Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More
The significance of databases in the realm of enterprise AI operations is paramount. Data serves as the backbone for training and grounding AI systems, and numerous research studies indicate that without a robust data foundation, AI initiatives are likely to falter. As trends such as vibe coding and agentic AI gain traction, the demand for agile database technologies that can be deployed in a serverless manner becomes increasingly critical.
This week, the value of databases was underscored by Databricks’ acquisition of the privately held serverless PostgreSQL startup Neon, founded in 2022. The deal, reportedly valued at an impressive billion, is particularly noteworthy considering Neon had raised only million in a Series B funding round just two years prior.
What adds another layer of intrigue is Databricks’ own identity as a data vendor, known for its data lakehouse platform. Historically, the company has positioned itself as an alternative to traditional databases, offering a data lake structure for user queries. This raises questions about what gaps Databricks aimed to fill with this billion-dollar investment and what it reveals about the evolving needs of enterprise AI.
At its core, this acquisition is about equipping developers with the tools necessary to create agentic AI applications. Neon reports that over 80% of the databases generated on its platform were initiated by AI agents.
What is serverless PostgreSQL and why does it matter?
While Neon may be a newcomer, the underlying technology of PostgreSQL is well-established, with roots tracing back to 1996. As a relational database platform, PostgreSQL features tables and rows, alongside a suite of stable functionalities that organizations have relied upon for decades. The open-source PostgreSQL database benefits from regular updates, with the latest stable version, PostgreSQL 17, released in September 2024.
The open-source nature of PostgreSQL has fostered widespread adoption and contributions, positioning it as a viable alternative to proprietary relational databases like Oracle. As of 2025, PostgreSQL stands tall, currently ranked as the fourth most popular database according to DB-Engines, trailing only Microsoft SQL Server, MySQL, and Oracle. The state of PostgreSQL 2024 report from Timescale highlights the database’s growing prominence as the preferred choice for AI applications, attributed to its reliability and accessibility.
However, PostgreSQL alone is merely a database; its operation as a serverless offering transforms its deployment dynamics. Serverless databases promise operational simplicity, allowing resources to be allocated on-demand rather than requiring a continuously running dedicated deployment. This flexibility is particularly appealing to developers aiming to expedite application development, especially in AI-centric projects where databases can be created and deployed programmatically.
The serverless PostgreSQL landscape has a lot of vendors
Every major cloud provider has introduced some form of PostgreSQL service over the years. Google offers multiple solutions, including AlloyDB, while Microsoft provides Azure Database for PostgreSQL, and AWS features Amazon RDS for PostgreSQL and Amazon Aurora. Each of these platforms also includes a serverless variant, providing databases on demand.
In addition to these giants, several smaller players contribute to the ecosystem, such as Timescale, EDB, and NetApp Instaclustr. Notably, Databricks previously acquired bit.io, another serverless PostgreSQL vendor and an early competitor of Neon.
However, the objectives and functionalities of bit.io differ significantly from those of Neon. “Together with the Neon team, we look forward to building the most developer and AI-agent-friendly database platform,” stated Phil Shin, senior director of corporate development and ventures at Databricks. He emphasized that the bit.io acquisition focused more on enhancing developer experiences, particularly in trial and self-service processes, rather than on PostgreSQL itself.
How serverless PostgreSQL fits into the enterprise database landscape
Despite its recent entry into the market, Neon’s serverless PostgreSQL implementation stands alongside established vendors like EDB, which has been operational since 2004 and offers a range of commercially supported PostgreSQL solutions.
Matt Yonkovit, VP of Product for EDB, remarked that Databricks’ acquisition of Neon signifies a strong endorsement of PostgreSQL as a foundational technology for AI and analytics. “It reinforces what we’ve long believed: Postgres is increasingly central to modern data stacks,” Yonkovit stated. He added that while serverless solutions are ideal for development and testing environments and can rapidly initiate AI projects, enterprises ultimately require the performance, compliance, and control that a sovereign platform provides as these use cases scale.
Yonkovit acknowledged that serverless architectures excel in handling short bursts and smaller workloads, offering rapid scalability and the ability to shut down entirely when idle, thus reducing costs related to compute, power, and storage. However, he also noted potential drawbacks. “A significant challenge with serverless is that sovereign data management can become messy because you can’t control where the data is processed unless you have a well-restricted pool of resources,” he cautioned.
The power of serverless PostgreSQL for agentic AI
Neon’s serverless PostgreSQL model distinguishes between storage and compute, enhancing its appeal for developers and aligning with AI-native requirements. This architecture facilitates automated scaling and branching, akin to the functionality of the Git version control system for code.
Hyoun Park, CEO and Chief Analyst at Amalgam Insights, highlighted Databricks’ pioneering role in deploying and scaling AI initiatives. “One of the big challenges in AI is managing the storage and compute associated with the data,” Park explained. “Each of these needs will be increasingly hard to predict in an agentic world where end-user prompts and requests may quickly lead to unexpected demands in storage or compute to solve the problem.”
Park emphasized that Neon’s serverless autoscaling approach is crucial for AI, as it allows projects to expand without the constraints of tightly coupled storage and compute resources. This capability is particularly beneficial for Databricks, supporting both agentic use cases and the custom models developed following its Mosaic AI acquisition.
What it means for enterprise AI leaders
For enterprises striving to excel in AI, this acquisition signals a pivotal shift in the infrastructure requirements necessary for successful AI implementation.
While the critical role of data in AI is well understood, the ability to swiftly deploy databases emerges as a fundamental requirement for agentic AI success. This acquisition underscores that even advanced data companies recognize the necessity for specialized serverless database capabilities to support AI agents that can create and manage databases programmatically.
Organizations must acknowledge that traditional database strategies may hinder their AI initiatives, while flexible, instantly scalable serverless solutions offer the dynamic resource allocation essential for modern AI applications. For companies still charting their AI roadmaps, this acquisition serves as a clear indication that database infrastructure decisions should prioritize serverless capabilities, enabling rapid adaptation to unpredictable AI workloads. Such a shift could redefine database strategy from a mere technical consideration to a competitive advantage in delivering responsive and efficient AI solutions.
