AI: The Time for Accountability Has Come, and Stablecoins Charge by the Millisecond

Show me the money! The grace period is over. More than three-quarters of American entrepreneurs have integrated artificial intelligence into their operations, 10% have deployed it end-to-end, yet over a third have reduced their technology spending in the past twelve months. These figures come from the latest EY barometer, conducted among 500 executives of established companies in the United States with at least $5 million in annual revenue. Their position, summarized by Forbes columnist Joe McKendrick, can be encapsulated in a phrase that Jerry Maguire would have appreciated: "show us the money".

In accounting terms, every dollar invested in AI must reappear somewhere, in revenue, margin, or retained customers. This climate of distrust also affects payments between AI agents, which are currently being reinvented.

Key Points

• According to the EY barometer conducted among 500 American entrepreneurs, over 75% have integrated AI and 10% have fully deployed it, but more than a third have reduced their technology spending due to a lack of measurable returns.
• In Europe, AI adoption reaches 76% among entrepreneurs surveyed by EY, with 60% reporting efficiency gains but only 38% showing a visible impact on revenue.
• Coinbase's x402 protocol totaled around 165 million transactions and $50 million in cumulative volume by the end of April, with an average ticket of $0.31, a level unattainable by banking rails.
• Google (AP2), Mastercard (AP4M), and Visa have all integrated stablecoin settlement for machine-to-machine payments, while 60% of stablecoin volume already comes from business transactions.

Growth remains the number one priority for 46% of respondents, ahead of financial stability (33%) and profitability (21%). However, the trade-off is sharp. 72% consider their business moderately or highly exposed to geopolitical disruptions, and budgets are focused on use cases capable of producing quantifiable results.

When asked about the areas where they expect the most value from AI, entrepreneurs already invested cite revenue growth through sales and marketing (57%), customer experience, and then operational efficiency. The blockage is not ideological. It is methodological.

"Many organizations still struggle to consistently determine the impact of AI."

Authors of the EY Entrepreneur Ecosystem Barometer

The cited barriers are data preparation, integration complexity, talent shortages, and security. On this last point, 99% of respondents claim to have formalized a risk management approach related to data and AI, and 88% anticipate an increase in their workforce over the next twelve months.

The same gap between usage and yield appears in Europe: in the second edition of the EY European barometer, covering 1,009 entrepreneurs, AI adoption reaches 76% compared to 61% a year earlier, with 60% reporting efficiency gains and 51% cost optimization, but only 38% showing measurable effects on revenue.

The counter before the concept: crypto rails charge AI per unit

This gap has a mundane cause. Most AI tools are sold on monthly subscriptions and per-user licenses, a mesh far too coarse to tie a precise expense to a specific revenue. Payments in stablecoins tackle the problem from the other end, charging for each call.

The x402 protocol, supported by Coinbase, reactivates the HTTP 402 "Payment Required" code, which has been dormant since the early days of the web: a software agent settles its request in seconds, in stablecoin (a crypto token pegged to a currency like the dollar), without a card or invoice.

At the end of April, it totaled approximately 165 million transactions and 50 million dollars in cumulative volume for 69,000 active agents, with an average ticket of 0.31 dollars. CoinGecko, for example, charges 0.01 dollars per request for its access point. No banking rail holds at this level, as the fixed part of interchange fees far exceeds the amount paid.

Major payment networks have taken their seats at the table. Google has integrated x402 into its AP2 (Agentic Payments Protocol), developed with over 60 partners, as the first facilitator in stablecoins.

Mastercard launched its AP4M (Agent Pay for Machines) in June with about thirty partners including Coinbase and Cloudflare, for micropayments dropping to fractions of a cent. A joint report from Visa and Artemis published in mid-July positions stablecoins as the preferred rail for micro-commerce between machines, while cards continue to handle traditional purchase amounts.

Corporate treasuries have already made their decisions.

The shift was underway before the arrival of autonomous agents. About 60% of the stablecoin volume comes from business transactions, and Visa's stablecoin settlement program operates at an annualized pace of 7 billion dollars.

According to data reported by Paybis, B2B accounted for 35.7% of stablecoin volume in 2023, compared to 97.8% in the first four months of 2026. The calculation is clear for a manager paying foreign providers: a few dozen dollars in fees instead of several thousand, and a settlement in minutes rather than days.

Misconceptions persist nonetheless. Nearly half of the companies surveyed by Paybis believe that a stablecoin transfer takes hours, a third imagines it at 3% fees when the actual cost remains below 1% end-to-end. Only 22.5% report using or planning to use stablecoins for their international payments. There is room for improvement.

The agentic takeoff remains bumpy. About half of the x402 volume resembles tests rather than actual commerce, daily transactions fell from 731,000 in December 2025 to 57,000 in February, and no legal framework clearly designates the responsible party in case of a payment error made by an agent.

For an entrepreneur seeking proof, the granularity nonetheless changes the nature of the discussion: a request billed at 0.01 dollars, timestamped and verifiable on a blockchain, directly compares to the revenue it has generated. A monthly flat fee, never. This is the crux of this emerging agentic economy: transforming an opaque AI expense into a traceable accounting line, even if it first requires addressing the issue of legal liability.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.