Just like last year, NVIDIA has used some of its time on the CES stage to introduce an upgraded version of its real-time image-upscaling technology. The new DLSS 4.5 promises sharper visuals with the 2nd Generation Super Resolution Transformer, which the company says will deliver better temporal stability, reduced ghosting and improved anti-aliasing. The DLSS 4.5 also includes Dynamic Multi Frame Generation, which maximizes frame per second to monitor refresh rate, delivering up to 4K 240Hz path traced performance.
The 2nd Gen Transformer Super Resolution is available now for all RTX GPUs, while the Dynamic 6x Frame Generation will arrive some time in spring 2026 for the RTX 50 series. It will also be available for more than 400 games through the NVIDIA app.
Another new feature is RTX Remix Logic, which allows real-time environmental reactions to in-game events. For instance, when a door is opened on screen, the NVIDIA tech can present changes to volumetric conditions, display different weather simulations with particles or alter materials. More than 30 different common events can be detected, and the RTX Remix Logic can make adaptations to volumetrics, particles, material properties and light properties.
NVIDIA noted that it now has native clients for both Linux and Fire TV. That’s the roundup of major gaming updates from the company’s CES presentation, but NVIDIA CEO Jensen Huang had plenty to talk about earlier today.
This article originally appeared on Engadget at https://www.engadget.com/gaming/nvidia-announces-dlss-45-at-ces-2026-053000128.html?src=rss
For better or worse, CES 2026 is already shaping up to be a big year for humanoid robots. Chinese company Agibot showed up with two: the roughly human-sized A2 and the slightly smaller X2, both of which were displaying their surprisingly impressive dancing abilities.
We watched both robots walk around, wave at passersby and show off their best moves. The larger A2 mostly kept its legs still and danced mainly with its arms. The smaller X2 on the other hand is a bit more nimble — it has a larger set of “feet” to give it more stability — and those abilities were on full display.
At the time we saw them, the robots were controlled partially by an Agibot rep using a dedicated controller, but the company told me the robots are able to move autonomously in spaces once they’ve been able to use their onboard sensors to map out their environmentThe company, which has already shipped several thousand robots in China and plans to make them available in the United States this year, says both the A2 and X2 are intended to provide a flexible platform so people can interact with the robots in a variety of situations.
Agibot envisions the larger A2 as a kind of hospitality helper robot that can greet visitors at museums or conferences (like CES) and provide directions or even walk alongside their human guests.
The smaller X2 on the other hand could be suited for educational purposes or other scenarios when you might want a robot with slightly more human-like movements. It could even be a good TikTok companion, as Agibot’s head of communications, Yuheng Feng explained to me. “Take a Tiktok video, for example,you can use that video to train the robot, [so] it can also dance exactly like you did in the video.”
The company hasn’t given details on when its robots might show up in the US or how much they might cost. Feng told me a lot will depend on how companies want to use them because their hardware is able to be customized depending on the use case. For now, though, we’ll just soak in the dance moves.
This article originally appeared on Engadget at https://www.engadget.com/ai/agibots-humanoid-robots-can-give-directions-and-learn-your-tiktok-dances-045049798.html?src=rss
The following article discusses adult themes.
One surprising find at CES 2026 comes from OhDoki, the Norwegian makers of The Handy, a smart stroking device. It turned up to the desert with two new products, the Handy 2 and the Handy 2 Pro. Fundamentally, they’re the same device, but the Handy 2 Pro comes with a far bigger battery, taking the usable life from one hour on the standard model all the way to five. That extra power enables you to run the 2 Pro far faster and harder than the original once you set it into Turbo mode. In fact, I was told the hardware can essentially be “overclocked,” although I wouldn’t recommend it. Because the motor action was so aggressive that even just holding it in the hand, I’d be worried about breaking bones.
