Life on the Edge: A Guide to Computing's Newest Frontier

 

"Artificial-intelligence companies are at the leading edge of saying "edge" in their earnings calls. "We see AI at the edge as a large growth opportunity," said chip maker Advanced Micro Devices recently. Western Digital, which makes memory, said that AI will drive demand for "flash at the edge."

Qualcomm led the industry with five "edge" mentions in its latest earnings call. And Procter & Gamble back in January touted Charmin Ultra Soft with "scalloped-edge perforations." That last one might not be a computing reference, but PG said its plush new rolls were delivering improved "end-use experience," which is just how a technologist would describe the matter.

I had some investment questions about the edge. Where is it? What happens there? Is it a falling hazard? And if the edge is so buzzy, is it bearish for the middle?

Amazon.com's Amazon Web Services, which should know a thing or two about edge computing, defines it as "the process of bringing information storage and computing abilities closer to the devices that produce that information and the users who consume it."

That makes it sound like the early 1990s. Back then, personal computers went obsolete faster than you could rewind a VHS tape while chugging a Crystal Pepsi. Processing oomph primarily lived on user machines-the "edge"-and each new version of the Microsoft Windows operating system required a lot more of it. Good times for CompUSA.

The internet and the rise of cloud computing shifted processing power to remote servers. PC and smartphone replacement cycles lengthened. Last year PC sales fell 14.8%, the second consecutive double-digit decline.

But there are early signs of a bounce. Canalys, a data firm, says that first-quarter PC shipments grew 3.2%, including a 4.2% increase for notebooks. It reckons that AI is behind the rebound. Companies are buying new machines before upgrading to Windows 11 and its centralized AI assistant, called Copilot, which runs in the cloud but is expected to run locally in the future. PC and chip makers are adding dedicated AI accelerators, including so-called neural processing units.

There are three reasons that the edge is due for a computing upgrade, says Daniel Pilling, a semiconductor analyst at Sands Capital. One is what's known as latency. If Siri wants our relationship to thrive, she's going to have to stop giving me the cold artificial shoulder when we drive through woodsy areas. More local processing, or "inferencing" as the AI folks say, can help. A second reason is a hidden cost benefit for companies. "If you're Microsoft and you sell us a Copilot, it's much better for the consumer to pay for the inference, right?" says Pilling.

A third reason for edge excitement is privacy, which is closely related to memory. "If you go on ChatGPT, it doesn't really know who you are when you come back," says Pilling. "In the future, we will probably want to have our own models that know us. And then maybe that model just resides on your phone for privacy reasons."

If you're thinking that this sounds like good news for Apple, you're not alone. "Are you a believer in the edge thesis that AI and processing on smartphones and devices like yours is going to have a huge role in AI?" an analyst asked CEO Tim Cook earlier this year. Cook said that AI was a "huge opportunity" but that he didn't want to give more details or get "out in front of myself."

This past week, Apple introduced an iPad Pro powered by a new inhouse processor called M4. It includes a neural processing unit that can handle 38 trillion operations per second, or TOPS. BofA Securities calls TOPS an imperfect but simple measure that Apple and rivals will likely use to tout performance.

Microsoft, which will host a Windows AI event on May 20, is expected to set a minimum requirement for AI PCs of 40 TOPS, which today's machines fall short of but new ones could meet by year's end. BofA sees this as good news for Buy-rated AMD, Arm Holdings, Nvidia, and Micron Technology, and Neutralrated Intel.

Meanwhile, Apple next month could host the "biggest unveil event for Cook and Cupertino in roughly a decade," wrote Wedbush Securities this past week. That's a reference to the company's yearly Worldwide Developers Conference, which typically focuses on software. This year, the company could give details on its AI strategy, including perhaps an AI App Store, which Wedbush predicts will lay the foundation for an iPhone sales "supercycle" this fall.

Don't let all this edge emphasis make the middle seem mundane, however. Pilling at Sands Capital sees an "avalanche of demand" coming there. Think of the primitive offerings when Apple first opened its App Store in 2011. "You had ringtones, right? Because the hardware wasn't good enough," he says. As phones improved, sophisticated new apps arrived by the millions.

With AI today, the public is focused on early applications like ChatGPT 4, but as chips get better, AI algorithms will improve exponentially, and machines talking to each other will discover new uses, says Pilling. "And with that use case explosion, you need way more chips."

Among Pilling's favorite stocks is Nvidia, which he says has the industry's top mathematicians and scientists, and a lead in the software behind its chips.

He also likes equipment maker ASML Holding for a process called lithography, and Lam Research for ones called etch and deposition.

 Another pick: Taiwan Semiconductor Manufacturing for fabrication. I asked about the China risk there. "If China were to control Taiwan tomorrow, then the ASMLs of the world would stop delivering spare parts, which would mean that all of the Taiwanese equipment would stop working within two or three weeks," Pilling says." [1]

1.Life on the Edge: A Guide to Computing's Newest Frontier. Hough, Jack. Barron's; New York, N.Y. Vol. 104, Iss. 20,  (May 13, 2024): 9.