Head over to our on-demand library to view sessions from VB Transform 2023. Register Here
This article is part of a VB special issue. Read the full series here: The future of the data center: Handling greater and greater demands.
Those of a certain age may remember an IBM commercial from 2006 showcasing the possibilities of radio-frequency identification (RFID).
In the now 17-year-old ad, a suspicious-looking guy in a trench coat quickly moves around a store, stuffing items in his pockets, drawing the attention of a security guard and other shoppers. When he walks out, he sets off what seems to be a security system — but the guard just informs him that he forgot his receipt.
At the time, the concept of autonomous checkout seemed pretty far-fetched and more in the realm of science fiction — but today it’s a reality. Computer vision, sensors, wireless, 5G, machine learning (ML), edge capabilities and other high-performance computing (HPC) technologies are set to transform retail as we know it.
VB Transform 2023 On-Demand
Did you miss a session from VB Transform 2023? Register to access the on-demand library for all of our featured sessions.
“It takes HPC to deliver the level of customization, precision and speed that keeps retail businesses out of the red these days,” Aron Ezra, chairman of Plan A Technologies — which has created software tools for major retail companies including McDonald’s, Kellogg’s, Microsoft, Apple, Ford and Volkswagen —told VentureBeat.
High-performance computing at the edge
With consumers increasingly demanding more of their in-person and digital experiences, retailers are embracing HPC and cutting-edge technologies including edge computing — in which HPC is done on-site (or near it) rather than at a remote data center.
The edge provides lower latency, faster response times, and data privacy, said Jordan Fisher, CEO of retail computer vision platform company Standard AI. This is particularly true for autonomous checkout.
Fisher pointed out that while the proliferation of self-checkout is a good introduction to the market, “we can go way beyond that.”
As the IBM ad foretold, cameras situated throughout a store tabulate customers’ items as they fill up their carts, and they are charged upon exit.
“They can walk in, walk out, get their receipt automatically,” said Fisher.
Computer vision capabilities can also help navigate shoppers to products and alert staff if an item is out of stock.
At this point, the emerging technology is easier to roll out in smaller, convenience-type stores because it is expensive, Fisher said. However, that price point will come down over time.
“This is early, but this is where retailers are heading,” he said.
Ezra agreed, noting that “as the demand for lightning-fast results grows (and it will never stop growing), edge computing has become an increasingly valuable concept.”
HPC is also critical in allowing retailers to take advantage of their wealth of data to drive insights.
“In 2023, the world doesn’t run on money,” said Ezra. “It runs on data — and lots of it.”
HPC harnesses and combines the processing power of arrays of computers to crunch huge amounts of data and solve highly complex problems “literally a million times faster than a desktop computer could on its own,” he said.
As an example: A family-run teddy bear company is looking to optimize product sales to save time and money. To achieve this, it needs to know whether kids in Orlando are more into stuffed dinosaurs or stuffed bears. Then it needs to make sure it can ship enough “tiny T-rexes” from its warehouse in Illinois to stores in Florida.
“Not doing so means being out of stock when a cadre of junior paleontologists visits the stores after Disney World,” said Ezra. “And you’ve got to do that for every store across the country or world.”
Omnichannel a must in modern retail
Another big reason for retailers to adopt HPC in digital retail: omnichannel experiences.
“[Omnichannel is] a bit of a buzzword that people are throwing around and applying to almost everything,” Ezra acknowledged, “but the fact of the matter is that today’s consumers demand a seamless, centralized digital shopping experience.”
Consumers expect to be able to get everything they need through a single app, and pick up where they left off if they switch platforms — and they want to be engaged throughout the entire experience.
High-performance data analytics (HPDA) allows retailers to provide personalization, product logistics, security and omnichannel experiences in near real-time and add predictive analysis, said Ezra. Enterprises can also take advantage of real-time information about orders, inventory and customer behavior.
Turning this “exponentially growing mass of data” into actionable insights about the market can help with targeted ads. Real-time intelligence can allow retailers to react to sudden changes in market conditions and improve error analysis. Ezra said that HPC can also be used to “sift through oceans of data” to detect security anomalies.
Furthermore, he said, AI and HPC can have a mutually beneficial relationship going forward. Natural language processing powered by HPC, for instance, allows for a much more personalized shopping experience.
Getting infrastructure right from the get-go
Cost is always the number one concern for retailers, Fisher noted, but in adopting HPC, he advised not to be short-sighted.
“It’s really important to think about this as an infrastructure investment — you want to make sure you pick the right infrastructure that will work today and in the future,” he said.
He used the analogy of laying out electricity lines in America: It was a massive undertaking, but it was important to get it right from the outset. A technology like autonomous checkout, for example, will fundamentally change retail, and it’s an expensive infrastructure layer that retailers should only install once. With time, new and different features can be built onto that infrastructure to “continuously pay dividends.”
“You want to lay the foundation once — and yes it’s going to be expensive, but you get to capitalize on that for a store’s life,” he said.
Factors to consider in adopting HPC
Retail enterprises must first consider whether they want to build their own HPC systems — which can be expensive, but offer full control — or rent out HPC clusters, said Ezra. Google and AWS offer such capabilities, as do smaller, more specialized companies.
Also, retailers need to ensure that they’re getting the power they need without overpaying, and they must consider where HPC data will be stored and the space requirements for that. HPC tools that are 100% cloud-native with virtual CPUs can save rack space and cost, but that also means data has to pass through more layers, “from the compute cores and back again,” he said, which could slow things down.
Whether they buy or rent, though, retailers must ensure that data is safeguarded with the appropriate level of encryption and security.
Looking ahead, evolving HPC capabilities offer exciting possibilities, said Ezra.
“Today’s computers are over one trillion times more powerful than the ones in the 1950s,” he said. “HPC takes that jump even further. It will be incredibly exciting to see what amazing new solutions will be enabled by it as we look ahead toward the coming decades.”
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.