There are plenty of useful new features, including using the Handy’s beefy battery to charge your other USB-C devices. In addition, the sleeves are now be mounted to the motor arm with a click on motion, so you don’t have to wrestle with the (very) weighty hardware when you don’t have to. There are even tripod screws on both the side and bottom of the unit, enabling you to mount the Handy 2 to any compatible equipment. And, if you don’t want to use your phone to control the action, there’s a dot matrix display on the side which will show you both speed and stroke length.
Naturally, the company’s representatives were keen to highlight the connectivity features beyond just connecting to your own phone. As well as allowing other people to control the device, it’ll sync with various VR and video content platforms as well as tube sites like Xhamster. And the company has recently teamed up with a cam platform for more remote opportunities for intimacy.
The Handy 2 and Handy 2 Pro launch January 6, with the regular model priced at $299 and the Pro going for $499.
This article originally appeared on Engadget at https://www.engadget.com/general/the-handy-2-pro-is-an-overclockable-sex-toy-044456140.html?src=rss
Last year in Las Vegas at CES 2025, Sony Honda Mobility displayed yet another revision of its now-familiar yet still-unreleased Afeela 1 sedan. With a starting price of $89,900 and a maximum range of 300 miles, it didn’t seem like a great value proposition. 12 months later, the math still hasn’t changed, but would-be buyers now have a slightly taller option to consider.
Meet the Afeela Prototype 2026, a version of the Afeela 1 sedan that’s been up-scaled to better suit American SUV proclivities. It offers a carbon copy of the smaller Afeela’s understated styling, including the external, nose-mounted LCD, and is due “as early as 2028.”
That’s code for “2029 is probably more likely” and is, sadly, all we know about the thing for now.
And what about the Afeela 1? Shugo Yamaguchi, president and CEO of Sony Honda Mobility of America, confirmed that it’s still on track for a release this year, still priced at $90,000 to start, and still set to be built at Honda’s Ohio factory. That last bit is particularly good news, as the global tariff situation has been rather fluid since we saw the sedan last year.
Unfortunately, the rest of the Afeela 1’s core attributes haven’t aged so well, particularly given it’s intended to appeal to an increasingly skeptical American market, a situation that has driven Honda to dial down its own US EV aspirations. 300 miles of range is an increasingly middling figure, especially for a machine priced within spitting distance of a Porsche Taycan.
To make the Afeela 1 compelling, Sony Honda Mobility is leaning into the tech play, launching with an advanced hands-off, eyes-on driver assistance system (a la GM’s Super Cruise). Thanks to the beauty of over-the-air updates, the Afeela 1 will eventually offer Level 4 autonomy, leaning on a bevy of integrated sensors, including a roof-mounted LIDAR pod. That would theoretically mean you could drive in Gran Turismo 7 while your car drove you to work.
At launch, though, the driver will only be able to play GT and other PlayStation games while the car is parked, thanks to built-in Remote Play functionality. Beyond gaming, Sony is building out a comprehensive set of APIs that will enable third-party developers to create immersive in-cabin experiences, including custom gauge clusters and infotainment skins. All that will be run on a Qualcomm Snapdragon Digital Chassis, with enough power to handle both the evolving active safety and infotainment duties.
That all sounds fun, but the bad news is that would-be buyers are going to have to wait a little longer to start customising their rides. Production on the Afeela 1 has slipped, and it will not ship to the first customers until the very end of 2026. Those customers will have to be in California, as that’s the only state where the sedan will be for sale. Arizona will come next, sometime in 2027, but nobody at Sony Honda Mobility will say which state will get the next bite of this apple, or when.
Will it be worth the wait? We’ll circle back later with more detailed impressions from inside this year’s iteration of the Afeela 1.
This article originally appeared on Engadget at https://www.engadget.com/transportation/evs/sony-honda-mobilitys-afeela-prototype-2026-puts-an-suv-spin-on-its-too-familiar-sedan-at-ces-043927882.html?src=rss
Garmin’s big announcement for CES 2026 wasn’t another fitness watch. Instead, it revealed it is adding food (and calorie) tracking to its Connect app. It combines AI image recognition with a rich food database, so you can monitor your calorie and nutrient intake without leaving the app — and even log some food through its watches.
