Snowflake and Databricks bank PostgreSQL acquisitions to bring transactions onto their platforms

Data analytics vendors have tried this before with limited success

In the past few weeks, both Snowflake and Databricks dipped into their respective acquisition funds and found the resources to buy specialist providers of PostgreSQL database systems.

In May, Databricks, a data lake maker originally built around Apache Spark, bought Neon for a reported $1 billion in equity. Neon, which builds serverless PostgreSQL services, has claimed 80 percent of the databases it provisions were created automatically by AI agents rather than by humans.

Then at the start of this month, cloud data warehouse biz Snowflake, which branched out into data lakes and transactional systems, paid $250 million to buy Crunchy Data. Crunchy Data provides PostgreSQL clusters for transactional and analytics systems spanning managed cloud services, Kubernetes deployments and on-premises solutions. Snowflake said it will use the purchase to create "Snowflake Postgres," to run any Postgres-dependent application directly on its data and analytics platform.

Databricks argued that putting Neon's serverless PostgreSQL architecture on the Databricks Data Intelligence Platform would help developers and enterprise teams efficiently build and deploy AI agent systems.

Speaking to the media, Databricks CEO and co-founder Ali Ghodsi said the plan with the acquisition was not just to appeal to the startups who already use Neon, but also to win over enterprise customers.

"About 70 percent of our customers… have legacy databases that they want to swap out, and they're saying it's too expensive, it's stagnated… not much has happened in the last 20 years in the old database transaction processing area. [They] would love to replace it with something that's modern and fit for the AI era. We spent our... last six months talking to those enterprises, really sussing out: would this [acquisition] be a fit for them. It turned out that they were all super excited," he claimed.

Henry Cook, Gartner senior director analyst for IT leaders, said both Snowflake and Databricks have been known as vendors of analytical systems, so their buys might help them into the operational and transactional part of the market – "over time."

"There is a trend towards greater integration between transactional and analytic systems, where the transactional systems feed the analytical systems and the analytical systems can provide analytical feedback for delivery by transactional systems, whether in real time or near real time. These moves could also be interpreted as a facet of that trend too," he said.

The idea is to allow users to implement applications that are a mix of analytical and transactional operations, either pure analytical or structured analytical, he said.

"It will also allow them to define mixed transactional and analytical applications in a cloud-independent manner, in that longer term providing a complete portable environment which will be attractive to many organizations," Cook said.

For the PostgreSQL community it would be a boon, because it offers wider choice and support for the open source database.

Gartner senior research director and analyst for the database market, Robin Schumacher, said data warehouse and analytic vendors who have tried to enter the operational/transactional DBMS market have mostly failed.

"Teradata and Vertica are examples. Snowflake introduced Unistore in hopes of doing the same, but has seen little interest from what I've seen. Part of the reason for these vendors falling short is DBMS users want a trusted platform for their operational systems and adding new, unproven transactional support in a previously analytic-only DBMS wasn't appealing to the market. But DBMS users know and trust PostgreSQL, so both Snowflake and Databricks stand a chance of bucking that historic trend with their purchases," he said. ®