The News: Databricks announced on June 26 that it is acquiring generative AI model-builder MosaicML for approximately $1.3 billion. MosaicML is known for its ability to build efficient AI models that cost less to train than many other large language models (LLMs). The company’s MPT LLMs, MPT-7B and MPT-30B, have been downloaded more than 3.3 million times. Customers include AI2 (Allen Institute for AI), Generally Intelligent, Hippocratic AI, Replit, and Scatter Labs.
Then on June 28 at the company’s Data + AI Summit in San Francisco, the company announced the launch of LakehouseIQ, a knowledge engine designed to enable technical or non-technical enterprise users to search, understand, and query their company’s data. According to the press release, Databricks envisions LakehouseIQ tacking the following issue: “…when employees need access to internal data to help complete their tasks, it can be difficult to quickly find and analyze the data they need…This bottleneck prevents businesses from truly embracing data and AI. Large language models promised to fix this problem, but so far, the results have been disappointing. General purpose models don’t understand the unique language of every business: they cannot process jargon or internal acronyms; they are not trained on the company’s unique data sets; and they do not understand organizational charts or know which teams should have access to what information.”
Read the full Press Release on MosaicML here.
Read the full Press Release on LakehouseIQ here.
Databricks’ MosaicML Acquisition, LakehouseIQ Launch, Data + AI Summit Show Gen AI Savvy
Analyst Take: Through a series of strategic moves and through their storytelling at their annual Data + AI Summit, data management and development platform player Databricks is proving to be very Gen AI-savvy and the company has the potential assume a leadership position in the Gen AI ecosystem, helping enterprises adopt Gen AI technology, applications, and use cases. Here’s why:
MosaicML Provides Evolved LLMs for Enterprise-Grade Uses
MosaicML’s products and evolved AI thinking are an excellent match for Databricks, with the potential to bring a more comprehensive and seamless approach to building generative AI outputs, particularly once the two companies’ platforms are integrated. The interest in LLMs has exploded with the debut of the ChatGPT interface, but LLMs have challenges, including training expense and accuracy issues. MosaicML is evolved LLM thinking. Their approach is to leverage LLMs for private domain data, specific to the enterprise using it. They are focused on model efficiency – MosaicML has developed faster and cheaper ways to train AI models. From the press release – “…automatic optimization of model training provides 2x-7x faster training compared to standard approaches. Combined with near linear scaling of resources, multi-billion-parameter models can be trained in hours, not days. With Databricks and MosaicML, training and using LLMs will cost thousands of dollars, not millions.” At the Data + AI Summit, MosaicML Co-Founder and CEO Naveen Rao shared a graphic which estimated an example of AI training costs for MPT-7B at $250,000.
LakehouseIQ Is a High-Potential Gen AI Use Case Delivering on the Promise of Big Data
AI leaders think and deliver on how to solve business and operational problems with AI. Databricks LakehouseIQ is a great example of this kind of leadership. The LakehouseIQ initiative feels like it is a pathway to the longstanding but unrealized vision of big data. The combination of the knowledge base (LakehouseIQ) and the interface (Assistant) with the data (Lakehouse) may unlock big data queries and analysis not only for technical professionals within the enterprise, but for business and operational users as well.
Databricks Delivered Great Gen AI Guidance at Data + AI Summit
At the Data + AI Summit, it was refreshing to hear a company talk plainly and knowledgeably about how enterprises should be thinking about AI. Databricks offered pragmatic guidance at a time when there are a lot of unknowns around generative AI.
Just two examples:
- In his keynote, Databricks CTO Matei Zaharia put up the following graphic: “Problem: Naively adding an LLM assistant doesn’t work.”
- During one keynote presentation, the speaker showed a graphic with the following text:
“Modeling techniques will quickly commoditize…Your data is your competitive edge.” This is contrary to a lot of the conventional messaging around LLMs, which is, the more data the better.
These examples give you an idea of the approach that was taken during the event – Databricks has the knowledge, culture, and vision to become a leader in helping enterprises navigate generative AI.
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Other insights from The Futurum Group:
Study: Generative AI Is Most Helpful to Less-Skilled Contact Center Workers
Facing the Unknowns of Generative AI, Key Industry Players Collaborate, Incubate, and Back Startups
Author Information
Based in Tampa, Florida, Mark is a veteran market research analyst with 25 years of experience interpreting technology business and holds a Bachelor of Science from the University of Florida.
