Rockset acquisition strengthens OpenAI's enterprise play
The GenAI pioneer went outside the AI realm for its second acquisition, looking to the smaller database analytics vendor for its real-time search and vector capabilities.
OpenAI's purchase of database analytics startup Rockset indicates the generative AI pioneer is looking to strengthen its enterprise offerings with real-time data and model fine-tuning capabilities.
The acquisition, made public on June 21, gives the GenAI vendor Rockset's cloud-native database technology to boost its models' ability to retrieve and generate more up-to-date information.
Those capabilities are critical for OpenAI to broaden the scope of its GPT LLM series with LLM features enterprises want, said Kashyap Kompella, CEO of RPA2AI Research.
"With capabilities for data ingestion, processing, search, and analytics and support for SQL, No SQL, and vector databases, Rockset plugs gaps in OpenAI technology and can become more suitable and appealing for enterprise use cases," Kompella said.
OpenAI's models are now largely pre-trained and not up to date. For example, the latest knowledge update for OpenAI's latest commercial LLM, GPT-4 Turbo, was in April 2023.
"For many enterprise use cases, if the AI response is based on pre-trained data, it just won't do," Kompella said. "The AI responses must contain timely, updated, and often, real-time information for them to be useful."
Two startups
Founded in 2016 by two former Facebook engineers, Rockset develops a database that includes indexing, SQL support, built-in connectors for data ingestion and developer tools. Most recently, the vendor added vector embedding support that lets users search and analyze unstructured data.
Rockset had raised a total of $105 million, most recently pulling in $44 million in August 2023. Neither company disclosed the value of the acquisition, but the all-stock transaction was reportedly a nine-figure deal.
Members of Rockset's team will join OpenAI, the company said in a blog post.
OpenAI, which started as a nonprofit research lab in 2015, has grown explosively amid the surge in GenAI technology and is now a for-profit company valued at more than $80 billion.
While products from the GenAI vendor, which also offers the Dall-E image-generating model series as well as multimodal models GPT-4o and Sora, have proved popular among consumers, the company has yet to establish a firm enterprise customer base.
"The biggest value for OpenAI is the potential to revolutionize data indexing and querying within their own platform," said Mike Leone, an analyst at TechTarget's Enterprise Strategy Group. "It means OpenAI will be able to improve trust in responses through faster access to larger amounts of data."
Meanwhile, some see OpenAI's move as a sign of a budding trend that could reverse the current paradigm of larger tech vendors investing in and acquiring smaller AI startups.
"This is the start of a trend. The data companies are moving from data to AI -- Snowflake, Databricks, Oracle, Google. Now what we're seeing is that the AI companies are moving toward data," said Sanjeev Mohan, principal at SanjMo, a data trends advisory firm. "At some point data, analytics and AI are all going to converge, so we are going to see more of these."
Rockset customers
In a few years, Rockset built a roster of well-known users, including Meta, payments vendor Klarna and JetBlue, according to the vendor.
Those enterprises will now have to find a new streaming database supplier, such as Materialize or SingleStore, and soon. In an FAQ on Rockset's site, the company said "customers paying month-to-month or in arrears will have until Monday, September 30th, 2024, 5 PM PDT to off-board."
For many enterprise use cases, if the AI response is based on pre-trained data, it just won't do.
Kashyap KompellaCEO, RPA2AI Research
"I don't think this is a good deal for Rockset customers. Rockset customers are finished," Mohan said. "They have to find a different database and migrate off of Rockset. It makes me sad that all this technology that was built by Rockset is now going away."
But for enterprises looking for a GenAI provider, the Rockset technology could help sway a decision toward OpenAI.
Despite its wide name recognition and innovative brand reputation, OpenAI is still mostly a consumer-facing company. Meanwhile, AI rivals such as Google; AWS; Meta; and even Microsoft, OpenAI's own partner, have much bigger enterprise footprints.
"OpenAI clearly understands that the next wave of GenAI innovation focuses on domain-specific enterprise data," said Kevin Petrie, an analyst at BARC U.S. "Companies need to organize their multi-structured data … in ways that language models and AI models can understand very quickly. Rockset provides these capabilities."
Senior news writer Eric Avidon contributed to this story.
Shaun Sutner is senior news director for TechTarget Editorial's information management team, driving coverage of artificial intelligence, unified communications, analytics and data management technologies. He is a veteran journalist with more than 30 years of news experience.
