As the artificial intelligence industry advances rapidly, the demand for computing power has surged significantly. Recently, several major banks, including Agricultural Bank of China, Bank of Communications, China Merchants Bank, and Shanghai Pudong Development Bank, have transformed into "computing power intermediaries," successively launching AI-themed bank cards. These cards offer users various AI-related benefits, drawing widespread attention.
Industry experts believe these AI bank cards can effectively fill the gap for dedicated financial products in the computing power consumption sector. It is anticipated that more small and medium-sized banks will follow suit with similar offerings. However, for large-scale implementation, challenges such as difficulties in computing power delivery and ambiguous regulatory boundaries remain. Furthermore, to build a core competitive edge for retail finance that can endure market cycles, banks should not limit themselves to being mere "computing power intermediaries." They must progress from simply "providing computing power" to "creating an ecosystem."
From "Capital Intermediaries" to "Computing Power Intermediaries": Multiple Banks Ramp Up AI Card Launches
On July 10, Agricultural Bank of China officially launched its KIMI co-branded credit card. This card is available in standard and premium platinum tiers, with different annual fees and corresponding AI benefits. Standard cardholders receive benefits such as Agent credits, KIMI Code credits, and website deployment functionality.
On the same day, Bank of Communications officially launched an artificial intelligence-themed debit card. Customers who bind the card to China UnionPay's mobile payment service can receive a dedicated AI service benefits fund (10 yuan per person) from leading large language model providers. Additionally, cardholders who register for and meet the requirements of the bank's asset upgrade promotion via its mobile banking app can receive an extra 100 yuan per person in AI service benefits. This fund can be used when making a single purchase of 199 yuan or more with the same card bound to UnionPay.
Shanghai Pudong Development Bank recently collaborated with China UnionPay and Alibaba Cloud to launch the SPDB Tech Elite Credit Card (Cloud Intelligence Edition). This product's benefits are precisely tailored to meet the essential need for computing power procurement, covering all Alibaba Cloud platform products that support UnionPay payments. These include tools like AI agents, general-purpose large language models, code assistants, and video generation models, providing comprehensive support for scenarios such as AI content creation, automated program development, and algorithm model debugging. The total computing power subsidy can reach up to 30 billion Tokens for the Qwen large model.
On June 12, CM BANK's American Express Engineer Credit Card introduced exclusive AI benefits. Users can choose one of three packages: 1 month of MiniMax Token Plan Max, 2 months of MiniMax Token Plan Plus, or 1 month of MaxClaw Basic plus 1 month of MiniMax Token Plan Plus. The highest-tier offering provides users with up to 1.8 billion Tokens of M3 model usage per month.
On June 30, Ping An Bank, in partnership with China UnionPay and Tencent Cloud, launched the country's first debit card for individual AI computing power consumers—the "AI Computing Card." This card integrates functions such as aggregated computing power payment, basic financial services, and exclusive computing power benefits. Cardholders can accumulate computing power benefits through daily spending.
Furthermore, on June 25, MYbank announced a collaboration with ecosystem partners including Alibaba Cloud, Sesame Enterprise Credit, and Amap to create the country's first AI benefits card for small and micro business operators. The upgraded Business Gold Card addresses the pain points of small business operations by offering services like Token trials, content generation, customer acquisition, business opportunity assessment, and store location selection.
Discussing the reasons behind banks' intensive launch of AI cards, Lou Feipeng, a researcher at Postal Savings Bank of China, believes the primary goal is to capture the entry point to the AI ecosystem. It also reflects a shift in banks' customer acquisition logic from "price wars" to "scenario-based competition."
"Traditional benefits like points and air miles have diminishing marginal utility, whereas 'computing power' is an essential production factor for the digital-native generation. By bundling computing power resources, banks embed financial services into high-frequency user scenarios like creation and development. They aim to lock in high-net-worth younger customer segments with the dual attributes of 'finance + technology,' building differentiated competitive barriers," Lou Feipeng stated.
Tian Lihui, a professor of finance at Nankai University, remarked that the dense rollout of AI cards by banks represents a paradigm shift in retail finance from "consumption rebates" to "production empowerment." Traditional benefits are trapped in homogeneous competition over items like Starbucks vouchers and mileage points, leading to rising customer acquisition costs. In contrast, AI computing power, as a new form of production factor, can precisely target high-net-worth groups like developers and content creators. "Such users have an average annual computing power expenditure exceeding 10,000 yuan, yet credit card penetration among them is less than 15%," he noted.
