Here is a quiet contradiction behind the global rise of Chinese artificial intelligence (AI). A model such as DeepSeek can answer a question in English, Spanish, or Arabic from day one — language is no barrier at all. But the moment a user in Brazil, Germany, or Indonesia tries to pay for that service, the company runs into a wall that has nothing to do with intelligence: local payment habits, fragmented tax rules, subscription billing, fraud checks, and a corporate structure split between entities inside and outside China.
A great model travels instantly. Money does not.
That gap is the subject of a recent move by Antom, the merchant-payment arm of Ant International (蚂蚁国际), the global affiliate of the company behind Alipay. On September 22, 2026, Antom announced a one-stop solution aimed specifically at Chinese AI companies expanding overseas, covering subscriptions, payments, tax, risk control, accounts, and treasury management, and designed to work across a company's separate domestic and foreign business entities. Several of China's best-known AI names — DeepSeek, Kimi, Qwen (via QwenCloud), and the design agent Lovart — are already using Antom, the company said.
To understand why a payment product is news, you have to understand how unexpectedly hard it is for an AI company to simply collect money abroad.
What Antom Actually Announced
Let us establish the facts before explaining them.
Antom describes itself as a one-stop platform for global payments, business growth, and digital services, under Ant International. The new offering targets the specific needs of AI businesses — subscription, payment, tax, risk control, account, and fund management — and is built to match the multi-entity structure common among Chinese firms, which typically operate one company in the Chinese mainland and others overseas.
The scale figures are broad. Antom says it supports more than 300 payment methods, reaches over 200 markets, and processes over 100 currencies, with a network built largely through acquisitions and partnerships — including Southeast Asian platform 2C2P, European provider MultiSafepay, Japan's KOMOJU, and the small-business platform EPOS. The same modular tools for billing, reconciliation, and tax can also serve SaaS (Software as a Service) companies, developer platforms, and digital-entertainment businesses, not just AI labs.
Liu Zheng, Senior Vice President of Ant International and Chief Executive Officer of Antom, framed the work in three parts in remarks reported by Chinese media: integrating fragmented domestic-and-overseas subscription systems; connecting the full lifecycle from acquiring and risk control to tax and multi-currency accounts; and embedding AI throughout so companies can "respond faster, grow faster, and transact more safely."
None of that sounds glamorous. That is exactly the point.
Why Collecting Money Is the Hard Part
To a reader outside the industry, "accepting payments" sounds like a solved problem. For an AI company selling globally, it is several problems at once.
1. Every market pays differently. Credit cards are not universal. In the Netherlands, shoppers prefer the local system iDEAL; in Brazil, the instant-payment network PIX; in India, UPI (Unified Payments Interface); across Southeast Asia, a tangle of e-wallets such as GCash and DANA. A company that offers only Visa and Mastercard quietly loses customers at checkout — abandoned carts it never sees. Offering dozens of these methods normally means separate technical and banking relationships in each market.
2. Tax follows the customer, not the company. This trips up even technical founders. If you sell a digital service to a consumer in the European Union, you generally must collect VAT (Value-Added Tax) at the buyer's local rate — between roughly 17 and 27 percent depending on the country — and file it with the correct authorities, with effectively no revenue threshold for non-EU sellers. In the United States, sales-tax rules vary across thousands of state, county, and city jurisdictions, and an "economic nexus" (a tax obligation triggered purely by sales volume, not physical presence) can apply once a company passes a state's threshold. The EU, United Kingdom, Japan, Singapore, and Australia each run their own regimes, and Japan even requires a local tax agent. An AI startup can be technically non-compliant abroad after just a handful of foreign sales without realizing it.
3. AI billing is unusual. Traditional software is sold as a flat subscription. AI is often metered — charged per token (the basic unit a model reads and generates), per API (Application Programming Interface) call, or per second of computing time — sometimes blended with monthly plans. A payment system has to handle variable, usage-based amounts triggered dynamically by the company's own backend, plus prepaid credit balances and recurring subscriptions at the same time.
4. AI is treated as a risky category. Because the field is new, some banks and payment providers view AI services as prone to chargebacks — a user might dispute a charge because they simply "did not like" the text or image a model generated. That requires more sophisticated fraud detection than an ordinary retailer needs, and a payment partner with enough credibility with banks to keep transaction approval rates high.
