March 22, 2018
HP Enterprise (HPE), the division dedicated to enterprise products and offerings, is joining the AI land grab. This week, it announces the launch of several “vertical AI solutions” to “accelerate deep learning training.”
According to the press release, the new offerings are designed to “help customers ramp up, optimize and scale artificial intelligence (AI) usage across business functions to drive outcomes such as better demand forecasting, improved operational efficiency and increased sales.”
The centerpiece is the hardware HPE Apollo 6500 Gen10, a high-performance computing (HPC) system for deep learning. The system is augmented with HPC Digital Prescriptive Maintenance Services, HPE AI Transformation Workshop, and a reseller agreement with WekaIO, a HPC file system provider.
Enthusiasm vs. Implementation
Despite enthusiasm for AI and what it can do, the majority of businesses may be sitting on the fence, still figuring out the right strategy for AI adoption. Research firm Gartner reports that, according to its recent survey of global CIOs, “only 4% of respondents had deployed AI” (”Lessons from AI Pioneers,” February 2018).
Based on survey respondent feedback, Gartner notes, “It makes sense to pursue small-scale plans likely to deliver small-scale payoffs that will offer lessons for larger implementations.”
This may explain HPE’s plan to augment the hardware offering with hand-holding and guidance from the workshops to help businesses identify the right areas to apply AI.
“Many customers we talk to just don’t know how to get started,” said Pankaj Goyal, HPE’s VP of AI Business & Strategy. “They’re struggling with step one.”
GPU-Accelerated Automation for Industry 4.0
HPE’s Digital Prescriptive Maintenance Services, the first in a series, “automates problem prevention and increases productivity of industrial equipment,” according to HPE. Goyal described it as “AI-enabled industry offering.”
HPE offers two delivery models. The company sells HPC systems that buyers can maintain on-premise. It also offers hardware as a service under its Flexible Capacity plan.
HPE Apollo 6500 Gen10’s “support for eight NVIDIA Tesla V100 GPUs delivers dramatic increases in application performance, enabling a 3x faster model training than previous generations,” HPE’s announcement states.
GPU maker NVIDIA has thrown its full force behind the rise of AI, championing the use of its GPU’s parallel processing power for the massive computation required in machine learning. NVIDIA also offers DGX, a line of GPU-accelerated hardware for AI research.
The manufacturing sector’s AI adoption may begin with robotics and factory maintenance, Goyal reasoned. AI-powered manufacturing systems may “automatically detect failures before they happen,” he pointed out. Prescriptive maintenance, as the name suggests, “will not only alert you of pending failures but also prescribe ways to fix them,” he added.
Cultivating AI Talent is the Key
AI adoption strategy will invariably require both access to HPC hardware and AI talent. The HPE AI-centric hardware can be managed and maintained just like any other equipment under the supervision of IT, but conducting machine learning is beyond the IT territory.
In “Lessons from AI Pioneers,” Gartner points out, “Lack of staff and skills to conceive and execute AI projects is a significant obstacle to progress for many organizations, so external service providers can play a key role in planning and delivering AI-powered software.”
For manufacturing, that may mean working with data scientists and big data experts to extract executable nuggets from equipment history data, to identify patterns that point to pending failures.
HPE’s firmware for HPE servers “has built-in machine learning capabilities,” Goyal says. “We’re continuing to build these capacities.”
HPE’s storage offerings are augmented by the InfoSight maintenance engine software, which “continuously learns as it analyzes [sensor] data, making every system smarter and more reliable.”
HPE’s AI-targeted offerings may benefit from a dose of InfoSight technology or something of the same variety.