Advanced Computing in the Age of AI | Saturday, May 18, 2024

HPE Acquires Startup Determined AI to Help Customers Train AI Models More Quickly 

Making machine learning easier to use is all the rage in enterprise computing as more and more businesses are finding out that adopting and integrating AI is harder to do than they first anticipated.

HPE, like other large technology vendors, has been hearing that message from customers and in response, has acquired Determined AI, a San Francisco-based startup that built an open source machine learning training platform that is designed to make the process faster and simpler.

Determined AI’s platform, which is also cloud neutral, uses the Horovod deep learning distributed framework which works with TensorFlow, Keras, PyTorch and Apache MXNet to make distributed deep learning easier to use. The platform is a software stack that can train AI models faster and at any scale, while removing complexity and lowering the costs of machine learning development, according to Determined AI.

A price tag was not announced for the deal, which was unveiled on June 21 (Monday).

HPE says it will combine Determined AI’s platform with its own AI and high performance computing (HPC) products and services to speed up and expand the work of ML engineers as they create and train new models to gain improved insights from their data.

“As we enter the age of insight, our customers recognize the need to add machine learning to deliver better and faster answers from their data,” Justin Hotard, a senior vice president and general manager of HPE’s HPC and mission critical solutions division, said in a statement. “AI-powered technologies will play an increasingly critical role in turning data into readily available, actionable information to fuel this new era. Determined AI’s unique open source platform allows ML engineers to build models faster and deliver business value sooner without having to worry about the underlying infrastructure.”

By bringing in Determined AI’s technology, HPE says it will enable researchers and scientists to spend less time on configuring complex training models so they can spend more time focusing on innovations and accelerating their time to delivery. The platform is built to simplify set-up, configuration, management and sharing of workstations or AI clusters that are running on-premises or in the cloud, according to HPE.

In one case study, Determined AI helped drug discovery vendor, Recursion, cut its machine learning modeling from three days to three hours.

Under the acquisition, the Determined AI team will join HPE’s HPC and Mission Critical Solutions business group.

Founded in 2017, Determined AI launched its open source platform in 2020. The platform has been adopted by customers across a wide range of industries, including biopharmaceuticals, autonomous vehicles, defense contracting and manufacturing, according to the company.

In a blog post on the Determined AI website, CEO Evan Sparks, CTO Neil Conway and chief scientist Ameet Talwalkar wrote that the acquisition by HPE will accelerate the company’s mission and technology as it works to make ML less stressful for enterprises.

“Over the last several years, building AI applications has become extremely compute, data and communication intensive,” they wrote. “Ten years ago, you could do cutting-edge computer vision research on a laptop. Today, you need a massive farm of GPUs or other specialized chips to remain competitive. These problems are not unique to vision or academic research – they are affecting organizations large and small, as we are hearing from our thriving open source community.”

The problem, they wrote, is that AI is rapidly becoming an HPC problem, which makes the acquisition by HPE a smart move for Determined AI’s customers and technology, they wrote.

“HPE is already a global leader in designing and delivering high performance computing systems and, via their acquisitions of Cray and SGI, they have decades of experience in the space working with some of the most sophisticated users on the planet,” the post continues. “We are thrilled about the opportunity to partner with HPE to deliver co-designed software and hardware and tackle some of society’s most pressing challenges.”

A spokesperson for HPE could not be reached for comment by press time today.

Rob Enderle, analyst

Rob Enderle, principal analyst of Enderle Group, told EnterpriseAI that making ML easier for users is a hot topic right now because the process requires specialized skills and creates its own challenges.

“On paper, it appears to shorten the time needed to do AI training significantly, and AI training is one of the biggest impediments to AI implementation at the moment,” Enderle said of Determined AI’s platform. “AI and robotics are the two most promising segments to go vertical this decade, or grow extremely rapidly with the revenue line nearly going vertical, so HPE is ensuring they do not miss out on that opportunity.”

Making ML easier for businesses is a big topic in a crowded marketplace, he said.

“I think IBM is still way ahead with Watson, and they have been working on training for some time,” said Enderle. “Some of the best AI training is coming from Nvidia now. [HPE] will need to make this [acquisition] work, and HPE does not have an excellent reputation for doing company mergers well. Historically, they are more likely to break the acquired company than benefit from it.” Enderle gave examples such as Autonomy, Palm, ArcSight and SGI to illustrate that trend.

Under the deal, HPE is committed to investing in and rapidly growing the Determined Training Platform as an open source project, according to the post. “Our customers and open source community members will continue to receive the same high level of service and support that they always have, from a team of experts who are intimately familiar with the challenges they are facing. Moving forward, we will be in a position to serve a much more global network of users, both large and small. We are excited to continue building new features to both our customers and our open-source community.”

Determined AI’s employees are a collection of ML and distributed systems experts, including key contributors to open source community projects such as Spark MLlib, Apache Mesos and PostgreSQL. The startup’s investors include GV (formerly Google Ventures), Amplify Partners, CRV, Haystack, SV Angel, The House and Specialized Types.

HPE's new Determined AI technologies will also be tying in with HPE's GreenLake cloud services for HPC as the company integrates the new platform into its products and services.