Naveen Rao’s MosaicML To Be Bought By Databricks For $1.3 billion
MosaicML co-founders Hanlin Tang, left, Naveen Rao, Michael Carbin and Jonathan Frankle. Photo MosaicML.
June 29, 2023
Naveen Rao and his co-founders of MosaicML, along with investors in the start-up, are among the early big winners in the race to build Artificial Intelligence (AI) software, at least in financial terms. MosaicML announced today that it was being bought by Databricks for roughly $1.3 billion. Both AI companies are based in San Francisco.
“MosaicML’s expertise in generative AI software infrastructure, model training, and model deployment, combined with Databricks’ customer reach and engineering capacity, will allow us to tip the scales in the favor” of an explosion of custom AI models that are built by many developers and companies from every corner of the world, said Rao, 48-years-old.
AI models need to be trained using massive amounts of data. The volume of data, and the rising complexity of analyzing it, could cost companies millions of dollars for computing and storage. MosaicML is a generative AI platform that empowers enterprises and developers to cheaply build and control their own AI.
Customers include Dream3D whose AI-powered tool makes it easy for designers to create studio-quality 3D computer graphics. The New York based startup uses MosaicML’s platform to train its models in order to save time and cloud-computing costs.
Twelve Labs, based in San Francisco, provides tools to video creators and advertisers to quickly and cheaply search raw footage, a video. or a pool of videos to identify specific content. It uses MosaicML’s tool to automate the processes and manage its graphics processing unit system.
Databricks’ customers include Comcast, Conde Nast and more than half of the Fortune 500 companies who use it to organize data in order to use Artificial Intelligence (AI) tools and reduce the related high costs of cloud-based computing. Databricks current annual revenue run rate exceeds $1 billion, according to Bloomberg. It has a valuation of more than $30 billion.
Prior to founding MosaicML, Rao co-founded San Deigo based Nervana, a machine intelligence platform which allows the application of neural network learning to nearly any kind of data problem. The goal was to change “the way we think about computation,” Rao notes on his LinkedIn profile. “In the previous phase of computation, we built machines that crunch numbers. In the next phase of computation, we'll need machines that peer into large data sets and find useful insight. At Nervana we changed the world of hardware for AI.”
In 2016, Nervana was acquired by Intel for $400 million. Rao and co-founder Hanling Tang stayed on at Intel till 2020.
Rao is a member of the Board of several companies including Luminous Computing, based in Santa Clara, California, and Neu Reality, an Israeli data science company.
Earlier, from 2012 to 2014, Rao was a researcher at San Deigo based Qualcomm, the $129 billion chip manufacturer, investigating how biologically-inspired neural networks and artificial neural networks can be used to efficiently solve difficult computational problems.
In 2012, he designed algorithms to optimize daily trading strategies to minimize transaction costs at a firm in the Boston area. In 2011, he worked for five months as a post-doctoral research associate at Brown University.
From 2005 to 2008, prior to studying for his Ph.D., Rao worked at W&W Communications in Mountain View, California, working on algorithms, architecture, modeling, logic design, and logic verification. Earlier he was an engineer at Sun Microsystems (2004-2005; 1997-1998); Kealia (2002 to 2004); Caly Networks (1999-2002); and Teragen (1999).
In 1996, while a student at Duke University, he interned at IBM.
Rao earned a Ph.D. in Neuroscience from Brown University (2007-2011); studied for a Maters in Electrical Engineering at Stanford, but did not complete the degree (1997-1998); and earned a BSEE in Electrical Engineering and Computer Science from Duke University (1993-1997).
In 2020, Rao and Tang co-founded MosaicML. They teamed up with Michael Carbin, a professor, and Jonathan Frankle, a doctoral candidate, both at the Massachusetts Institute of Technology, who had published work on how to make machine learning more efficient.
The four co-founders realized that there is no single technology for reducing AI computing costs, but instead a mosaic - hence the company's name - of different approaches emerging among researchers, according to Reuters. MosaicML built tools that estimate the tradeoffs between the speed, cost and accuracy of using different kinds of computing hardware for training an AI model.
MosaicML’s technology, combined with the Databricks Platform, will offer customers a simple, fast way to retain control, security, and ownership over their valuable data without high costs, the companies state. Also, they add that training and using large language models, which power generative AI, will cost thousands of dollars, not millions.
MosaicML which has more than 60 employees, raised $37 million in venture funding from DCVC, Lux Capital and Playground Global. It was valued at $222 million at its last funding round in October 2021, according to CBInsights.
Following the completion of its acquisition by Databricks, expected later this year, investors in the start-up will have made at least a 600% gain in under two years. Co-founders Rao, Tang, Carbin and Frankle will likely make millions of dollars.
Rao describes himself, on his LinkedIn profile, as a “Neuroscientist, AI perpetrator, processor architect, entrepreneur, and investor. My career has been about how we can build synthetic intelligence and make it available everywhere.”
FOR MORE UNIQUE STORIES ON INDIANS AND INDIA:
For access to stories each week email: gitimescontact@gmail.com
or follow via LINKEDIN or TWITTER or FACEBOOK
(c) All rights reserved. Copyright under United States Laws