The food tracking works similarly to existing apps like MyFitnessPal, Noom, LifeSum and many others. Still, Garmin hopes to make its companion app the best place for tracking nutrition without having to leave its app – and tying it into your other fitness goals.
This makes sense in a few ways. If you already use Garmin to track your exercise, the app can provide a more accurate estimate of your calorie expenditure. It’ll personalize it based on height, weight and gender, which you’ve likely already entered on your Garmin profile. It can even offer calorie and macronutrient recommendations tailored to your fitness goals.
However, it unveiled the news at CES 2026, a tech conference held in Las Vegas, which seemed like a troll. Las Vegas, where you’re never far from a three-foot margarita, a BLT Bloody Mary or fried chicken and waffles. In fact, Garmin selected one of my favorite restaurants, Yardbird, to showcase its new nutrition-tracking feature. Yardbird does incredible chicken and waffles. It’s possibly the worst possible place to start monitoring my calorie intake, regardless of the 15,000-odd steps I take each day here.
In addition to the food database, you can use the camera to log food. With AI-powered image recognition, the Connect app makes a generally good guess at what you’re eating. During the media briefing lunch, the app swiftly recognized grilled salmon, corn and several other vegetables. The quantities were not remotely accurate, however: Five kernels of corn do not equate to one cup. You can manually adjust the serving size, but unfortunately, it can’t make a more informed estimate at this point. A Garmin spokesperson noted that the company is looking into how to improve accuracy here. (It’s a common issue for any app or service attempting to use AI image detection to analyse meals.) It was, however, quick and easy.
You can also create customized meal entries (with the correct ingredient quantities), making it easier to track your regular breakfast habits or midday coffee order. Better still, you can monitor nutrition and calorie intake on compatible Garmin watches. You can even log your favorite (and recently logged foods from your wrist without needing to use your phone. Voice commands also work on compatible Garmin wearables.
The feature is now available to Garmin Connect+ subscribers, who pay $6.99 per month. This also includes Active Intelligence, which already offers AI-powered insights into your workouts. It can now factor in your nutrition.
Subscribers will also be able to get daily, weekly, monthly and even annual reports on calorie targets (and what you ended up consuming). But let’s start monitoring that once I’ve left Vegas, please.
This article originally appeared on Engadget at https://www.engadget.com/wearables/garmin-food-tracking-feature-connect-plus-ces-2026-043231948.html?src=rss
OhSnap won our hearts (and a best of CES award) last year with the MCON, its tiny Bluetooth game pad that magnetically connects to your phone. At the time, we tried a pre-production unit, but in the year since the MCON has been finalized and went on sale last month. I just got a chance to try the final version here at CES, and just a quick demo was enough to convince me that OhSnap has made some solid refinements over the last year.
Just as we saw last year, the MCON attaches magnetically to an iPhone or any compatible Android phone (the company also includes a magnetic ring in the box for Androids that don’t have Qi2 yet). It has the full array of controls, including four face buttons, a D-pad, two joysticks and index finger buttons and full bumpers. It’s just sleeker and feels more solid than last year’s prototype, and while mounting your makes it feel a bit top-heavy, it’s not awkward enough to cause real issues when playuing (at least that’s how I felt after my brief demo).
The controller also has grips that unfold to help balance things, and the plate that attaches your phone to the controller has a kickstand. So if you want to drop your phone down on a tray table on a plane and play just holding the controller, feel free. All in all, the MCON feels like a flexible and high-quality device that’s probably worth the $150 OhSnap asks for it.
The company also has some new accessories to show off. There’s a $70 TV dock that powers your phone and outputs video from USB-C to HDMI. You could use this dock with any Bluetooth controller hooked up to your mobile phone, not just the MCON. There’s also a tiny $30 adapter you can put on your keychain that has a USB-C plug that goes into your phone; plug an HDMI cable into the other end and you can broadcast your games to a TV anywhere you go and play them with a controller.
