Databricks has agreed to acquire artificial intelligence (AI) startup MosaicML in a mostly stock deal valued at US$1.3 billion ($1.95 billion), marking Databricks' latest efforts to build an ecosystem for enterprises to train their own AI models.
The deal comes at a time when AI development is surging in the tech industry, with rapid AI thrusts coming from big tech including Google and Microsoft, and venture capital investors pouring billions into funding AI startups in recent months.
MosaicML provides software tools designed to make it cheaper to carry out AI work, which often involves training AI algorithms on huge troves of data using expensive computer chips.
Databricks said the deal would combine its AI technology with MosaicML's language-model platform, creating a one-stop solution for enterprises to train, fine-tune and use its models.
"We think we can together make it very affordable for any organization to train their own models, own their own on their own private data and keep that private IP to themselves without having to use one centralised model," said Ali Ghodsi, chief executive at Databricks.
Databricks, which sells software tools for building AI systems, has been an advocate for open-source models, which it argues could rival the models players like OpenAI and Google are offering.
MosaicML has released multiple open-source foundation models it has pre-trained, which is the category of core technology behind services like OpenAI's ChatGPT.
Founded in 2021, MosaicML has raised US$64 million to date from investors including Lux Capital and DCVC.