2023-06-27 15:10:06
Databricks has acquired MosaicML and its LLM model building platform for $1.3 billion. The objective is to allow customers to create or adjust models on their own data hosted on the Databricks platform. In partnership with Nvidia, Snowflake (competitor of Databricks) will offer a similar solution.
Leading competitors in the niche of cloud data management platforms, Databricks and Snowflake are positioned in the customization of large language models (LLM). Both are about enabling companies to develop custom generative AI applications using their own data.
To do this, Snowflake has announced a partnership with Nvidia. Customers will be able to create LLM models on NeMo, the platform Nvidia offers to train new models or train on pre-trained models. The procedure is carried out using data hosted securely on the Snowflake Data Cloud, underlines the Snowflake press release. The LLMs can thus be refined (fine-tuning) without moving the data.
Databricks buys MosaicML for $1.3 billion
For its part, Databricks will offer a similar service through the acquisition of MosaicML. According to the indications of the website of this Californian start-up, its platform makes it possible to train, pre-train or refine LLM models. With the Databricks Lakehouse and MosaicML technology, Databricks customers will have access to “a quick and easy way to maintain control, security and ownership of their valuable data without high costs”, highlights the official announcement. of the acquisition of MosaicML. The transaction is valued at approximately $1.3 billion.
It’s no surprise to see Databricks and Snowflake riding the wave of generative AI. The advent of large language models, such as the one nested at the heart of ChatGPT, has indeed boosted companies’ interest in data mining solutions for data scientists, showed a recent study by Databricks.
1687888749
#Databricks #Snowflake #companies #harness #data #GenAI