Photo credit: Walmart
The world of opportunity for AI continues to expand, offering businesses unprecedented opportunities for insights and efficiencies.
Walmart just unveiled its AI-enabled store and is continuing to show the important role AI can play in helping them understand their operations better with real-time access to an incredible amount of information about their inventory, improve customer service.
The concept is innovative for a few reasons that I’ll get into shortly, but what I liked about the story is that Walmart is not removing people from the retail location (like cashier-less stores), but is instead re-deploying their staff to help improve the customer experience by keeping stock replenished, ensuring stale/over ripe foods are removed and increasing availability of personnel for customer questions, support and staffing check out lanes. (A bonus is we won’t have to hunt around a parking lot for a cart ever again because alerts will go out before they run out!)
The unveiling comes on the heels of announcements about their robot janitor, and to me underscores the value Walmart places on new technologies to their future competitiveness in the ultra competitive retail space.
Taking it to the Edge
Walmart estimates its 50,000 square foot retail space will generate 1.6TB of data a second. They have installed a data center on premise that customers can see and have installed kiosks for customer interaction. They’ve said it’s important to make all the technology or how it’s being used visible and accessible to help put customers at ease with the concept.
What made the announcement of this retail concept even more interesting it that, Walmart is not only using on site servers to handle the data (visible to shoppers behind a glass wall) but it’s taking AI to the edge to take advantage of edge computing’s speed and responsiveness for AI inference.
In 2017, Gartner predicted that, just as cloud computing had started to eat enterprise data centres with born in the cloud applications and workloads, “edge computing will eat the cloud”. Now, before you start writing comments: I am not saying all AI applications will take place at the edge. There are times and applications for cloud computing, edge computing and a hybrid approach. Walmart’s new AI-powered store is a prime example of hybrid strategies at work.
The analyst went on to note that responsiveness and the need for real-time interaction are drivers of pushing workloads to the edge but that “technologies for the edge will be completely different, much more dynamic, much more evolutionary and competitive.”
I agree technologies at the edge are different, dynamic and evolutionary – and we’re taking bold steps to accelerate innovation, as we recently unveiled at data-centric innovation day with a portfolio of products to move, store and process data including edge-computing solutions like Intel® Xeon® D processors, Intel® Optane Memory, and Intel® Agilex™ FPGA.
Why AI Needs an Edge
Statisca projects the number of connected devices worldwide will hit 74.44 billion by 2025, up from 26.66 billion this year. Cisco estimates IoT devices will produce 5 quintillion bytes of data produced every day.
All of these devices are transmitting valuable data back to the enterprise, but as I mentioned last time, only a small fraction of that data is being used or analyzed. We need to enable businesses to accelerate data-centric decision making and moving workloads to the edge is a good strategy for some situations. For example, when a retailer is using cameras to monitor stock levels or freshness, recognition apps need to have answers quickly. They can get those answers faster on the edge.
Ecosystem for AI
At its core, a single AI project need to blend processes associated with:
• capturing data
• getting data ready to use
• creating, securing and fine-tuning models
• deploying solutions at scale in the real world.
But, while the high-level processes might be similar, the AI space is becoming increasingly complex, and since companies have different initiatives and priorities so a one size fits all solution doesn’t work.
That complexity is why businesses need a connected ecosystem of partners that span a wide range of expertise and skill sets as well as knowledge on deploying the right combination of hardware, software, and accelerators. Businesses are also wrestling with decisions around environment, infrastructure, and the complexities around managing workloads across multiple architectures.
Are you ready to help your clients tackle the complexities of AI? As with many digital transformation projects for the data-centric era, the channel is well positioned to help customers on their AI journey from the cloud to the edge.