The product incorporates technology from MosaicML, acquired by Databricks for $1.3 billion in June.
By Kenrick CaiForbes team
Databricks made the largest acquisition of the generative AI era to date when it bought MosaicML in June for $1.3 billion. This is a nearly 6x markup from MosaicML’s previous valuation of $222 million – a “bargain,” if you ask Ali Ghodsi, CEO of Databricks: “Looking back, I ‘would have paid even more. »
On Wednesday, Databricks announced a new product that Ghodsi hopes will confirm his sentiment. His company became one of the world’s hottest business technology companies, last valued at $43 billion in September, by selling its flagship data software “Lakehouse” to companies looking for a way to store and analyze data both structured (think: bank transaction spreadsheets) and unstructured (like images and raw text files). The new offering, called the Data Intelligence Platform, will inject MosaicML’s generative AI expertise into the Lakehouse.
“I think of the lake house as a little platform. This is the foundation, and this is the skyscraper on top,” Ghodsi said. Forbes.
Now, instead of needing to know coding languages like Python to analyze a company’s data, employees will be able to ask questions in plain English to get insights about the data, as if they were using a version ChatGPT primitive – allowing many more users to make use of the data. This is powered by the large language models of MosaicML, the underlying technology that has fueled the current AI boom. Customers can train these custom models using proprietary datasets they have stored in Databricks.
“I have no doubt that this is the future of all data platforms: Databricks, Snowflake, [Google’s] BigQuery and so on.
Among the beta testers, Ghodsi said Databricks is seeing many CEOs and executives using the new product to answer questions they would otherwise have had to ask technical staff to uncover. At Tufts Medicine, some doctors use the tool to review patient data. “It allows many more people within the organization to get insights and do the type of work that only data scientists could do before,” Ghodsi said. “I have no doubt that this is the future of all data platforms: Databricks, Snowflake, [Google’s] BigQuery and so on.
Like Databricks, ranked #2 on ForbesThe Cloud 100 list prepares for a highly anticipated IPO, it faces a two-front war against its longtime data platform rivals (Snowflake, its closest parallel, trades on the public market with a market cap of $55 billion) and the new crop of AI model providers led by OpenAI (#1 on Cloud 100) and Anthropic (#73). Databricks announced that its revenue exceeded $1.5 billion in September; OpenAI is quickly catching up, with $1.3 billion, according to an October report in Information.
With ChatGPT, OpenAI has lowered the bar for consumers to access AI, and its revenue has so far reportedly been dominated by chatbot subscriptions. But more and more companies are leveraging large models like GPT-4 or Anthropic’s Claude 2 to build their own AI applications, like productivity tool Notion which earlier this week released a new feature based on AI. Databricks helps other customers approach the problem from a different angle: instead of using one massive model, it helps them create smaller, tailored models.
“What’s getting the most interest is people who have very sensitive data and want to create their own AI,” Ghodsi said. “We help them do it.” Databricks is already at capacity with the 15,000 GPUs it leases for these purposes, Ghodsi said, which means some customers will have to wait. The new product, which lowers the bar for workers to access data on top of these custom AI models, represents another step in Databricks’ quest to carve out a niche in the rapidly evolving field of AI .
However, there is still a learning curve to use the product. When Ghodsi demonstrated it to Forbes, two queries returned the message “Oh no!” An internal error occurred. Databricks then clarified that a sample dataset was used for the demo and that a user needed to be knowledgeable about the topic to ask the right questions that would return useful answers. So the product probably won’t make data scientists obsolete anytime soon. “I think they will continue to exist and become even more important because they are the experts who can fix the errors that you see,” Ghodsi says.
MORE FROM FORBES