North America’s Crypto Economy: AI Payments, Dollar Stablecoins and Tokenized Treasuries

AI agents are starting to pay for data, software tools and other services on-chain, and North America supplied much of the infrastructure behind that activity during the 12 months ending July 1, 2026. More than 138 million x402 transactions had taken place on Base by the end of the period. The figures suggest that machine payments are shifting from isolated trials toward repeat use, though the market remains early and its activity is concentrated in a handful of service categories.

That shift sits inside a broader regional story. The United States and Canada received $3.02 trillion in on-chain value during the reporting period, down 1.9% from the prior period. The U.S. accounted for $2.78 trillion, a 2.2% decline amid a sharp market downturn. Canada moved in the opposite direction: its $247.7 billion crypto economy grew 2.3%. Stablecoins, tokenized assets and AI payment infrastructure all helped define the region’s activity, even as their users and effects increasingly crossed borders.

Small payments give agents room to transact

x402 is a payment system built by Coinbase for software that can make purchases without a person approving each transaction. Its relevance comes partly from transaction size. Credit card networks are poorly suited to processing charges worth fractions of a cent, while stablecoin transfers can settle for less than a cent. That makes it possible for an agent to pay a small fee each time it retrieves information or calls a service, instead of relying on a larger subscription or bundled purchase.

Circle described a division of use: card payments can cover purchases initiated by consumers through agents, while stablecoin transfers can handle machine-to-machine payments too small for card processing economics. In the 16 months ending September 2026, more than 87% of recorded x402 payments fell between one hundredth of a cent and ten cents. The most common payments were between one and ten cents.

Those amounts suit services that deliver value in small units, such as a single data request through an application programming interface (API). They also change the economics of software that repeatedly calls other software. A developer can set an agent to buy the information it needs as it works, while each service receives a small payment for its contribution. That arrangement could support pay-as-you-go access, but it also leaves developers to manage funds, permissions and spending limits across many automated transactions.

The network’s activity is still small beside conventional payments, and transaction counts alone don’t show whether each purchase produced useful work. Yet wallet behavior offers signs that x402 users are becoming more deliberate. The researchers counted wallets that spent on at least two separate days, excluding accounts that tried the system once and stopped.

Among those wallets, the share funded for a second time within 30 days rose from 70.6% for wallets first funded in the third quarter of 2025 to 89.3% for those first funded in the second quarter of 2026. The newest group is the latest one for which a complete 30-day observation window was available. A second funding suggests the wallet’s initial balance was used and the operator chose to replenish it, though it doesn’t reveal what the agent accomplished with the money.

The delay between first funding and first x402 payment also shortened. Wallets created in late 2025 were a median of 73 days old when they made their first payment. By the third quarter of 2026, the median interval had fallen to 18 minutes. Over the same period, the share of wallets making a first payment within an hour of funding rose from 4% to 67%.

Taken together, those measures point to wallets funded for a specific task and used soon afterward, rather than funded long in advance and left idle. That’s a useful indicator of repeatable use, but it is not proof of durable demand across the wider market. The data covers a new payment system, and the observed wallets represent a subset of users who made multiple payments.

Agents are buying information and other software services

Early x402 activity included one-off experiments and payments connected to memecoins such as PING. The reported activity has since spread into services with a clearer role in automated work. Millions of transactions went to code review, financial intelligence, data and search APIs. These services offer information in formats an agent can process directly, which makes them a natural match for small, on-demand payments.

Other observed purchases included gaming, token issuance, gambling services, payment tools and software for building agents. Model inference—the computation that produces an AI model’s response—was one of x402’s largest transaction categories. In that sense, agents are buying from other agents: a system performing a task can pay an outside model for a response, then use that response in a larger workflow.

Coinbase Ventures head Hoolie Tejwani said that agents are becoming economic actors, with each one functioning as a new blockchain user. He also pointed to the role of open standards and internet-native payments in systems built on x402. The claim captures the potential scale of machine activity, but it also raises a practical question: if a service accepts payments automatically, how does the person who configured the agent set and enforce its boundaries?

