[2/2] Databases in 2025: A Year in Review by Andy Pavlo (Category Databases)
Continue. story About the overview of the database market from Andy Pavlo, I will tell about the remaining theses.
ия Money, mergers and acquisitions 2025He was rich in big deals in the data world. The hunt for PostgreSQL continued: Neon went to Databricks, Crunchy Data went to Snowflake. ($250 million) IBM did not stand by and took DataStax. (Cassandra) roughly3 billion Salesforce Spent on Informatica’s Oldest ETL Platform (~$8 billion)
- Even MariaDB Corp. She managed to buy herself twice in the same year, returning the breakaway cloud daughter SkySQL to her wing. The surprise of the year was the merger of Fivetran and dbt Labs – two complementary companies merged into an ETL colossus, aiming for an imminent IPO.
Investors cooled to the new DBMS Hype around the vector bases slept, ventures in 2025 They invested less in young DB startups, preferring to carry money in AI. But industry leaders continue to get huge rounds:
- Databricks raised $5 billion (two-round) yearly
- ClickHouse – $350 million
- Supabase - Total $300 million (Series D + E) And even very young projects like SpiralDB. (Vortex creators) They raise tens of millions.
Not everyone survived. A number of companies did not survive the year. FaunaDB – an ambitious distributed DBMS startup with strong consistency (Pavlo was their technical consultant and advised to add SQL, but he was not listened to.)
- PostgresML project (Machine learning inside PostgreSQL) Didn't take off either.
- Team Hydra. (connector DuckDBPo️Postgres) Who went where?
- Even the classic: Apache Derby (Java-DBMS 1997 year) "Read only" is announced - development stopped due to lack of Maintainers. GPU databases also suffered a difficult fate. Startup supergroup Voltron Data with an investment of $110M did not have time to release its GPU-accelerated engine Theseus
- HeavyDB (OmniSci/MapD, pioneer of GPU-DBMS) Quietly ceased to exist independently – the team was absorbed by Nvidia Andy Pavlo summarizes: niche solutions like Kinetica and Sqream are still somewhere on the periphery, and the classic CPU-oriented GPU database solutions have not yet been able to squeeze significantly. The performance of modern column systems on the CPU is already so high that not raw optimizations are crucial, but the convenience for developers and the “smartness” of the query optimizer. However, Pavlo suggests that the 2026We will hear about new attempts to use the GPU at full capacity – the race continues.
Larry Ellison is the pinnacle of success For the founder of Oracle, this year was truly a triumph. In 81 Larry Ellison topped the list of the richest people on the planet for the first time “crowned” as the richest person in the world (condition estimated at about $393 billion). Oracle soared in value thanks to cloud deals in the field of AI, and in September 2025 Oracle shares have risen 40Percent a day, making Ellison not only No.1 in the world, but also the richest man in history. (Inflation-adjusted beat even Rockefeller). After a couple of months, the price of Oracle rolled back and Larry lost about $.130 Billions – on the scale of a billionaire is like for us to lose one salary, jokes Pavlo. Within a year, Oracle/Ellison made major headlines: Oracle was a key partner in the deal to “Americanize” TikTok and financed the attempted takeover of Warner Bros. (through the family-owned Paramount). Andy Pavlo admits that he was even happy to hear the news that “databases have made someone the richest person on earth” – something positive has finally happened. He mocks Ellison’s critics: Larry has already conquered the corporate database world, won regattas and built a luxury spa network for the techno-rich, so why not buy a news channel? Allison doesn’t pay attention to conversations, he does what he wants – the fans and his new wife support him, and the rest doesn’t matter.
#Database #Architecture #Management #DistributedSystems #Software #Engineering #Future