trillion, and guarantees amount to $137 billion. This figure represents explosive growth compared to $350 billion in 2023. Moody’s analysts noted that the pace of growth for these commitments is even more noteworthy.

华尔街见闻2026/09/23 23:26