Databricks has announced a definitive agreement to acquire MosaicML, a pioneer in large-scale language models (LLM).
This strategic move aims to make generative AI accessible to organizations of all sizes, allowing them to develop, own, and secure their own generative AI models using their own data. .
The acquisition is valued at approximately $1.3 billion (including retention packages) and demonstrates Databricks' commitment to democratizing AI and strengthening its Lakehouse platform as the leading environment for building generative AI and LLMs. I am.
Naveen Rao, co-founder and CEO of MosaicML, said:
“At MosaicML, we believe in a world where anyone can add their voice and perspective to building and training their own models. Working with Databricks allows us to make that belief a reality.
We started MosaicML to solve the tough engineering and research problems needed to make training at scale more accessible to everyone. The recent wave of generative AI has put this mission front and center.
Together with Databricks, we will tip the scales in the best interest of the many people. And we do it as fellow researchers-turned-entrepreneurs who share the same mission. We look forward to continuing this journey with the AI community. ”
MosaicML has gained recognition for its cutting-edge MPT large-scale language model, with millions of downloads for MPT-7B and the recently released MPT-30B.
The platform has demonstrated how organizations can leverage their own data to quickly build and train their own cutting-edge models cost-effectively. Prominent customers such as AI2, General Intelligent, Hippocratic AI, Replit, and Scatter Labs leverage his MosaicML for a variety of generative AI applications.
The primary goal of this acquisition is to provide organizations with a simple and fast way to develop, own and protect their models. By combining the power of Databricks' Lakehouse platform with MosaicML's technology, customers can maintain control, security, and ownership of their valuable data without incurring prohibitive costs.
MosaicML's automatic model training optimization enables 2x to 7x faster training compared to standard approaches, and near-linear scaling of resources allows you to train models with billions of parameters within hours. can. As a result, Databricks and MosaicML aim to reduce the cost of training and utilizing LLMs from millions of dollars to thousands of dollars.
Combining Databricks' unified data and AI platform with MosaicML's generative AI training capabilities creates a robust and flexible platform that can serve the largest organizations and address a variety of AI use cases.
Once the transaction closes, the entire MosaicML team, including its renowned research team, will join Databricks.
MosaicML's machine learning and neural network experts are at the forefront of AI research, working to improve model training efficiency. They have contributed popular open source base models like MPT-30B and training algorithms that power MosaicML's products.
The MosaicML platform will be supported, expanded, and integrated over time to provide customers with a seamless, integrated platform to build, own, and protect their generative AI models. The Databricks and MosaicML partnership gives customers the freedom to build their own models, train them with their own data, and develop differentiated intellectual property for their businesses.
Completion of the proposed acquisition is subject to customary closing conditions, including regulatory approvals.
(Photo by Glen Carrie on Unsplash)
See also: MosaicML's latest model outperforms GPT-3 with just 30B parameters
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expos in Amsterdam, California, and London. This event coincides with Digital Transformation Week.
