Industry Expert: Fed AI Adoption Hinges on Power, Cost

An official with data storage technologies provider Pure Storage told MeriTalk that issues holding back Federal government adoption of artificial intelligence (AI) technologies include the high cost of super-computing resources and the continued reliance of government agencies on legacy IT systems that were not created with AI applications in mind.

“AI is absolutely exploding,” said Gary Newgaard, public sector vice president Pure Storage, in an interview with MeriTalk. “The challenge for our Federal and state and local customer base is how do they get off their legacy systems and quickly modernize.”

Bringing AI to Government

According to industry research group Gartner, 80 percent of enterprises will deploy AI by 2020–and the Federal government will be no exception. But while agencies are racing to figure out how they can turn the piles of data they have into actionable information, the government’s reliance on bulky, out-of-date legacy IT systems is holding them back.

Newgaard explained that while the government has typically been an early adopter of new technology, fully embracing AI remains a challenge.

In addition to legacy system issues, the computing power needed to run AI technologies and programs often is too costly for agencies, he said.

“You’ve got supercomputers that are hugely expensive,” he said. “So, you get a lot of projects that don’t get forward momentum because they either can’t get time on the bigger boxes or they don’t get the focus they need.”

Lowering Costs, Increasing Simplicity

To address that problem, Pure Storage, an all-flash storage provider, and NVIDIA, a leader in the graphics chip space, launched what they are billing as “the industry’s first comprehensive AI-ready infrastructure.” The technology, dubbed AIRI, will make AI more accessible for the government, the company says.

“You don’t need anyone to operate it, it comes in 50 inches, and you can set it up in hours,” Newgaard said.

AIRI is powered by Pure Storage’s FlashBlade technology and four NVIDIA DGX-1 supercomputers. The solution is supported by the NVIDIA GPU Cloud deep learning stack and the Pure Storage AIRI Scaling Toolkit, according to a news release.

Newgaard said Pure Storage wanted to work with NVIDIA for a very simple reason–NVIDIA is good at what it does.

“The quality of their products is superb,” he said. “It’s a match made in heaven from a technological standpoint and it gets the solution out into the market in an affordable manner.”

Newgaard said the greatest benefit of AIRI is its simplicity.

“From an end-user standpoint, supercomputers are big and complex,” he said. “The great thing about this AIRI solution is its simplicity. It’s simple to own and operate. When people do upgrades and technology refreshes on supercomputers, they are used to a 3-5 year life space. The AIRI solution can be upgraded on the fly, with no downtown.”

Helping Federal Agencies

The great challenge with AI is having enough computing power to quickly analyze a mountain of data. AIRI, according to Pure Storage and NVIDIA, can give the government the computing power it needs.

Newgaard has a very clear idea of how AIRI can help the government.

“It’s going to help where minutes and second make a difference,” he said. “It can make a significant difference in protecting the nation and finding the cure for cancer. Speed and accuracy have always been the pot of gold we are all chasing. When you are looking for a needle in a basket full of needles–that’s where machine learning comes through.”

No Comments

    Leave a Reply

    Recent