BNB Chain has launched BNB Agent Studio v3, a new version of its toolkit aimed at improving how AI agents hold, spend, and earn funds within the BNB Chain ecosystem. According to Techub, the release adds support for Turnkey wallets, making wallet integration broader for agent-based use cases. It also simplifies access to tBNB, a change designed to make testing or ecosystem interaction more straightforward for developers and users working with AI agents. The update also brings in Travala.com’s MPP payment function and expands support for the b402 protocol. BNB Chain said these changes are intended to strengthen the financial operating capabilities of AI agents and improve user experience across the ecosystem. The announcement was attributed to @BNBCHAIN.

Atul Khekade, co-founder of XDC Network, said AI agent payments are removing delays tied to banks, clearing houses, and business hours, turning payments from a waiting process into a background function that runs at the speed of the underlying network. He said this points to new forms of payment, including possible changes to how invoicing and credit card payments work. Research cited in the report shows that AI agents carried out more than 176 million on-chain transactions between May 2025 and April 2026, with total value above $73 million. Of those transfers, 76% were below the $0.30 fixed fee threshold commonly associated with card payments. USDC accounted for 98.6% of settlement volume. To support high-frequency, low-value payments, several companies have introduced related tools and standards. The report names Stripe’s Machine Payment Protocol, Google’s Agent Payment Protocol, Cloudflare’s wallet and payment tools, and Mastercard’s Agent Pay for Machines. XDC is also entering the segment through XDCAI.tech, using the open x402 protocol originally from Coinbase.

Stripe used an investor letter dated Aug. 19 to do more than announce its acquisition of OpenRouter, which it described as the largest deal in the company’s history. The letter lays out a broader thesis: in the AI era, businesses will need to manage two core digital flows, one for money and one for intelligence. Stripe says it already built the first and is now assembling the second through OpenRouter and earlier acquisitions including Bridge, Privy, and Metronome. The company argues that building economic infrastructure for the internet and for AI is essentially the same job. It points to several shifts already under way: faster company formation, rising usage-based pricing, AI agents nearing the status of independent economic actors, and rapid stablecoin adoption. Stripe says more than 5 million businesses already use its platform and that funds flowing through Stripe account for about 2% of global GDP. It also says 88% of companies on the Forbes AI 50, including OpenAI and Anthropic, are built on Stripe. In Stripe’s framing, OpenRouter gives developers a reliable way to manage the “intelligence pipeline,” much as Stripe did for revenue. The company says OpenRouter runs the world’s largest and most trusted token routing engine and that token consumption on the platform has compounded at 9% per week year to date.

OSL Group, the stablecoin payments and trading platform, has unveiled OSL AgentPay, a stablecoin payment infrastructure built specifically for AI agents. The infrastructure targets machine-to-machine payments, where agents settle small-value transactions without a developer intervening in each step. The new product is designed to give developers' artificial intelligence agents a fully autonomous payment workflow, enabling high-frequency micro-payments and value exchange between agents. OSL AgentPay supports multiple stablecoins including USDT, USDC, and USDGO. It also works with several payment protocols such as x402, AP2, and MPP. The system offers eight core capabilities: execution interface, multi-asset route selection, stablecoin abstraction, micro-payment support, zero on-chain fees, multi-protocol compatibility, multi-wallet compatibility, and global fiat on/off ramps. Developers only need to express a payment intent, specifying the amount, asset, and recipient. Their AI agent can then direct AgentPay to handle routing, signing, and settlement. Looking ahead, OSL plans to integrate Banxa's global fiat on/off ramp capabilities and the treasury operations of its enterprise stablecoin USDGO into AgentPay.

Visa and Mastercard are building out payment rails for AI agents, though the market is still tiny by card-network standards. Visa launched Intelligent Commerce Connect on April 9, a single integration layer that lets merchants accept payments from four agent protocols. Mastercard followed on June 10 with Agent Pay for Machines, aimed at high-frequency, low-value transactions between machines and supporting settlement across cards, bank accounts and stablecoins. On the crypto side, teams are already showing working demos: XDC said it connected Anthropic’s Claude to onchain payments in July and demonstrated an AI ordering and paying for coffee. Even so, a joint study by Visa and Artemis found that x402 and MPP processed more than 110 million transactions over the past 12 months, with total volume of just $136 million, or about $1.24 per transaction. The gap between technical capability and real-world adoption remains large, and questions around identity, authorization, risk controls and accountability are still unresolved.

Tiger Research argues that crypto wallets are quietly being rebuilt for a world where AI agents, not humans, initiate large volumes of online payments. In that model, a single user request could trigger dozens of machine-made transactions, while individual API calls or data queries may cost as little as $0.001 or even $0.00001. That makes traditional card infrastructure, with chargebacks and fixed per-transaction fees, poorly suited to the task. The report says more than 10 companies are already developing wallets specifically for agents. Coinbase, Binance and other firms are investing despite little visible near-term revenue because they want to secure future users and transaction flow before agent activity scales. Using Coinbase’s disclosed metrics and a base of 9.2 million monthly transacting users, Tiger Research estimates that annual revenue under an aggressive scenario could reach $50.37 billion, roughly 7x Coinbase’s current total revenue. The report also says wallet payment histories could become the basis for credit models and revenue-based financing for agents, much as Stripe Capital used merchant payment data. Still, Tiger Research stresses that this remains a forward-looking scenario. Payment reliability, fragmented protocols such as x402, AP2 and MPP, and unresolved legal and KYC questions continue to limit the market today.

Tiger Research argues that crypto wallet providers are laying the groundwork for AI agents long before mass adoption arrives. The report says autonomous agents will need to make huge volumes of tiny payments for API calls, data access, and online services, often worth fractions of a cent, which makes traditional card networks ill-suited for the job. In that setup, programmable wallets become the starting point for machine-to-machine commerce. The report points to growing industry activity, with more than 10 companies already developing wallets tailored to AI agents. It says firms such as Coinbase and Binance are not chasing near-term revenue. Instead, they are positioning for a future user base and future transaction flow. Using Coinbase’s disclosed figures and a model based on 9.2 million monthly transacting users, Tiger Research estimates that annual revenue tied to agent activity could range from about $84 million in a conservative case to roughly $50.37 billion in an aggressive case, or around 7x Coinbase’s current total revenue. Tiger Research also says wallet transaction histories could later support agent-focused lending models similar to revenue-based financing. Still, the firm stresses that this remains speculative. AI agents still make payment mistakes, payment protocols remain fragmented, and the legal and compliance status of agents is unresolved.

Web3 AI network X-Agent has announced the open-source release of xagentAI/xpense after completing a deep integration with OKX Agentic Wallet and launching a paid MCP tool on the OKX marketplace. The project is positioned as a payment control plane for AI agents, aimed at filling a missing control layer in autonomous spending, budget risk management, and delivery verification. According to the announcement, existing standards and tools including HTTP 402, x402, MPP, MCP, and OKX Agentic Wallet already support signing and micropayment rails, but they do not fully address the upper-layer controls required when machines spend funds on their own. X-Agent said xpense is designed to answer practical questions such as who sets spending limits for an agent, how repeated charges can be prevented, and how a system can verify that a paid service was actually delivered. The project lays out a seven-step deterministic flow covering Intent, Policy, Route, Pay, Deliver, Verify, and Account. It also adopts a dual-state-machine model that separates payment settlement from service delivery. X-Agent said the system also relies on policy-engine controls, a payment-aware router, and OKX infrastructure including TEE-based key isolation and x402 gas-free USDC settlement.