During the company’s Q2 2023 earnings call, Intel CEO Pat Gelsinger expressed strong confidence in AI, telling investors that the business intends to “build AI into every product that we build.”
Meteor Lake, Intel’s first consumer chip with an integrated neural processor for machine learning activities, will go on sale later this year. (AMD recently followed Qualcomm and Apple by doing the same.)
However, Gelsinger seems to anticipate that AI will eventually be included in every product Intel sells, contrary to Intel’s earlier suggestion that only its high-end Ultra CPUs might have those AI coprocessors.
Gelsinger frequently extols the virtues of the “four superpowers” or “five superpowers” of tech firms, which originally encompassed both AI and cloud. However, he now contends that AI and the cloud are not inextricably linked.
Today, you may start to see people using the cloud to play around while writing a research paper using ChatGPT, which is kind of awesome, right? Kids are simplifying their homework assignments in this manner, of course, but you won’t do that for every client because being AI-enabled requires that it be done on the client, isn’t that right? The cloud is not accessible. A circular voyage to the cloud is not possible.
Real-time language translation during zoom calls, real-time transcription, automation inferencing, relevance portraying, generated content and gaming environments, real-time creator environments through Adobe and others doing those as part of the client, new productivity tools—bbeing able to generate local legal briefs for clients one at a time—isn’t that right? There will be a plethora of AI enablements, and those will be client-centred across every area of consumer, developer, and enterprise efficiency use cases. These will be at the edge as well.
A circular voyage to the cloud is not possible. If you wanted to round-trip inferencing from, say, a neighbourhood convenience shop to the cloud, you wouldn’t have the latency, bandwidth, or cost structure to do so. Everything will take place at the client’s edge.
At another point in the discussion, he stated, “AI is going to be in every hearing aid in the future, including mine.” They won’t set up a dedicated 10-megawatt farm, regardless of whether it’s for a client, an edge platform for retail, manufacturing, and industrial use cases, or an enterprise data centre.
On one hand, it makes sense that the CEO of Intel would say this. The types of chips that power the AI cloud are produced by Nvidia rather than Intel. Because it offered the ideal picks for the AI gold rush, Nvidia’s market cap soared to $1 trillion. Intel must make its own entrance.
However, it’s also true that not everyone wants everything on the cloud. This includes Microsoft, a cloud provider that still generates a sizable portion of its revenue from the sale of licences for Windows PCs.
At AMD’s chip introduction in January, Windows CEO Panos Panay teased that “AI is going to reinvent how you do everything on Windows,” and he wasn’t just saying that for show. After Copilot was unveiled in March and integrated into Windows, my colleague Tom now thinks Microsoft’s new AI-powered Copilot tool will transform Office documents forever. However, Copilot will cost $30 per user each month and is currently powered by the cloud.
The one to watch for will be Windows 8. Windows 12 is reportedly the target of Intel’s Meteor Lake, which includes a built-in neural engine.
