Intel's plan to put AI in every single chip

Intel wants AI in literally everything it makes.

During Intel's Q2 2023 earnings call, CEO Pat Gelsinger made a massive declaration to investors: the chipmaker plans to build AI into every single platform and product going forward.

Here is how Intel plans to pivot the AI conversation away from massive cloud data centers and straight onto your desk.

No more cloud round-trips

Intel is preparing to roll out Meteor Lake later this year, marking its first consumer processor equipped with an onboard neural processor. AMD, Apple, and Qualcomm have already introduced similar silicon, but while Intel previously hinted that only top-tier Ultra chips might get the AI coprocessor, Gelsinger made it clear that this technology will eventually reach everything the company sells.

Gelsinger argued that relying entirely on the cloud does not make economic or technical sense for everyday computing. In his words, users cannot simply make a round trip to the cloud for every single task.

Latency, limited bandwidth, and massive server costs get in the way. Whether it is real-time translation in a Zoom meeting, continuous transcription, Adobe creative tools running on the machine, or local legal brief generation, edge execution solves bottlenecks that remote servers cannot handle. Gelsinger pointed out that local convenience stores and retail sites are not going to erect dedicated 10-megawatt server farms just to run inferences. He even joked that future hearing aids, including his own, will run dedicated AI.

Playing to Intel's strengths

Intel's pivot is also a strategic defense. Nvidia captured the lion's share of the current generative boom, sailing past a trillion-dollar valuation on the back of data center GPUs. Intel missed that initial wave and needs a narrative of its own.

Fortunately for Intel, key partners like Microsoft also want local compute. While Microsoft is pushing its $30-per-month cloud Copilot, Windows licensing remains a cornerstone of its business. Windows chief Panos Panay previously hinted that AI would redefine the operating system, and leaks suggest that Meteor Lake's onboard neural engine is specifically tailored for the next release of Windows.

Why it matters

For regular users, shifting AI from distant data centers to local silicon means lower latency, better battery management, and genuine data privacy. Tasks like real-time audio translation, noise removal, and local text indexing will not have to expose private files to third-party cloud servers.

We think this pivot is Intel's smartest available move, but it carries plenty of risk. While Intel still dominates the sheer volume of PC shipments, having a neural chip onboard means nothing if software developers do not build for it. Unless Windows and major creative suites seamlessly integrate these local engines to deliver obvious time savings, consumers will not see an urgent reason to upgrade their hardware just for an idle AI processor.