Wu Peng finally received his long-awaited IT investment budget.

As the head of the IT department of a rural commercial bank in the central region, he has often received complaints from business departments in the past two years. For example, traditional centralized databases have become difficult to maintain stability and high performance in the face of rapid growth in business data volume; Centralized databases are no longer able to support large-scale concurrent access for business partners, and some high concurrency businesses have encountered "account locking" issues, making it difficult to provide 24/7 services for loans and other businesses.

In April 2026, the board of directors of this rural commercial bank held an extraordinary meeting and decided to allocate nearly 10 million yuan in special funds to replace the database architecture of core businesses such as deposits and loans with domestically produced distributed databases.

In the past two months, Wu Peng has been in close communication with five domestic database vendors, hoping to finalize the specific technical architecture of domestic distributed databases as soon as possible.

Unexpectedly, when he proposed requirements such as "lightweight local deployment, low-cost migration, minimalist operation and maintenance, and vendor on-site operation and maintenance services" to these database vendors, three small and medium-sized vendors quickly gave up, leaving only two industry-leading vendors still providing domestic distributed database local deployment solutions.

These requirements are not excessive, I don't know why small and medium-sized manufacturers choose to withdraw, "Wu Peng said somewhat puzzled.

Wu Peng stated that compared to four years ago, banks have made a "qualitative" upgrade in their performance and operation requirements for domestic distributed databases. In 2022, banks' requirements for domestic distributed databases were focused on "whether they can be used", but now their demands have shifted to "whether they can operate stably, iterate quickly and be easy to operate, and maximize cost control".

IDC China Research Manager Wang Nan analyzed to reporters that on the one hand, the core systems of top database vendors' domestically produced distributed databases continue to increase in verifiable cases of local deployment, and data migration services and ecological construction are becoming increasingly mature, attracting more banks to favor them; On the other hand, many small and medium-sized database vendors have difficulty making breakthroughs in core financial scenario services, and their market share continues to shrink, forcing them to leave.

According to IDC (International Data Corporation) data, as more banks accelerate the "import substitution" of core system databases, the local deployment market size of domestically produced distributed databases in China's financial industry will reach 2.02 billion yuan by 2025, a year-on-year increase of 37.6%. Over the past year, more than 60 small and medium-sized database vendors have chosen to withdraw, while industry leaders such as Ant Group's self-developed native distributed database OceanBase, Jinzhuan Xinke, Tencent Cloud, Huawei Cloud, and Alibaba Cloud have seen their market share increase from 86.9% in 2023 to 90.9% in 2025. As of now, there are about 100 database vendors still operating in China, which means that about 38% of companies have chosen to exit in the past year.

This also indicates that the database service industry in the financial industry has already passed the period of import substitution dividends and has begun to enter the stage of compound competition of capability+ecology+scenario, "said Wang Nan. Affected by the booming development of "AI+finance", banks have put forward new demands for domestic distributed databases since the beginning of this year, that is, generative AI (artificial intelligence) needs to be integrated with database development.

Banks are no longer satisfied with whether they can be used or not

In order to better complete the transformation of the bank's core system database, Wu Peng sought advice from the IT department heads of multiple banks.

The other party clearly stated that this work cannot be simply understood as "moving the business data of core businesses (deposits and loans, transfers and remittances, account management, etc.) from one database to another", but essentially involves the iterative reconstruction of the bank's overall financial infrastructure.

In this process, banks must ensure that the new financial infrastructure operates safely and reliably. Any minor malfunction (such as a network card being incompatible with the operating system) can affect the performance and orderly operation of the entire distributed database.

This made Wu Peng realize that his initial focus on whether distributed databases could be used was completely outdated. Nowadays, banks place greater emphasis on whether the distributed database technology route can achieve the following aspects: controllable data migration risks, strong performance (capable of carrying massive data and high concurrency transactions), sufficient compatibility (able to reduce upgrade and transformation costs), high availability at the financial level (ensuring stable operation of the system 24/7), and convenient local deployment and operation (improving operation efficiency and reducing costs).

