TiDB has expanded its cloud platform with new database and storage services for enterprise and AI applications. The move follows the creation of nearly 3 million database instances on TiDB Cloud Starter by AI agents in the past six months.
The additions include managed databases running in customers' own cloud accounts, persistent storage for AI agents and PostgreSQL support on TiDB Cloud Starter. PingCAP, the company behind TiDB, also plans to open source the core code of TiDB X, the architecture used in TiDB Cloud, by the end of 2026.
The announcement reflects a shift in how database suppliers are positioning themselves as businesses adopt AI systems that generate software and carry out tasks with limited human input. Those systems are creating large numbers of database instances and filesystems, raising questions about cost, operational control and how to retain an agent's work between sessions.
TiDB is adapting its cloud platform to support both conventional production workloads and newer agent-based applications on the same underlying data system. It says the approach draws on work with large enterprises, AI model developers and agent platform operators.
Cloud options
Among the main additions is Bring Your Own Cloud, or BYOC, now in public preview. The service is designed to let customers use TiDB Cloud's managed database operations within their own cloud accounts while retaining control of their data environments.
TiDB Cloud Filesystem has also entered public preview. It gives AI agents a persistent workspace, allowing code, context files and other saved work to remain available when an agent starts a new session in a new sandbox.
PostgreSQL support on TiDB Cloud Starter is also now in public preview. This extends the service to teams working in the PostgreSQL ecosystem, alongside TiDB's established MySQL compatibility.
Another part of the expansion is TiDB Cloud Premium, scheduled for general availability in November. The service is intended for business-critical applications and includes automatic scaling, isolated resources and additional security features.
Enterprise use
TiDB pointed to customer deployments to illustrate the scale of workloads now being placed on the platform. For its Forge SQL hosted relational database service, Atlassian consolidated more than 750 PostgreSQL clusters into 16 TiDB clusters supporting more than 3 million tables.
Elsewhere, Manus runs more than 1.5 million TiDB Cloud clusters to support applications created by its agents. Moonshot AI, the company behind Kimi, uses more than one million persistent filesystems for coding agents and has cut its data infrastructure costs by an order of magnitude with TiDB Cloud.
TiDB also cited Pinterest as an example of a more traditional application migration. According to the company, Pinterest used the database to simplify its data infrastructure, retire several in-house systems and reduce maintenance overhead, with infrastructure cost savings of around 50%.
Open-source plan
The planned release of the core code of TiDB X extends the group's open-source model to the architecture behind its cloud service. That could matter for customers seeking greater visibility into the technology they rely on, particularly as database systems become more central to both internal applications and AI-driven services.
Database vendors have increasingly tried to appeal to organisations that want managed services without giving up control over deployment or portability. TiDB's emphasis on BYOC and open-sourcing TiDB X suggests it is trying to balance those demands while broadening its reach beyond its existing MySQL base.
For businesses experimenting with agent-based software, persistent storage may prove one of the more practical additions. Many AI agents operate in temporary computing environments, so they need a way to retain code, files and context if they are to continue tasks across multiple sessions.
That requirement differs from the needs of standard transactional systems, but database and storage providers are increasingly treating the two as connected. TiDB argues that a shared data foundation can support both established enterprise applications and the systems being created by AI agents.
"Every team should be able to start small on a data foundation that can grow with its ambitions," said Max Liu, Co-Founder and Chief Executive Officer at TiDB. "TiDB has been battle-tested at extraordinary scale by large enterprises and frontier AI pioneers. We're bringing the capabilities shaped by that experience into a managed cloud platform, so more organizations can scale on the same foundation without having to build the specialized teams and manage the infrastructure complexity behind it."