The reported data suggests that oversight may need to focus on the arrangements around a payment. A wallet’s funding source, its spending authorization, the service it pays and its transaction pattern can each provide information about the person or organization that deployed the agent. Individual payments may be too small and frequent for manual review. Automated controls could instead set budgets and permitted destinations, then flag behavior that departs from an agent’s usual pattern.

Responsibility still rests with the people and organizations that fund and configure the software. The agent itself doesn’t explain who approved its budget or why it paid a particular service. Funding records, known recipients and patterns of activity could help trace that relationship, but the source provides no evidence that these controls are consistently in place across x402. For developers, the challenge is to let an agent transact quickly while preserving a clear record of its permissions and spending.

U.S. wallets lead several x402 service categories

Among the agent-to-service transactions that could be assigned a location, U.S.-funded wallets accounted for 30%. The country’s share was higher in several service categories tied to building and running agents: 88.1% of geolocated model inference payments, 41.8% of payments to agent frameworks and tool catalogs, and 23.2% of payments to data and search APIs.

Wallets attributed to China showed a different distribution. They accounted for nine in ten tracked payments to security and code-analysis services, with nearly all of that category’s activity going to AgentLISA, a service that uses a language model to review Solidity smart contracts. Chinese-linked wallets also represented 46.7% of payments for market and financial intelligence and 35% of payments to payment and settlement services. Their share of model inference payments was 0.1%.

The report suggests that differences in model access may help explain the gap. Chinese-developed open-weight models, including DeepSeek and Qwen, can be used without the metered charges associated with U.S. services such as ChatGPT and Claude. Local preferences for homegrown models may also shape which services agents use. The available payment data shows where spending occurred; it doesn’t establish why a user selected one model or market over another.

BlackRock digital assets research head Will Su said the firm is watching agent payments because automated systems could create demand for frequent, low-value transactions at any hour. That interest reflects the scale these systems might reach if they become part of ordinary software workflows. For now, the evidence is more specific: U.S. wallets dominate several x402 categories, while other regional patterns vary by service.

These figures also depend on location estimates. The data covers transactions that could be geolocated, so it should not be read as a complete map of all agent activity. Still, the category-level differences show that machine commerce is already tied to a supply chain of models, tools, data and security services, with distinct regional centers of demand.

Dollar stablecoins and bitcoin play different roles

The wider U.S. crypto market shows a split between digital dollars moving outward and bitcoin activity becoming more concentrated in the country. The U.S. share of global bitcoin activity rose through each reporting period following the January 2024 launch of spot bitcoin exchange-traded funds. Stablecoins, meanwhile, formed a smaller share of U.S. crypto activity than they did elsewhere, and that gap widened during the 2026 period.

The pattern reflects different needs. People outside the United States may use a dollar stablecoin to hold or transfer dollar value without an American bank account. U.S. residents already have access to domestic dollars, while some institutions and investors use bitcoin as a scarce asset. Su said interest in bitcoin as a monetary alternative remained strong amid rising public debt and spending around the world.

During the 2026 period, the United States became a net importer of bitcoin through cross-border flows for the first time since 2022. Transfers below $10,000 remained net outbound in every measured period. Larger transfers ran the other way: the bracket above $10 million alone accounted for 53,200 bitcoin arriving in the country. The figures show different patterns by transfer size, but they do not identify every sender’s motive.

Dollar stablecoins also move through channels that don’t appear in conventional payments statistics. The report says 99.1% of stablecoin order-book volume was tied to dollar-pegged tokens. Each token represents a claim on reserves held largely in short-dated U.S. debt. Circle reported $73.3 billion of USDC in circulation at quarter-end, 19% higher year over year, alongside $14.8 trillion in on-chain volume for the quarter.

Circle described stablecoins as a routing layer between traditional currencies, local stablecoins and global dollar liquidity. That use matters because moving a position between assets, locations or trading venues can be a separate function from holding the position itself. It also means that digital dollars may be used by businesses and institutions as plumbing between different forms of money.