In early June, when Wu Peng proposed these new requirements to five database vendors, the three small and medium-sized database vendors changed their previous "responsive" attitude and instead proposed to renegotiate the new database technology architecture and local deployment path with the company's technical team. After two weeks, they all informed each other that they could only withdraw from the bidding due to insufficient technical capabilities.

Wang Zijian from the Software Development Center of Beijing Bank also found that the bank's demand for distributed database architecture is continuously upgrading. Wang Zijian stated that on the basis of "product usability and ease of use", banks have put forward a series of new requirements for distributed database performance and service providers: to achieve 100% independent research and development of database core code; Need to have elastic scalability characteristics, able to cope with the explosive growth of business data volume and meet the needs of business elastic scaling with ease; We can provide continuous optimization tools such as compatibility assessment, data migration, monitoring, etc., to completely solve the pain points of traditional centralized databases in terms of scalability, availability, and cost.

The iterative upgrade of banks' demand for distributed databases is far from over, "said Wang Zijian. Starting from the construction of the data base in 2022, the data governance and labeling work of Bank of Beijing is still ongoing, and there is still a long way to go before achieving "intelligent questioning" in business scenarios (using AI intelligent customer service to accurately understand user questioning intentions and provide professional responses). Database assistance is needed to build stronger data analysis capabilities.

Forcing Database Vendors

In 2021, as the country continued to promote financial innovation, Feng Yu sensed that domestic distributed database services would have broad development prospects, so he joined a small and medium-sized database vendor as the Director of Operations.

In recent years, the market size of distributed database services in the financial industry has maintained an average annual growth rate of over 30%, which confirms Feng Yu's prediction. However, his company did not get a share of it.

Three years ago, large state-owned banks and joint-stock banks were concerned about whether domestic distributed databases could be used, and were willing to try non full stack and dual track parallel models. With the deep understanding of bank database operation and maintenance by our technical team, we won some business opportunities, "Feng Yu recalled. Nowadays, "product availability" is no longer the priority consideration for banks when choosing database vendors. Banks are more focused on long-term service capabilities and database ecosystem construction strength, which has led to a continuous widening gap in product competitiveness between small and medium-sized database vendors with limited R&D investment and industry-leading enterprises.

With the completion of local deployment and transformation of distributed databases by large state-owned banks and joint-stock banks, the new business opportunities in the financial industry database service market from 2025 will mainly focus on small and medium-sized banks, as well as non bank financial institutions such as securities firms and insurance companies. The latter particularly values whether database vendors have accumulated enough verifiable cases of distributed database services and can provide low-cost long-term operation and maintenance services and database iteration upgrade technical support.

Feng Yu admitted that three years ago, his database vendor could still sporadically obtain distributed database renovation contracts from two or three banks. But from 2025 onwards, this business will have almost no revenue, and the company can only rely on hundreds of thousands of yuan in operation and maintenance revenue from previous orders to support the overall operation of the enterprise. In the past year and a half, the accumulated losses have exceeded 6 million yuan.

Feng Yu said that driven by factors such as the upgrading of bank demand, the distributed database service market is showing a development trend of "the stronger the stronger, the weaker the weaker".

At present, OceanBase serves over 400 financial institutions, covering various types of banks such as state-owned large banks, joint-stock banks, city commercial banks, and rural commercial banks, among which banks with trillion dollar scale account for nearly 70%. In the non bank sector, 75% of top insurance institutions, 80% of top securities firms, and 60% of top fund institutions choose Ocean Base to be equipped with core business systems.

Weng Rui, General Manager of OceanBase's Financial Government and Enterprise Business Unit, revealed to reporters that in the past 15 years, one of their core jobs has been to catch the most critical business loads in the financial industry. Therefore, they explored the corresponding best practice solutions through the business scenarios of Ant Group and Alibaba. These scenarios are highly isomorphic to the core systems of financial institutions in terms of concurrency scale, consistency requirements, and fault tolerance standards.