It’s a pretty clever and comprehensive set of products, and while there are tons of mobile game controllers, the OhSnap feels like one of the more portable and sleek options out there. OhSnap is working on something even slimmer, though. The company showed off two new prototype controllers, the MCON Lite and MCON Slim.
As the names suggest, they’re both more compact than the original model. The Lite uses concave joypads rather than full sticks, and the Slim makes things even smaller by using concave touch pads. Your’e not going to get quite the same quality experience with those as you would with the bigger version, but they are noticeably thinner; OhSnap says they’ll be cheaper, too. The MCON Lite is scheduled to arrive this summer, while the Slim should launch in the fall (around the same time as the iPhone 18).
This article originally appeared on Engadget at https://www.engadget.com/gaming/another-look-at-ohsnaps-mcon-a-magnetic-game-controller-for-your-phone-042424092.html?src=rss
Every year at CES, Intel and AMD announce their newest processors. This year, Intel debuted its first line of chips made with its 18a process, which stands for 18 angstroms, or under 2nm. To make that a little less nerdy, that means these chips can fit a lot of tech into a small area, which means big gains in performance.
Colloquially called Panther Lake, the new Intel Core Ultra Series 3 chips are focused on laptops and mini-PCs. But that doesn’t mean they don’t put out impressive numbers. During its keynote, Intel promised “77% faster gaming performance,” and for everyone actually using their PCs to do work, “60% better multithread performance.” Impressively, it also said its new chips would also enable “up to 27 hours of battery life,” rivaling Apple’s M-series chips and other ARM chips from manufacturers like Qualcomm. Battery life has been a sore spot for both Intel and AMD for a while, and I’m excited to get my hands on machines that use Panther Lake processors.
And then there’s the AI. Intel says its top Intel Core Ultra Series 3 models have 50 NPU TOPs and 180 TOPS in total when you combine the NPU and GPU numbers. To translate, that basically means faster AI performance for developers who don’t want to bother with the cloud, and would prefer the speed and privacy of a locally downloaded AI model.
Speaking of that GPU, Intel’s upgraded its integrated graphics this generation to the Intel Arc B390, which has twice the cache of its prior GPU and 50% more cores. I already mentioned that 77% increase in gaming performance, but AI developers will also see a 53% increase in performance over Intel Core Ultra Series 2, and a two times performance increase over Intel Core Ultra Series 1.
So, better computers. That’s pretty par for the course for CES, but there is a “one more thing” here. Intel is coming for AMD’s dominance in handheld gaming PCs.
Until now, most handheld gaming PCs have used AMD chips, with those that have opted for Intel getting hit in reviews for buggy or poorly optimized performance. The Steam Deck uses an AMD chip, and so does the Xbox handheld. Intel says it’s going to change that.
After announcing the new integrated GPU and walking through features like frame generation and ray tracing, the company said it “will be launching an entire handheld gaming platform with Panther Lake.”
That means big moves, and soon—Panther Lake won’t last long before getting replaced. The company didn’t say much more, but it did show a slide with partners set to use Intel chips in their handhelds, including Acer, MSI, and hey, Microsoft. I guess we’ll see another Xbox handheld model soon.
And that’s about it for Intel this year. The new chips are smaller, stronger, and more efficient, but smartly, the company is also planning to use them to shore up its weaknesses, specifically in gaming and battery life. That makes sense. With developers like Apple having famously ditched Intel throughout the decade, the company has been on the backfoot.
According to Intel, the earliest machines powered by Intel Core Ultra Series 3 will start accepting pre-orders on January 6, with availability beginning on January 27.
New tech at CES often makes categorization hard. Is Cozyla’s latest device a digital family calendar, a TV or a smart home hub? After seeing it, I’d say the answer is, yes? Cozyla calls the Calendar + Max the “largest interactive hub in it’s category.” And indeed, it’s the size of a big flatscreen, but the main interface is just like any other Cozyla device, an Android-based family calendar and planner with shared schedules, notes, meal plans, and chores. It syncs with Google, Yahoo and Apple calendars as well as Outlook, and it will run apps from the Google Play store. (Yahoo owns Engadget, but has no say in our editorial coverage.)