Bitcoin accumulation and stablecoin use have different sources of demand. Institutional bitcoin allocations can shift with market conditions. Su said exchange-traded products saw outflows in the first half of 2026 as capital moved toward AI-related investments, then resumed inflows in the third quarter as that investment theme cooled. Stablecoin adoption may build more gradually as users incorporate digital dollars into payments and transfers.

Dollar dominance remains difficult to displace in trading activity. Euro-backed stablecoins accounted for 0.45% of order-book volume, despite changes intended to make it easier for issuers to operate in Europe. Singapore and Hong Kong have also set up frameworks for locally denominated stablecoins. The figures suggest that creating an alternative is only part of the task: users need a reason to hold and spend it.

Tokenized assets add another channel for U.S. debt

The tokenized real-world asset market reached $36.07 billion at the end of the second quarter of 2026, growing 135% period over period. U.S. Treasury debt was its largest category, with $15.42 billion on-chain, or 42.8% of tokenized asset value. That amount grew 125%, from $6.86 billion in the previous period.

Some smaller categories grew faster in percentage terms. Corporate credit expanded 3,156% and private equity 1,340%, taking each from a base of tens of millions of dollars to a few billion. Treasury debt’s growth added more value in dollar terms than the next three categories combined. The market is broadening, but its largest component remains U.S. debt.

Tokenization represents an asset on a blockchain, which can make it available to holders through digital wallets and connect it to other on-chain services. Securitize communications head Tom Murphy said asset managers were launching tokenized funds, companies were exploring on-chain equity, and financial firms were building stablecoin infrastructure. He also pointed to tokenized assets appearing in lending, collateral and trading markets.

People holding tokenized assets were spread unevenly across countries. U.S. residents made up 11.7% of the global tokenized real-world asset holder base and held 11.4% of its value. For tokenized Treasuries, geolocated value was led by Italy at 29.9%, Portugal at 18.8% and Singapore at 18.2%; the United States ranked fourth with 10.5%.

The ranking by number of holders looked quite different. Japan had 16.4% of holders but just 0.4% of the value. The gap between the holder count and the value held suggests that some countries have many people with relatively small positions, while a smaller number of large holders account for substantial value elsewhere. The dataset doesn’t show how these holdings are distributed within each country.

Access through a blockchain wallet may make a Treasury product available to someone who can’t easily use a U.S. brokerage account or buy a large wholesale position. That expands the potential reach of U.S. debt while giving holders in other countries a dollar-linked asset. It also means that tokenization can carry financial value across borders without requiring the holder to be based in the country that issued the underlying asset.

Lending figures show that tokenized assets are used differently depending on the asset. Wallets lending private credit grew 1,618%, and private credit made up 77% of the value lent against tokenized assets. Lending involving commodities grew 875%. Treasury lending contracted 23%, the only category to fall.

One explanation offered in the report is that tokenization gives traditionally illiquid private credit a form that can be pledged as collateral. Treasuries already have established routes to liquidity in conventional finance, so a tokenized Treasury may be more useful to hold or transfer than to borrow against. The figures indicate differing use patterns, though they don’t show why each holder chose to lend or retain a given asset.

Securitize’s view is that users may eventually focus less on whether an asset is “crypto” and more on whether it gains speed, programmability and global reach from blockchain infrastructure. The source also describes activity on permissioned systems such as Canton, where selected institutions can settle tokenized securities and cash. Such activity isn’t included in public-chain holder counts, so the reported market figures don’t capture every tokenization effort.

Canada’s crypto activity is tied to cross-border use

Canada’s crypto economy grew 2.3% to $247.7 billion between July 2025 and June 2026. It ranked seventh globally for domestic and cross-border peer-to-peer activity, placing first among G7 economies in the report’s peer-to-peer measure. Stablecoins accounted for 44% of Canadian-dollar order-book volume in June 2026, up from 30% a year earlier. USDC led this category, while Canadian-dollar stablecoins saw little trading activity.