Tian Lihui added that at a deeper level, banks' customer management logic is shifting from "stratification by asset size" to "stratification by behavioral scenarios." Banks are no longer solely looking at AUM figures but are identifying genuine needs through professional behaviors. As Tokens become the "new coffee vouchers," bank cards evolve from payment tools into gateways for digital productivity. This is both a breakthrough strategy in a saturated market and a strategic leap for the financial industry from being "capital intermediaries" to becoming "computing power intermediaries."
From "Providing Computing Power" to "Building an Ecosystem": Forging a Core Competency in Retail Finance
With the rapid development of the AI industry, the number of domestic AI R&D professionals and ecosystem participants has grown substantially, creating demand for specialized financial services in scenarios like computing power procurement.
Interviewees generally agree that following the initial trials by major banks like Agricultural Bank, Bank of Communications, and CM BANK, more small and medium-sized banks will follow with their own AI card offerings. However, large-scale implementation still faces multiple pain points.
"It is expected that some small and medium-sized banks will follow suit within the next six months, but achieving scale faces three practical bottlenecks: cost, standards, and liability," analyzed Tian Lihui. "First, cost volatility is significant, with monthly fluctuations in computing power prices exceeding 20%, making it difficult for banks to lock in redemption costs. Second, standards are fragmented, as Tokens from different providers are not interchangeable, forcing users to switch between platforms. Third, liability is ambiguous; when model interfaces fail, responsibility attribution is unclear."
"Furthermore, compliance risks require heightened vigilance. The cross-use of financial transaction data and AI usage behavior data must obtain separate, explicit user authorization. Physical isolation firewalls must also be established to prevent the outflow of raw data," Tian Lihui pointed out. "Currently, some banks simply equate Token benefits to 'points redemption' without clarifying data flow boundaries, which easily risks violating the 'Personal Information Protection Law.' The truly sustainable path lies in embedding computing power services into a closed-loop financial scenario. For example, using AI-generated financial report analysis to directly trigger credit approval, rather than merely gifting isolated Token allowances."
Lou Feipeng also expects small and medium-sized banks to follow quickly, but large-scale implementation faces challenges related to computing power delivery difficulties and ambiguous compliance boundaries. For instance, technically, banks lack computing power scheduling capabilities, and reliance on third parties can lead to a fragmented user experience. Regarding compliance, the flow of user data between financial and computing power platforms carries risks of privacy leakage.
Some viewpoints suggest that the current AI bank cards offered by various banks primarily cater to niche customer segments like developers and content creators, potentially having limited appeal to ordinary users. So, how can banks avoid letting AI benefits become just another homogeneous marketing gimmick and instead build them into a sustainable, integrated financial and technology ecosystem?
In response, Lou Feipeng stated that to avoid homogenization, building a robust ecosystem is essential. Banks should not act merely as distributors of computing power but should integrate resources from multiple parties to create a "payment—computing power—application" closed loop. For example, they could develop low-threshold AI applications for ordinary users, such as intelligent investment advisors or family AI assistants, transforming computing power into tangible service experiences. Simultaneously, establishing developer communities and promoting internal ecosystem circulation through events like hackathons can make the benefits genuinely foster user loyalty and commercial value.
Tian Lihui expressed that the core of avoiding homogenization lies in shifting from "providing computing power" to "building an ecosystem." Currently, most products remain at the superficial level of "swipe card, get Tokens." However, the practice of MYbank demonstrates that when an AI store location tool is directly embedded into the small business loan approval process, Tokens cease to be an abstract benefit and become a tangible boost to operational efficiency.
"Banks should focus on two key factors: segmentation and scenario integration," Tian Lihui noted. "First, segmentation must be precise. For ordinary users, emphasize lightweight tools like intelligent income and expenditure analysis. For tech-savvy customer groups, offer higher-level permissions like concurrent Agent access. Second, scenarios must be internalized, deeply linking computing power consumption with financial behaviors. For example, dynamically adjusting credit limits based on the quality score of AI-generated financial reports."
Tian Lihui believes that making computing power the "invisible engine" of financial services, rather than just a marketing tactic, is crucial for building a positive cycle of "using the card—gaining computing power—improving efficiency—increasing loyalty." This is precisely the core competency that enables retail finance to navigate through periods of market saturation.
Comments