Weng Rui said that after years of effort, OceanBase's distributed database solution can support dual active across clouds and data centers, automatically switching traffic in case of accidents; Support seamless scaling of applications through native distributed architecture; A database supports both trading and analysis, and an integrated storage engine supports real-time analysis; Compatible with Oracle and MySQL syntax, as well as upstream and downstream ecosystems.

Wang Nan candidly stated that in the current distributed database market, the production level SLA (Service Level Agreement), controllable migration risks, and long-term service capabilities built by industry leaders are becoming a competitive barrier for them to continuously expand their market share, making it difficult for small and medium-sized database vendors to resist. Moreover, with the rapid development of "AI+finance", banks hope that distributed databases can become the data foundation to support the AI era. Faced with the huge investment of industry leaders in promoting the integration of databases and generative AI, the survival situation of small and medium-sized database vendors with insufficient technology research and development budgets is becoming even more difficult.

The AI era is intensifying differentiation

Since the beginning of this year, when Feng Yu communicated with banks about the cooperation of distributed database business, he found that the latter attaches great importance to the integration and development of distributed databases and generative AI.

The IT department heads of two joint-stock banks directly informed that their senior management has determined that two forces will reshape the bank's data foundation in the future: first, more AI intelligent agents will be applied in business scenarios such as customer service, credit, product marketing, and wealth management; Secondly, the massive amount of unstructured data such as customer service voice, face-to-face voice, due diligence images, financial texts, social texts, etc. in banks cannot be directly accessed and analyzed by traditional databases, resulting in significant business value being buried and urgently needing to be activated.

Therefore, the two joint-stock banks mentioned above have put forward new requirements for database service providers: distributed databases need to have stronger performance, achieve multimodal data, hybrid search and real-time analysis, not only support numerous AI agents to provide better financial service experience, but also assist generative AI in activating the value of these data.

At the end of last year, Wu Peng's rural commercial bank established an enterprise level knowledge base and provided remote banking customer service through the knowledge base. But he found that once there is a business continuity interruption in the knowledge base, it will directly affect the financial service capabilities of the entire bank. Therefore, he is communicating with two leading database vendors in the industry on how to build a more stable and vertical business knowledge base through the integration of generative AI and databases.

Faced with these new demands, Feng Yu is somewhat powerless. In recent years, the decline in orders for distributed database services has led to a lack of sufficient funding for Feng Yu's database vendor to research how to promote the integration of generative AI and distributed databases.

Feng Yu has collaborated with two AI technology research and development companies to promote "database AI" through the latter's AI technology. But he soon discovered that data fragmentation (data scattered in various business systems of the bank forming data islands), lack of real-time data analysis foundation (transactions and analysis are not in the same database), serious AI illusion situation, and high security compliance requirements (unstructured data governance security system is not sound enough) have become the four major obstacles to promoting database AI. In the end, after a brief period of ineffective cooperation, Feng Yu "peacefully broke up" with two AI technology research and development companies.

Feng Yu stated that over half of the more than 60 small and medium-sized database vendors that withdrew from the market in the past year were constrained by a lack of ability to integrate generative AI with distributed databases.

Faced with the trend of the integration of generative AI and databases, OceanBase has recently launched the "Financial AI Integrated Data Solution", which integrates transaction, analysis, and multimodal data in a single engine, supports mixed retrieval of vectors, text, and other data, and builds data branching capabilities into distributed databases to enable AI to achieve secure exploration and iteration in isolated environments, resolving governance risks caused by data operations. Weng Rui revealed that this scheme has been implemented in internal innovation projects such as Alipay AI Pay.

Wang Nan stated that the integration of generative AI and distributed databases will bring about a new round of architecture revolution and paradigm shift. The former means that AI capabilities will be deeply integrated into the database kernel, further surpassing traditional database architectures in terms of efficiency, low latency, and security; The latter means that databases are evolving from passive storage to active understanding, and semantic understanding, similarity reasoning, and cross modal association will become the core capabilities of the new generation of databases. This will inevitably drive the distributed database of the financial industry to accelerate into a high-quality competition cycle, intensifying the reshuffle of the entire industry.

(At the request of the interviewee, Wu Peng and Feng Yu are given pseudonyms)