5. Chinese companies face an extra structural problem. A Chinese AI firm expanding overseas typically has to manage money moving across separate legal entities — a mainland operating company and foreign entities that hold the international business. Reconciling subscriptions, settlements, and tax across that split is a specialized headache that Western payment tools were not built to address.
Antom's pitch is that one integration handles the entire stack rather than the company stitching together separate tools for each piece.
What Antom Is — and Where It Came From
Antom is not a new fintech startup riding the AI wave. The name is built from "from Ant to merchants," and the underlying business traces back to 2012, when it began handling global e-commerce payments for merchants on AliExpress. By 2016 it had expanded to serve large international brands; from 2022 it folded regional payment partners under one brand; and in 2024 it launched what it calls the first AI assistant for merchant payment services, Antom Copilot.
Today the company says it serves over a million merchants and emphasizes AI across four product lines:
- Antom Copilot — a business assistant across the merchant lifecycle, recommending payment methods and technical setups, analyzing payment success, and helping with disputes and risk rules.
- Antom Shield — a fraud-prevention system making real-time decisions, trained (the company says) on large global data sets.
- A global payment manager — smart routing that chooses among payment channels; Antom claims it can raise success rates while cutting integration work.
- A+ Rewards — a marketing engine connecting merchants to mobile wallets and consumer traffic.
The growth figures the company cites are strong: for the first ten months of 2025, direct acquiring volume from non-Alipay users reportedly rose more than 70 percent year over year. In November 2025 it also launched an EPOS360 app to extend services from large enterprises down to small and medium businesses, starting in Singapore with backing from ANEXT Bank, one of the city-state's digital wholesale banks.
The Global Context: Stripe, Adyen, Paddle
Placing Antom against Western players clarifies what it is — and what makes the Chinese offering different.
Stripe is the developer-friendly payment processor: flexible and relatively cheap to start, but the merchant remains the legal seller and largely handles its own global tax compliance, with tax and billing features charged separately. Adyen is the enterprise platform behind companies such as Uber and Spotify, with its own global acquiring network and transparent interchange-based pricing, but a longer onboarding process and minimum volumes aimed at larger firms. Paddle is a Merchant of Record (MoR) for software: it legally becomes the seller to the customer and folds tax, fraud, and subscriptions into one all-inclusive fee — convenient, but at a higher percentage and with less checkout control.
Antom does not position itself exactly as any one of these. Its stated differentiators are deep coverage of Asian and emerging-market payment methods, local acquiring licenses across key regions, and — crucially for its target customers — an explicit understanding of the multi-entity architecture of Chinese companies and the specific needs of AI and SaaS billing. For a Chinese founder, the appeal is a partner that already knows how money, tax, and corporate entities connect between the mainland and overseas markets, rather than forcing the founder to explain that structure from scratch.
What to Keep in Mind
A few qualifications keep the story grounded.
First, the headline figures — 300-plus payment methods, 200-plus markets, over a million merchants — come from Antom itself and describe network coverage, not the number of methods any single merchant necessarily uses. Claims such as improved payment-success rates or a 30-percent average lift in merchant sales are company-reported marketing figures, not independently audited results, and will vary widely by business.
Second, this is a crowded, competitive field. Chinese firms already have options ranging from global processors to other cross-border payment specialists, and Western players continue to expand into Asia. Antom's AI-specific package is a real product, but it is also a competitive move aimed at a fast-growing category — not the only viable route abroad.
Third, the AI partnerships Antom cites are, so far, largely described at the level of "working together" on payment acceptance; the precise depth and terms of each relationship are not public. Readers should treat them as commercial collaborations rather than exclusive arrangements.
The Bigger Picture
Step back, and the announcement points to something larger than one company's product launch.
The world tends to watch the AI race through the most visible layer — which model tops a benchmark, who has the most parameters, whose chip is faster. But underneath every global AI service sits a far less glamorous layer of infrastructure: the plumbing that lets a stranger on another continent pay, safely and compliantly, for a question asked in their own language. That plumbing is easy to ignore until it fails — and when it fails, it blocks growth as surely as a technical shortfall.
Chinese AI models crossed borders remarkably fast. The businesses behind them are now discovering that going global is not only a model problem. It is a money problem — payments, tax, risk, and the architecture that ties them together. Whether or not Antom ends up the winner, the quiet race to build that financial plumbing is now part of the story of Chinese AI going abroad.