The significance of databases to trendy enterprise AI operations can’t be overstated.
Information helps to coach and floor AI, and a number of analysis studies present that with out correct information, AI efforts are inclined to fail. With developments like vibe coding and agentic AI, it’s additionally more and more necessary to have database know-how that may be spun up as wanted in a serverless method to trendy improvement efforts.
In that setting, it ought to come as no shock that databases are a very invaluable commodity.
This week, that truth was on show with Databricks‘ acquisition of privately held serverless PostgreSQL startup Neon, which was based in 2022. The deal is reportedly valued at a staggering $1 billion, which is stunning given that hardly two years in the past, the corporate raised $46 million in a collection B spherical of funding.
What can be significantly fascinating is that Databricks itself is an information vendor, with its information lakehouse platform. At varied factors within the firm’s historical past, it has positioned itself as an alternative choice to conventional databases, offering an information lake construction the place customers could make queries. So what was lacking, and why did Databricks have to spend a billion {dollars}? What does it imply and say about what enterprise AI actually wants?
Functionally, it’s all about assembly builders’ must construct agentic AI functions. In response to Neon, over 80% of the databases created on its platform had been created by AI brokers.
What’s serverless PostgreSQL and why does it matter?
Whereas Neon is a startup, the core database know-how that it’s based mostly on shouldn’t be new.
PostgreSQL is among the oldest and most established open-source database platforms, relationship again to 1996. It’s a relational database know-how, which means it has tables and rows alongside extraordinarily steady options that organizations have trusted for many years. The core open-source PostgreSQL database is now up to date in a yearly launch cadence. The newest steady replace was PostgreSQL 17, which debuted in Sept. 2024.
As an open-source know-how, PostgreSQL has loved broad adoption and contributions. At one level, it was usually in comparison with different proprietary relational database choices, together with Oracle instead possibility. In 2025, although, PostgreSQL stands by itself.
DB-Engines at present ranks PostgreSQL because the fourth hottest database in use at this time, behind Microsoft SQL Server, MySQL and Oracle. The state of PostgreSQL 2024 report from Timescale identifies the open-source database’s rising prominence because the database of selection for AI functions. The database’s well-established and understood nature and broad availability are among the many quite a few components that make it engaging.
PostgreSQL by itself is simply the database, although. Working it as a serverless providing is an operational and deployment exercise. The promise of any serverless database is ease of operations. Quite than requiring a devoted database deployment that frequently runs with devoted assets, serverless is spun up on demand as wanted. It’s a deployment possibility that’s significantly engaging to builders as a strategy to construct functions rapidly. AI-based improvement is much more interesting as databases will be constructed and deployed programmatically.
The serverless PostgreSQL panorama has plenty of distributors
Each cloud hyperscaler has some type of PostgreSQL service and has for years.
Google has a number of choices, together with AlloyDB, Microsoft has Azure Database for PostgreSQL, whereas AWS has Amazon RDS for PostgreSQL and Amazon Aurora. Every of them additionally has some taste of serverless providing, that’s, a database accessible on demand.
Quite a few smaller distributors exist, together with Timescale, EDB and NetApp Instaclustr. In reality, almost two years in the past, Databricks acquired serverless PostgreSQL vendor bit.io, which was additionally an early rival of Neon.
Because it seems, the objectives and capabilities of bit.io are fairly totally different from Neon.
“Along with the Neon crew, we look ahead to constructing essentially the most developer and AI-agent-friendly database platform,” Phil Shin, senior director of company improvement and ventures at Databricks, instructed VentureBeat. “In distinction, the bit.io acquisition was not really about Postgres however concentrating on developer experiences, particularly within the trials and self-service course of.”
Shin added that the bit.io acquisition had a huge impact on Databricks’ seamless signup expertise.
How serverless PostgreSQL suits into the enterprise database panorama
Whereas Neon has solely been round for a number of years with its serverless PostgreSQL implementation, industrial vendor EDB has been in enterprise since 2004. EDB has a collection of its personal commercially supported PostgreSQL choices.
Matt Yonkovit, VP of Product for EDB, instructed VentureBeat that the acquisition of Neon is a powerful vote of confidence in Postgres as a foundational know-how for AI and analytics.
“It reinforces what we’ve lengthy believed: Postgres is more and more central to trendy information stacks,” Yonkovit mentioned. “Serverless is a good match for dev/take a look at environments and for rapidly jumpstarting AI initiatives—however as these use circumstances scale, enterprises want the efficiency, compliance, and management of a sovereign platform.”
Yonkovit famous that serverless is well-suited for brief bursts and smaller workloads. It may possibly scale up and down rapidly or shut off totally when idle, which considerably reduces prices related to compute, energy and storage. Nonetheless, in his view, there are tradeoffs.
“A major problem with serverless is that sovereign information administration can change into messy as a result of you possibly can’t management the place the info is processed until you’ve a well-restricted pool of assets,” Yonkovit mentioned.
The facility of serverless PostgreSQL for agentic AI
Neon’s serverless PostgreSQL method separates storage and compute, making it developer-friendly and AI-native. It additionally permits automated scaling in addition to branching in an method that’s much like how the Git model management system works for code.
Amalgam Insights CEO and Chief Analyst Hyoun Park famous that Databricks has been a pioneer in deploying and scaling AI initiatives.
“One of many huge challenges in AI is managing the storage and compute related to the info,” Park instructed VentureBeat. “Every of those wants shall be more and more exhausting to foretell in an agentic world the place end-user prompts and requests could rapidly result in sudden calls for in storage or compute to resolve the issue.
Park defined that Neon’s serverless autoscaling method to PostgreSQL is necessary for AI as a result of it permits brokers and AI initiatives to develop as wanted with out artificially coupling storage and compute wants collectively. He added that for Databricks, that is helpful each for agentic use circumstances and for supporting the customized fashions they’ve constructed during the last couple of years after its Mosaic AI acquisition.
What it means for enterprise AI leaders
For enterprises trying to cleared the path in AI, this acquisition indicators a shift in infrastructure necessities for profitable AI implementation.
Information is crucial for AI; that’s not a shock. What is especially insightful, although, is that the power to quickly spin up databases is important for agentic AI success. The deal validates that even superior information corporations want specialised serverless database capabilities to help AI brokers that create and handle databases programmatically.
Organizations ought to acknowledge that conventional database approaches could restrict their AI initiatives, whereas versatile, immediately scalable serverless options allow the dynamic useful resource allocation that trendy AI functions demand.
For corporations nonetheless planning their AI roadmap, this acquisition indicators that database infrastructure choices ought to prioritize serverless capabilities that may adapt rapidly to unpredictable AI workloads. This may rework database technique from a technical consideration to a aggressive benefit in delivering responsive, environment friendly AI options.
