The US$38 billion San Francisco software company Databricks has inked an agreement to acquire the city’s artificial intelligence (AI) startup MosaicML for US$1.3 billion.
The deal announced on Monday is aimed at creating a leading AI platform that can provide companies and organizations with the resources needed to develop their own large language AI models similar to the well-known ChatGPT at an affordable cost while ensuring the security of their data. The transaction represents approximately US$21 million for each of the 62 employees working for MosaicML.
The merger is intended to combine the capabilities of Databricks’ “Lakehouse” AI technology with MosaicML’s large language model (LLM) platforms like the MosaicML Pretrained Transformer (MPT)-7B and more recently released MPT-30B technology. Databricks says MPT-7B has had over 3.3 million downloads from customers such as the Allen Institute for AI, Scatter Lab and Generally Intelligent.
The computing costs associated with developing ChatGPT have been described by OpenAI’s CEO as “eye-watering.” Databricks and MosaicML say training and using LLMs through utilization of the companies’ combined software will cost thousands rather than millions.
“Every organization should be able to benefit from the AI revolution with more control over how their data is used. Databricks and MosaicML have an incredible opportunity to democratize AI and make the Lakehouse the best place to build generative AI and LLMs,” said Ali Ghodsi, CEO of Databricks.
.@alighodsi joined @jonfortt @CNBCClosingBell to discuss Databricks’ agreement to acquire @MosaicML. Watch now to hear what this announcement means + Ali’s take on the future of #generativeAI.
— Databricks (@databricks) June 26, 2023
Read more: Huawei’s Tech4Nature initiative launches new phase to boost coral reef conservation using AI
Read more: DiagnaMed Holdings launches medical chatbot powered by ChatGPT
Transaction is yet to close and subject to certain conditions and regulatory clearances
Databricks says its data and AI technology can provide a platform robust enough to cater to the world’s largest organizations when combined with MosaicML’s generative AI training capabilities. The company believes the combined tech will be flexible enough to address a broad range of cases where the use of AI is deemed necessary.
Databricks says MosaicML’s machine learning and neural networks experts engage in pioneering AI research to improve model training efficiency.
“We started MosaicML to solve the hard engineering and research problems necessary to make large-scale training more accessible to everyone. With the recent generative AI wave, this mission has taken centre stage,” said MosaicML’s CEO Naveen Rao.
Databricks opened a new office in Herzliya, Israel at the end of May and made the AI 50 list from Forbes in 2022.
rowan@mugglehead.com