But this super-sized version can also stream shows and movies with whichever apps you subscribe to. The photo integration, via the app, will take pics from your phone and stick them on the bigs screen. You can also create a slideshow and photo walls to look at when the calendar is off.
The smart home integration lets you view your security camera and video doorbell feeds on a much larger scale than your phone’s display can offer. The display has a camera of its own so you can make FaceTime and Zoom calls.
The Max comes on a built-in stand that you can roll around the house, along with a battery that should power the whole thing for around six hours. When I first heard about it, I found myself wondering where I would put Calendar+ Max in my house. The kitchen is an ideal centralized calendar location, but I don’t want to watch Miss Scarlet in there. Luckily, the wheels make that conundrum moot. This can go anywhere you want it to, and swings into portrailt orientation that looks great for chore mode, and landscape orientation for watching shows.
Possibly the best part is the lack of subscription required. Like with all Cozyla screens, there’s no ongoing cost to use all the features. The Calendar + Max has no release date or price available just yet, but if I ever decided to organize my family’s life beyond barely contained chaos, I can see myself considering it.
This article originally appeared on Engadget at https://www.engadget.com/home/smart-home/cozyla-showed-off-a-4k-55-inch-digital-family-calendar-at-ces-040352802.html?src=rss
Intel will formally launch its first Core Ultra Series 3 laptop processors later this month, the company announced at its CES keynote today. Codenamed Panther Lake and targeted, at least for now, at high-end ultraportable PCs, the Core Ultra 3 chips will also be the first to use Intel’s 18A manufacturing process, the company’s effort to catch up with the chip manufacturing technology of Taiwan Semiconductor (TSMC).
The launch will start with 14 chips across 5 product families, which Intel says will be used in “over 200” PC designs. The first of these will be available on January 27th, with others following “throughout the first half of this year.”
Credit:
Intel
Credit:
Intel
We wrote about the basic building blocks of Panther Lake when Intel released details late last year. In many ways the chip is a retreat from the Lunar Lake design, sold as Core Ultra 200V, which used chiplets manufactured mostly outside the company and on-package RAM rather than memory in a DIMM slot or soldered to the mainboard. At the time, Intel said these moves were made in the interest of saving power and extending battery life, as were decisions like removing Hyperthreading support from the P-cores.
When you come across an AI video on Instagram, or watch ChatGPT respond to your query, do you ever think about how that content was generated? Beyond the actual programs and prompts, generative AI takes an enormous amount of compute to support, especially as it skyrockets in popularity. As such, AI companies are looking for more power than ever, which means, of course, turning to those that make the hardware.
During a Monday evening keynote, AMD’s CEO Dr. Lisa Su showed off the hardware that will soon power everything from ChatGPT to the AI videos overwhelming your feeds. Su introduced “Helios” against a backdrop of dramatic music, the company’s upcoming AI rack, that packs a staggering amount of computing power into a rack that weighs nearly 7,000 pounds.
Each “cross-section” of these racks, if you will, is powered by four key AMD pieces of hardware: The company’s new AMD Instinct MI455X GPU, the new AMD EPYC “Veince” CPU, the AMD Pensando “Vulcano” 800 AI NIC, and the AMD Pensando “Salina” 400 DPU. There are some staggering stats here: Helios is capable of 2.9 exaflops of AI compute, and comes with 31 TB of HBM4 memory. It offers 43 TB per second scale out Bandwidth, and is developed with 2nm and 3nm architecture. The rack has 4,600 “Zen 6” CPU cores, and 18,000 GPU compute units. In other words, this isn’t your average piece of hardware.
Su’s pitch is that the AI industry is in need of this additional compute power. She notes how the world used one ZettaFlop of computing power in 2022 on AI technology, compared to 100 ZettaFlops in 2025. (For the curious, one ZettaFlop has a value of 10 to the power of 21.) It’s no surprise: AI is everywhere, and many of us are using it—whether we know it or not. Some of us are using it overtly, generating AI videos or running chatbots daily. But others are using AI quietly embedded in functions, like live translation.
