HPE Launches All Out AI Offensive, Pledges 30-50 Percent Savings Over Custom Appliances
Hewlett Packard Enterprise Wednesday launched a massive artificial intelligence (AI) sales offensive with Proliant and Apollo hardware platforms that undercut AI appliance prices by as much as 30-50 percent.
At the same time, HPE pledged an AI channel charge that will include solution provider behemoths, like CDW, and systems integrators.
The new HPE Platform for Rapid Deep Learning Application Development is comprised of HPE Proliant DL380 and the HPE Apollo 6500 systems, integrated tightly with Bright Computing's software for managing high performance computing clusters and NVIDIA's Tesla V100 GPUs.
"This brings high performance computing power with deep learning to partners and customers at a fraction of the cost it was available previously," said Pankaj Goyal, vice president of the Artificial Intelligence Business for HPE. "These (image classification algorithm) reference designs from our labs are performance tested, best-in-class that are at least 30 to 50 percent cheaper than the nearest appliance. We are lowering the barrier to get started with AI and simplifying the adoption for our customers and channel partners."
Exact pricing was not available on the hardware platforms. "Our core philosophy is customers should have a low entry point on infrastructure costs," said Goyal.
Besides the Image Classification Design, HPE is also making available a Deep Learning Cookbook from its AI team designed to help partners select the best hardware-software stack for other deep learning tasks including multilayer perception (MLP) and video algorithms, said Goyal.
The new offerings strike at the heart of complaints from customers frustrated with the multitude of frameworks, tools and AI offerings that were far too complicated and confusing to put a stake in the ground in the fast moving AI applications market, said Goyal.
"We are giving our customers and channel partners an easier way to get started," he said. "What we are providing is an integrated software development environment with pre-embedded deep learning libraries and tools and auto updates."
The simplified approach is aimed at increasing the AI sales pipeline for partners, said Goyal. "We are training partners and developing joint marketing sales motions for AI targeted to specific workloads," he said. "We will be working jointly with partners to create a sales pipeline."
Erik Krucker, CTO at Comport Consulting, an HPE Platinum partner and No. 379 on the CRN Solution Provider 500, one of the top healthcare solution providers in the country, said he was excited about the stepped-up pace of AI innovation coming from HPE.
"It's a pretty big deal for HPE to put together cost-conscious reference architectures around cases such as image classification," he said.
"It's great that HPE is investing in the reference architectures so customers know these solutions are rock solid."
The image classification reference design on Proliant and Apollo could have a big impact on driving improved diagnostic capabilities from radiology reports that are analyzed by AI-based systems, said Krucker.
"AI is much better suited to figure out over time what has changed with a particular radiology image than humans could ever be," said Krucker. "Today, radiology image classification is a very human-intensive process fraught with problems. You have someone looking to see if there is an anomaly in an image compared to a previous image. This provides hospitals with the analytics to make better, more informed decisions. This is allowing clinicians to focus in on certain areas versus looking at a blank canvas. It puts the energy, focus and time where clinical capabilities should be focused."
Comport has been a leader in the AI healthcare charge and is already working closely with hospitals with the highly touted NVIDIA Tesla V100 GPUs, which are powering AI breakthroughs on Proliant and Apollo systems. "The fact that HPE is developing a deeper relationship with NVIDIA is great for us," said Krucker. "A lot of the AI number crunching and analytics is enabled by NVIDIA GPUs."
Key in the future is going to be marrying the Proliant and Apollo reference architectures with new technology breakthroughs like Nimble's AI-based Infosight predictive analytics and Aruba Niara. "That is going to allow us to put together entire AI-based solutions rather than selling bits and pieces," said Krucker.
Krucker applauded HPE CEO Meg Whitman for driving an innovation renaissance with a wide range of breakthroughs. "Meg has made HPE a technology leader," he said. "It's great Meg is focusing on innovation around both hardware with SGI and software like Nimble Infosight and Aruba Niara. It's great to see that HPE is not just stuck in the old world of pushing boxes. Hardware does matter, but the solution stack matters just as much. This is the first step in how HPE puts those two things together for AI."