Skip to main content

Intel CES 2026 keynote live blog: Panther Lake debuts in Core Ultra Series 3 laptop CPUs, Arc B390 IGP announced

Jim Johnson, Intel’s Senior VP and GM of the Client Computing Division, is set to take the stage in Las Vegas

Intel Core Ultra Series 3
(Image: © Intel)

We're on the ground at Intel's CES 2026 keynote event in Las Vegas, Nevada. Jim Johnson, Intel’s Senior VP and GM of the Client Computing Division, is set to take the stage to give us the rundown on all things related to the latest Core Ultra Series 3 processors. You'll see those updates below as the event begins.

Intel’s Core Ultra Series 3, part of the Panther Lake family, is the company’s first client processor with a compute tile built on the new 18A node. Compared with the Lunar Lake family, the new Panther Lake processors are expected to deliver a 50 percent performance uplift while maintaining a similar power profile.

Refresh

Our Tom’s Hardware team is reporting on-the-ground from the Delfino ballroom at the Venetian Hotel in Las Vegas and virtually at home. The event is set to start at 3:00 PM PT / 6:00 PM ET.

Tom's Hardware

(Image credit: Future)

Intel’s CES 2026 keynote is sandwiched in between Nvidia, where CEO Jensen Huang spoke earlier, and AMD, where CEO Lisa Su will likely launch a counterassault to the new Core Ultra Series 3 processors for mobile.

Tom's Hardware

(Image credit: Future)

A surprise appearance by Intel CEO Lip-Bu Tan on the stage at the Venetian. "Intel is delivering on its commitment to ship its first 18A products by the end of 2025. Intel is ramping all Core Ultra Series 3 die packages as we speak. Intel is uniquely positioned to deliver this kind of product, he says, across design, process, and packaging."

Tom's Hardware

(Image credit: Future)

Lip-Bu Tan says that the Core Ultra Series 3 processor will usher in the "next evolution of the PC."

Tom's Hardware

(Image credit: Future)

Jim Johnson has just taken the stage. He explains that Intel and the industry are at a strategic inflection point, particularly when it comes to geopolitics. 18A is the "center of that effort," and it's in high-volume production.

Intel's Jim Johnson

(Image credit: Future)

Digging into RibbonFET, talking about the first gate all around (GAA) transistors. Intel has been talking about this and PowerVia for years, shortly after Pat took over as CEO. Now it's finally coming to fruition, which is a really big deal. Leadership manufacturing has once again become a strategic pillar of Intel, he says. The next wave is AI, if any of you guys weren't aware.

— Jake Roach

Intel

(Image credit: Future)

18A is "the most advanced process in the world," and Intel actually has a compelling argument for that. Rehashing new architecture, but highlighted ray tracing capabilities on the iGPU in Panther Lake. The goal on Panther Lake was to scale, but Jim Johnson says he "can't wait" for people to see the GPU in Panther Lake. Intel says it has "nailed its goals" with Panther Lake.

— Jake Roach

Intel

(Image credit: Future)

Panther Lake moves the GPU tile to its own chiplet, giving Intel greater design flexibility. Completely redesigned the low-power island with dedicated cache and its own power rail, particularly for workloads like web browsing. Beyond the SoC, Intel is using other features to improve efficiency, like IPU, Intel's Wi-Fi solution, and adaptive charging.

— Jake Roach

Intel

(Image credit: Future)

Foveros allows Intel to deliver far greater flexibility, which OEMs have "been asking for for two generations." Streaming 4K video requires 1/3rd the power of previous generations. Puts to bed the "myth" that x86 can't be power efficient, and "that's the last time I'm going to say that."

— Jake Roach

Intel

(Image credit: Future)

Dan Rogers is now out on stage. He admits that Intel has been "lacking in key features, driver support, and performance" when it comes to long history of Intel GPUs. However, Intel is "light years" ahead of where it was a few years ago with GPU.

— Jake Roach

Intel

(Image credit: Future)

Intel Arc B390 iGP announced. Intel claims it is 2X as fast as AMD in "select titles" and 73% faster on average. They're calling this a "discrete-class gaming performance" in a laptop. That could be the case if the B390 is actually as powerful as they suggest, though I suspect we're still looking at 1080p, medium settings, and plenty of help from XeSS.

The first integrated graphics to ship with multi-frame generation. That's true, but it kinda ignores Lossless Scaling, and the various other ways to generate frames on Nvidia and AMD GPUs (including iGPUs for AMD).

Intel

(Image credit: Future)

Jeff from EA is now out to talke about integration with Battlefield 6. Claimed up to 145 fps in Battlefield 6 with the B390, including XeSS with 4X frame generation. That's impressive, no doubt, but you're also looking at a base frame rate of around 35 fps, which isn't going to feel great to play. Battlefield 6 is getting a native XeSS 3 integration, which will excite dozens of people.

Here's a blink and you might miss it announcement: Intel is launching an entire handheld gaming platform with Panther Lake, and we'll have more details on it later this year. Probably Panther Lake in handhelds, which could be a big deal if the iGPU is as performant as Intel suggests.

Jim Johnston is back, and he's talking about AI and how it's being used in his own family. His son leads a software team, and AI-enabled coding is "completely overhauling" how they make products. He says engineering isn't about syntax, which fair, but pointing to vibe coding as the solution falls on deaf ears for all those junior developers. He adds, "think of what CAD tools did for mechanical and civil engineers."

Jim Johnson is pointing to examples of different apps that use on-device Gen AI, and he started with a GPU example. It's very telling that Intel is leaning into the GPU with Panther Lake, while its competitors like Qualcomm are leaning heavier into the NPU. Maybe that NPU wasn't as important as the industry pushed a few years ago.

Intel

(Image credit: Intel)

70B model in a 32K context, which is pretty big for a mobile SoC… and by pretty big, I mean really only rivaled by massive SoCs from Apple and Ryzen AI Max. Intel has shipped close to 4 Zetta OPs to the market, which is equivalent to about 40 data centers. I don't know if that's true, and I suspect it's not, but it's a cute figure.

Intel has a new AI Super Builder platform, which looks like an interconnected system of local and cloud AI. We've heard a lot about hybrid AI, but it looks like Intel is branding it

— Jake Roach

Intel

(Image credit: Future)

Perplexity CEO is now on stage, and he proclaims that deep tech is the real thought leadership in AI, and you're damn right it is.

Intel

(Image credit: Future)

AI feels slow to a lot of people, which is fair. If you're using ChatGPT every day, going out to the data center for every query really adds up. I think that makes a bigger difference for AI users than privacy, though that's arguably the more important reason for local AI.

— Jake Roach

Arvin is making an economics argument for local AI compute, which hits home right now with RAM prices out of control and utility costs rising in areas where AI data centers are running.

Intel

(Image credit: Future)

Panther Lake is going to the edge, which is a big change for Intel. It normally segments out edge and embedded solutions, but they all fall under the Series 3 umbrella, and Intel is actually drawing attention to them.

Intel

(Image credit: Future)

Intel

(Image credit: Future)

"This will be the most broadly adopted AI PC platform Intel has ever shipped." Over 200 designs coming from Intel's partners, with preorders starting tomorrow. Global availability coming January 27.

Intel

(Image credit: Future)

And with that, the keynote is over. Intel closed things out with some cool "Panther" sound effects.