Su welcomed reps from OpenAI, maker of ChatGPT, and Luma AI, which creates generative AI video content, to talk about how additional compute helps their programs. But during Luma AI’s demonstration of its hyperrealistic video generations, all I could think about was how this type of content is already tricking people into thinking its real, when it’s entirely fabricated—not to mention the impact on human artists. AMD is optimistic about AI, and the data centers powering it, but critics have been pushing back, citing concerns with the impacts on the communities companies are building these data centers in.
Helios will likely be a major success for AMD, but it comes at an interesting time for tech, and AI in general. AI is more popular than ever, but it’s also more controversial than ever. I see hardware like Helios only fueling the fire in both directions.
In addition to Helios, Su announced the AMD Ryzen AI 400 series. These newest chips comes with either 12 “Zen 5” CPU cores and 24 threads, 16 RDNA 3.5 GPU cores, a 60 TOPS XDNA 2 NPU, and memory speeds of 8,533 MT/s. AMD says the Ryzen AI 400 series is 1.7 times faster at content creation and 1.3 times faster at multitasking whe compared to Intel Core Ultra 9 288V.
These new chips will ship soon in a number of major PC brands, including Acer, Asus, Dell, HP, Lenovo, Beelink, Colorful, Gigabyte, LG, Mechrevo, MSI, and NEC.
Intel, AMD, Nvidia, and other chip companies usually have some kind of news to announce at CES to kick off the year, but some of those announcements are more interesting than others. Sometimes you see new chips with significant speed boosts and other new technologies, and sometimes you get rebranded versions of old silicon meant to fill out a lineup or make an existing architecture seem newer and more exciting than it is.
AMD’s Ryzen CPU announcements this year fall firmly into the latter camp—these are all gently tweaked variants of chips that launched in 2024 and 2025.
Credit:
AMD
Credit:
AMD
Credit:
AMD
Credit:
AMD
Let’s start with the Ryzen AI 400 series. Officially the follow-up to the Ryzen AI 300 chips announced in June 2024, these processors offer some modest clock speed improvements and faster memory support. The new Ryzen AI 9 HX 470 has a peak boost clock speed of 5.2 GHz and support for LPDDR5x-8533, for example, up from 5.1 GHz and LPDDR5x-8000 for the Ryzen AI 9 HX 370, and its built-in neural processing unit (NPU) is capable of 60 trillion operations per second (TOPS) rather than 50 TOPS.
That’s
The centerpiece
For as long as I can remember, I’ve always wanted to have a keyboard computer — that is, a computer where the entire machine was stuffed inside of a keyboard. Maybe I caught a glimpse of the Commodore 64 at an impressionable age, but for whatever reason, the idea has always intrigued me. Now HP is bringing the concept back with its new EliteBoard G1a “Next Gen AI PC.”
It’s an IT administrator’s dream: It looks a typical desktop keyboard, but it has the full power of a Copilot+ AI PC inside. You can equip it with Ryzen 5 or 7 CPUs and their embedded Radeon 800 GPUs, up to 64GB of RAM and as much as 2TB of NVMe SSD storage. All you need to do is add a monitor and a mouse, and you’ve got a full-fledged desktop setup.
The more I think about it, the more sad I am that the arc of the computing industry trended towards standardized desktops and laptops. There was a brief spark of interest with the UMPC (ultra mobile PC) trend in the 2000’s, which Engadget covered extensively as a young blog, as well as ASUS’s Eee keyboard. But they couldn’t survive the rise of the smartphone and tablet. It turns out putting an entire computer behind a screen was more compelling than stuffing them into a keyboard.
I was able to test out an early EliteBoard prototype, and while the experience wasn’t perfect, it’s still an intriguing computing option. I had trouble setting it up initially because it only had two USB-C ports on its rear, which meant I had to find a way to power it through one port and pass a video signal through the other. Thankfully, my Anker USB-C charging hub was able to juice it up, and I also had a USB-C hub with an HDMI port, which allowed me to connect to my monitor.
