Databricks, a leading provider of AI software and cloud services, announced on Monday that it has entered into a definitive agreement to acquire MosaicML, a generative AI platform that enables users to train and deploy custom AI models with their own data. The deal will cost Databricks $1.3 billion.
MosaicML was founded in 2021 by a team of AI experts and researchers from MIT, Stanford, and Intel. The company's mission is to democratize generative AI, a branch of AI that can create novel content such as text, images, audio, and video. MosaicML offers a suite of products that simplify the development and deployment of generative AI models.
Databricks said that the acquisition will enhance its AI capabilities and allow it to offer a comprehensive generative AI platform for enterprises and developers. By combining Databricks' data lakehouse architecture, unified analytics engine, and open-source frameworks with MosaicML's generative AI software infrastructure, model training, and model deployment solutions, Databricks said it will enable customers to build, own, and secure generative AI models with their own data without high costs or complexity.
The acquisition comes at a time when generative AI is gaining momentum and popularity across various industries and applications. Generative AI models can create realistic and diverse content that can be used for entertainment, education, marketing, research, and more. However, generative AI also poses significant challenges in terms of data privacy, security, ownership, quality, and ethics.
MosiacML Seeking to Compete Directly with OpenAI
MPT-30B is the latest addition to the MosaicML Foundation Series of models, which are open-source, commercially-licensed models that can be easily integrated into applications or fine-tuned on custom data. MPT-30B has 30 billion parameters, making it one of the largest language models available to the public. It was trained on a diverse corpus of text, including web pages, books, news articles, social media posts, and code.
According to MosaicML, MPT-30B achieves state-of-the-art results on several natural language processing benchmarks, such as GLUE, SuperGLUE, and SQuAD. It also demonstrates strong performance on generative tasks, such as text summarization, text completion, and code generation.
MosaicML claims that MPT-30B is more accessible and affordable than GPT-3, which is only available through a limited and expensive API. MosaicML allows users to download MPT-30B for free and run it on their own infrastructure, or use MosaicML's platform to train and deploy their own models.
Databricks said that by acquiring MosaicML, it will address these challenges and enable customers to leverage generative AI in a responsible and scalable way. The company said it will continue to support MosaicML's existing products and customers, as well as invest in developing new generative AI solutions and services.
Databricks was founded in 2013 by the original creators of Apache Spark, an open-source framework for large-scale data processing and machine learning. The company offers a cloud-based platform that combines data engineering, data science, machine learning, and analytics. Databricks has more than 5,000 customers across various industries.