That preference creates a tension: a large share of Canadian stablecoin activity uses a currency issued outside Canada. Canadian Web3 Council executive director Morva Rohani said demand for dollar stablecoins can support U.S. debt through the reserves backing those tokens, rather than creating equivalent demand for Canadian debt. The same dynamic can apply elsewhere when users choose dollar-based tokens over local alternatives.

Rohani said Canadian policymakers were following the trend. Canada has cleared a path for Canadian-dollar stablecoin issuers, and some prospective issuers are focused on domestic and international payments. So far, those tokens remain small in market value and use compared with dollar stablecoins. The gap illustrates the difficulty of turning a local-currency product into a widely used payment option.

Canada also ranked first for cross-border flows in the report’s big-money measure, while placing seventh on its grassroots adoption measure. Its large foreign-born population may help explain the strength of those flows. About 23% of the country’s population was born abroad, the highest share in the G7, Rohani said. She identified remittance routes to India, the Philippines, Nigeria, Pakistan and China as heavily used.

Stablecoins can reduce the cost and friction of sending money, particularly for people who make frequent transfers to support family, repay obligations or manage assets in another country. The report examined Canada-to-India, Canada-to-China and Canada-to-Philippines activity, three routes with fast-growing cross-border flows. Together, those destinations represented 19.4% of Canada’s outbound cross-border stablecoin value, up from 10.9% a year earlier.

Remittance-sized stablecoin and bitcoin transfers from Canada to the three countries totaled $2.36 billion during the period, a 141% year-over-year increase. The activity covered 3.56 million transactions and averaged 75,100 unique senders per month. The average transfer was $663.

Those crypto transfers remain one part of much larger remittance routes. Global remittances were about $905 billion in 2024. India received roughly $129 billion, China about $48 billion and the Philippines about $40 billion. Canadians sent an estimated $8.3 billion abroad in 2023, much of it to those same destinations. Against those totals, the measured $2.36 billion on crypto rails is a growing but limited share.

Cost may help explain the interest. The World Bank put the average cost of sending money from Canada near 6.7%, with bank transfers above 7%. A stablecoin transfer can settle for a cent or less regardless of the amount, according to the source. The comparison describes settlement cost; it doesn’t account for every expense a sender may face, such as converting between currencies or accessing a wallet.

Domestic peer-to-peer volume moved with temporary visa populations

Canada’s domestic peer-to-peer activity also tracked the number of people holding temporary visas. The report compared activity with study permit holders from China, India and the Philippines, three major remittance routes, starting in mid-2021. As the permit-holder population grew, domestic peer-to-peer crypto volume rose; as the population and permit issuance later fell, activity declined as well.

The relationship is associational. It doesn’t prove that temporary visa holders caused the change in peer-to-peer volume, and the report doesn’t quantify how much activity came from any one group. A plausible connection is that people with regular cross-border financial needs may be more likely to try a settlement method that can lower transfer costs. Anecdotes in the report include landlords accepting bitcoin for rent, residents paying bills in crypto and small businesses accepting digital asset payments; these examples don’t establish how common such use is.

The broader regional pattern is a split between where infrastructure is built and where it is used. U.S. companies and developers feature prominently in x402, dollar stablecoins and tokenized Treasury products, while users across the world contribute demand. Canada illustrates how that demand can run through foreign-currency rails even as local alternatives remain small.

Efforts to build alternatives are underway in several markets, including euro-denominated stablecoins, locally issued tokens in Singapore and Hong Kong, and Brazil’s Drex pilot for a digital real. Their presence shows that competition is not limited to technical design: a token also needs users, liquidity and services that make it useful. The data in this report points to adoption as the test that will determine whether local options can gain ground.

For agentic commerce, the near-term signal to watch is whether wallet replenishment and rapid first payments persist as x402 reaches more services. The payment data already shows small transactions flowing through automated systems, but lasting scale will depend on useful services, clear spending controls and a way to link an agent’s activity to the people who funded it.