Sadly, the overall setup was a jumble of wires, and not the clean layout I expected from a keyboard PC. Once I was able to start up Windows though, I was less annoyed and more amazed that the keyboard contained an entire computer. I suppose I shouldn’t be too surprised, as Intel’s Compute Stick placed a functional PC in a much smaller case, but unlike the failed product, the EliteBoard actually felt usable. I was able to load up several browser windows with tabs, edit a few photos and even play a few light games, like Vampire Survivors. And yes, typing on it felt pretty great too.
Since I was testing out prototype hardware, I agreed not to benchmark the EliteBoard. But from the performance I saw, it felt about the same an entry-level laptop. Basically, it’s perfectly suited to its main task: Being a boring office computer. Back in my IT days, I certainly would have preferred deploying a few light keyboards instead of the tank-like Dell desktops we typically ordered.
While the EliteBoard is targeted at commercial users, HP is considering it an experiment to see how people respond to a keyboard PC. There’s a chance we could see one that’s eventually meant for mainstream consumers. I’m not sure if that’s exactly, necessary, though. The whole concept of a desktop PC mainly appeals to tinkerers and IT folks. And for anyone who wants to get their hands on an EliteBoard soon, there’s really nothing stopping you.
This article originally appeared on Engadget at https://www.engadget.com/computing/hps-eliteboard-made-me-believe-in-keyboard-computers-again-033000955.html?src=rss
The whole AI PC trend didn’t exactly set the world on fire last year, but, like clockwork, AMD is still ready to deliver a new batch of AI chips at CES 2026. The Ryzen AI 400 processors will offer some slight speed upgrades over last year’s chips, and notably, they also include AMD’s first Copilot+ processors for desktops. Sure, the Copilot+ program didn’t really go anywhere, but as I’ve argued, it at least served as a template for building capable AI PCs. Now we just need some genuinely useful AI features in Windows — Recall and Copilot’s voice commands aren’t really compelling enough on their own.
AMD’s first AI desktop chips, the Ryzen 8000G series, arrived in 2024 with relatively underpowered neural processing units (NPUs) for AI tasks. The Ryzen AI 400 chips, on the other hand, feature 60 TOPS XDNA 2 NPUs (up from the 50 to 55 TOPS in Ryzen AI 300 hardware). That places them well above the 40 TOPS NPU minimum for Copilot+ systems. For most consumers, NPU speeds don’t really mean much yet, but if you’re running AI models on your system you can expect slightly faster inferencing from AMD’s previous chips.
The top-end Ryzen AI 9 HX 475 offers up to 12 Zen 5 CPU cores, 5.2GHz max boost speeds and up to 8,533 MT/s memory speeds. The line scales down to the four-core Ryzen AI 5 430, but even that model supports speedy 8,000 MTS RAM and offers a 50 TOPS NPU.
AMD isn’t giving us many specific details on the Ryzen AI 400 chips at CES, but broadly, it claims they’ll offer up to 30 percent faster multi-tasking, 70 percent faster content creation and 10 percent faster gaming than its previous chips. The company also says you’ll see 70 percent better “unplugged connectivity” on Cinebench nT, which is hopefully a sign that you’ll see improved performance overall on battery.
This article originally appeared on Engadget at https://www.engadget.com/computing/amds-ryzen-ai-400-chips-are-a-big-boost-for-laptops-and-desktops-alike-033000635.html?src=rss
While it’s nice to see desktop support in AMD’s new Ryzen AI 400 chips, demanding gamers and enthusiasts will likely be more intrigued by the company’s next batch of Ryzen AI Max+ chips, as well as the new Ryzen 7 9850X3D with 3D V-Cache. The former will make its way into small desktops and a handful of workhorse laptops, while the latter is another option for gamers who want the speed bump of 3D V-cache without shelling out for the $700 9950X3D.
