Edge computing — which is to say, the concept of running advanced AI on a device, rather than from the cloud — is the underpinning behind many of the most-anticipated technological advancements coming down the pipeline.
Take a self-driving vehicle, as a major example. An autonomous car has to be able to respond to unexpected events on the road, like a child running across the street, in milliseconds. That's not nearly enough time to be able to rely on the car to communicate with an AI system hosted in a cloud like Amazon's or Microsoft's and get a response. By placing the emphasis on the car's own processor, however, the vehicle can be relied on to make decisions instantly.
While the battle over supremacy in the edge computing sector isn't as well-known or contentious as the cloud wars, juggernauts like Amazon, SAP, and Oracle are all vying with each other and a slew of startups for an early leadership position in what experts predict could be a $65 billion billion market by 2024.
But one segment — the race to develop the chips to support edge computing — is quickly heating up.
At Intel, edge computing is becoming an important piece of the tech giant's business, even as the company is roiled by executive turnover and delays to its next generation of PC processors.
In 2019, Intel says its edge computing unit generated $10 billion in revenue, a significant chunk of its $72 billion in total annual revenue. The company declined to elaborate on how much of an increase that was from prior years. And Intel is bullish on expanding its customer base, which already includes corporations like Audi, CenturyLink, and Rakuten Mobile.
"This is just about helping our partners do business in a more simple and effective way using all sorts of digital technology," Intel Vice President Dan Rodriquez told Business Insider.
But Intel's ambitions were dealt a blow with graphics processor specialist Nvidia's impending $40 billion acquisition of SoftBank-owned Arm Holdings— all part of a strategy by CEO Jensen Huang to double down on edge computing.
Nvidia already makes the graphics chips that have become critical in the push to artificial intelligence. Arm, for its part, provides the design that underpin most smartphones — and, increasingly, servers in data centers around the world.
By purchasing Arm, according to analysts, Nvidia can combine its GPU prowess with Arm's CPUs to create a powerful system designed for both high-performance computing tasks like AI, as well as basic processing.
The rivalry is looking to get even more intense after Nvidia surpassed Intel earlier this year as the most valuable chip maker in the US. Despite the looming threat, Intel's strategy remains steady, according to Tom Lantzsch, senior vice president and manager of the "Internet of Things" group.
"I don't think it changes anything for us," he said. "If anything, it maybe pushes us to go faster, but competition always does that."
'We have a very wide scale'
One reason why Intel is so bullish it can lead in the burgeoning edge computing industry is because it already has 15,000 customer deployments under its belt among 1,200 customers.
"We have a very wide scale," said Rodriquez. "With the capabilities that we're investing in, plus our ecosystem and all of our partners, we're very confident that we're going to be very successful."
Audi, for example, is using Intel's tech to improve quality control in its manufacturing process. Prior to adopting more sophisticated machine learning capabilities, each of 5,000 welds included in every one of the 1,000 vehicles assembled every day at one facility would need to be inspected by hand — an impossible task given that amounts to 5 million welds that would need examination every day.
Now, after the German carmaker tapped a predictive quality control system from Intel powered by edge computing technologies, the amount of weld inspections possible in a day increased by over 100-fold and labor costs were cut by as much as 50%.
"If customers plan to have this kind of capability and add this kind of capability that we can demonstrate with these solutions, we can demonstrate to them immediately the business impact," Lantzsch said.