Last year, AMD debuted its Ryzen AI Max chips as a way to create a single piece of silicon with powerful CPU cores, GPU cores, NPUs and integrated memory, similar to Apple’s home-brewed chips. At the time, AMD VP Joe Macri also noted that the existence of Apple Silicon helped make the Ryzen AI Max chips possible.
“Many people in the PC industry said, well, if you want graphics, it’s gotta be discrete graphics because otherwise people will think it’s bad graphics,” Macri said at last year’s CES. “What Apple showed was consumers don’t care what’s inside the box. They actually care what the what the box looks like. They care about the screen, the keyboard, the mouse. They care about what it does.”
At CES this year, AMD is unveiled the 12-core Ryzen AI Max+ 392 and eight-core Ryzen AI Max+ 388. Both chips feature boost speeds up to 5GHz, 50 TOPS NPUs and GPUs capable of 60 TFLOPs. We’ve seen the earlier Ryzen AI Max chips in the Framework Desktop and the ROG Flow Z13, and we were generally impressed with its performance. For small systems, it was powerful enough that we really didn’t miss having dedicated GPUs.
As for the Ryzen 7 9850X3D, it’s an 8-core chip that can reach up to 5.6GHz boost speeds with 104MB of combined L2 and L3 cache. Like all of AMD’s X3D chips, it uses 3D V-cache technology to vertically stack additional cache memory. In comparison, the standard 9850HX chip has 76MB of L2 and L3 cache.
AMD says the new Ryzen AI Max+ chips and the 9850X3D will ship in the first quarter. There’s no pricing information on the latter, yet, but recent leaked listings suggest it may go for around $200. Rumors also point to a massive dual-cache (192MB!) 9950X3D2 chip coming soon.
This article originally appeared on Engadget at https://www.engadget.com/computing/amds-new-ryzen-ai-max-chips-and-ryzen-7-9850x3d-court-desktop-enthusiasts-at-ces-2026-033000587.html?src=rss
After relaunching its Area-51 brand with a new 16-inch laptop in 2025, Alienware is teasing two new laptops at CES 2026 in an attempt to reach an even larger audience. Dell’s gaming brand plans to launch both entry-level and ultra-slim models later this year, and in the meantime, it’s rolling out updates to its existing desktops and laptops to tide people over.
Alienware isn’t sharing all the details of its new laptops at CES, but we do have the highlights. The company’s new “ultra-slim gaming laptop” will be around 17mm or 0.67 inches thin, and come in either 14-inch or 16-inch variants. While the 16-inch version of the laptop will feature NVIDIA discrete graphics and “new highly efficient CPUs,” it’s not meant to be a gaming powerhouse like the Area-51. Instead, Alienware suggests the laptop will work for gaming and also “creative projects, productivity and everything in between.”
The entry-level laptop is similarly not at Area-51-levels of power, but Alienware claims it’ll deliver “strong gaming performance” at its “most accessible price point yet.” That should ideally put the new laptop under the $1,199 starting price of the more streamlined Alienware 16 Aurora laptop.
On top of those two new models, Alienware is bringing new anti-glare OLED panels to a selection of its Alienware 16X Aurora and Alienware 16 Area-51 laptops, along with new Intel Core Ultra 200HX chips. The new display panels reach 620 nits of peak HDR brightness and have a 0.2ms response time for even smoother gameplay. The Alienware 18 Area-51 is also getting an upgrade to Intel Core Ultra 200HX chips, while the Alienware Area-51 Desktop will ship with AMD’s new Ryzen 7 9850X3D chips.
There’s currently no pricing available for Alienware’s new laptops, or the updated versions of its older models. The updated Alienware 16X Aurora, Alienware 16 Area-51, and Alienware 18 Area-51 laptops will be available in Q1 2026. The update Alienware Area-51 Desktop is coming in February 2026.
This article originally appeared on Engadget at https://www.engadget.com/computing/laptops/alienware-teases-new-super-slim-and-entry-level-gaming-laptops-at-ces-2026-033000700.html?src=rss