Welp, it finally happened. We complained—and Dell listened. Dell’s XPS laptops have officially risen from the grave to haunt CES 2026 in Las Vegas, after Dell unceremoniously kicked the brand (and all others) to the curb in 2025 for a simplified lineup. Honestly? They’re looking pretty darn slick—I know I’m glad to see them again, and I’m convinced I’m not the only one. The XPS 14 and XPS 16 are light, sleek, and pack both aluminum and Gorilla Glass. It’s the kind of polish I’ve come to expect from this brand, and Dell appears to have delivered on that front. Visually, Dell has fully leaned into the XPS identity again. The minimalist design is cleaner than previous iterations, with the XPS logo boldly staring out from the front lid, making it all the more recognizable. The real kicker? The performance. These laptops pack Intel’s new Core Ultra Series 3 processors and Intel Arc graphics, which Dell says are up to 78 percent faster for AI tasks and more than 50 percent faster for graphics than before. There’s a big jump in battery life this year, too. With the LCD configuration of the laptop, Dell claims up to “27 hours of regular use or 40+ hours of local video playback.” Dell And yet, somehow, they’re still surprisingly light. The XPS 14 weighs just three pounds. The XPS 16? Around 3.6 pounds. That makes them Dell’s thinnest XPS laptops yet. Speaking of the display, this is a nice highlight that’ll appease the movie aficionados among us. You can pick the 2880×1800 OLED version (with Dell’s tandem technology) if you long for a truly beautiful picture, or you can stick with the more power-efficient 2560×1400 LCD model. Both have InfinityEdge screens and a 1–120Hz variable refresh rate, so your stuff should look crisp and multitasking shouldn’t overwhelm it. Dell also plans on releasing a brand new XPS 13 in late 2026. Not only will it be smaller and lighter than previous versions, but its price will be more accessible as well. Neat! Dell’s XPS 14 and 16 arrive January 6, 2026, starting at around $1,650 and $1,850. More options are coming in February. Read...

TCL is making some bold claims about its new SQD-Mini LED Series at CES, with the company crowing that its flagship X11L “ends the debate” between OLED’s superior blacks and LED’s superior brightness. The company promises that its SQD Mini-LED achieves 100 percent of BT2020 color (that’s very good), while noting that its technology is largely devoid of the color crosstalk, color bleed, and color blooming that Micro RGB can suffer from. Adding to the X11L’s luster is an uber-thin 0.8-inch design, with a completely flat back for a superior flush mounting experience. The perfectly flat back of the X11L makes wall-mounting nearly flush. TCL, like all the other TV vendors at CES, is touting the advanced AI capabilities of its TCL Super Resolution AI processor, which “ensures accuracy in every area” by “leveraging enhanced AI color, AI contrast, AI clarity, AI motion, AI upscaling, and AI sound.” In other words, TCL says it has improved the processing speeds and algorithms on its TV’s chips, resulting in better picture quality. Integrated Google TV with Gemini is featured to enhance voice control and offer help in finding content. Sound features include Audio by Bang & Olufsen and—of course—expansion via TCL’s own surround components, which include a wireless subwoofer and Dolby Atmos FlexConnect speakers. As you might expect, all this goodness won’t be cheap: the 75-inch model is $7,000, the 85-inch is $8,000, and the wall-filling 98-inch model is $10,000. But the best always costs more, and while we haven’t actually seen the X11L, the specs are impressive indeed. Read...

Micro RGB is shaping up to be the TV industry’s overriding theme at CES 2026, and Samsung is going all in. The new technology resides in the middle ground between increasingly mainstream mini-LED TVs and still ludicrously expensive micro-LED sets. Where micro-LED TVs use self-emissive pixels that can be turned on and off individually–much like an OLED TV–micro RGB TVs still rely on a backlight. But that backlight consists of red, green, and blue LEDs that reduce the need for a color filter. It’s a significant improvement on conventional LCD TVs, but it still relies on LCD shuttering. A host of Micro RGB TVs Samsung plans to ship eight new Micro RGB TVs in 2026, including a massive 130-inch model in its R95H series as well as the 115-inch model MR95F. The TV manufacturer is also showing 65-, 75-, and 85-inch models in its R95H series, along with 100-, 98-, and 85-inch models in the step-down R85H series. Some Samsung TVs, such as the newly reissued Frame Pro, will be compatible with the minty fresh wireless version of Samsung’s One Connect breakout box, which previously used a thread-like optical cable to connect to the TV. That leaves you need to hide just the TV’s power cord. It’s not quite the magic of DisplaceTV series, but Samsung is getting there. QLED Samsung is also showing a new giant-sized model of its The Frame TV that measures 98-inches, just in case your local art gallery has some extra large digital creations it wants to display. The company has also added a 55-inch model to its The Frame Pro lineup, which provides a brighter picture. OLED Samsung hasn’t forgotten its stellar RGB OLED series either. It’s refreshing the S95H series, which offers modest improvements to our favorite “puppies on black velvet” (i.e., fantastic black and warm color) TV viewing experience. AI and other stuff Samsung’s AI Sound now creates “stems,” aka separate music, voice, and audio effects streams. This is a feature that’s been around in the pro audio arena for a while, but not in real time. The feature is said to let you mute the announcers while still hearing the actual game and crowd noise. Now that’s progress! The same algorithms are used to create a more immersive sports audio experience called AI Soccer Mode, where the effects are boosted to make it seem more like you’re sitting in the stadium. Google Photos The company is also working to integrate Google Photos into its smart TVs for, so users can interact with their personal images, and manage their photo libraries directly from their TV. Speaking of such, my favorite quote from the show so far is from the president of Samsung’s display division, Seok Woo Yong: “Samsung TVs are not just screens. They are entertainment companions….” Err… In a world that’s moving away from live social interaction, that’s exactly what we need our TVs to be. Read...

Beatbot builds some of the most sophisticated—and the most expensive—robot pool cleaners on the market, and it’s once again raising the stakes at CES 2026 with the introduction of its AquaSense X. This doesn’t look like just an incremental advance over the TechHive Editors’ Choice-winning AquaSense 2 Ultra, it’s Beatbot’s first foray into self-cleaning technology, courtesy of a key component that lives on land. The Beatbot AquaSense X AI Robotic Pool Cleaner looks and feels a lot like 2025’s Ultra, but the addition of the new Beatbot AstroRinse Cleaning Station is an entirely new concept, not just for Beatbot but for the industry as a whole. This box-like docking station is a sizeable device atop which the AquaSense X robot rests when not in use. Once placed on top of the device, the AstroRinse gets to work, flushing the AquaSense’s filter during a three-minute cycle. Debris is captured in the 23-liter basket underneath; Beatbot says for a typical pool owner, the disposable AstroRinse debris bag will need to be replaced only once every two months (based on a cycle of two cleanings per week). The AstroRinse also recharges the unit after the cleaning cycle is complete. The Beatbot AquaSense X, sitting atop its self-cleaning station. Beatbot As for the Beatbot AquaSense X robot itself, while it looks familiar, many of its systems have been upgraded under the hood. Its AI vision detection system now recognizes 40 types of debris (up from 20 on the Ultra), allowing the unit to search for debris on the pool floor instead of simply traversing the area on a preset path. The new detection system can also search for debris on the surface of the pool: Like its predecessor, the AquaSense X can rise to the surface and float, operating in a skimmer mode as part of its cleaning cycle. The unit also has improved detection of permanent pool features, such as steps and ledges, thanks to dual ultrasonic sensors on the underside of the robot. Lastly, the unit now integrates with various smart home environments—including Amazon Alexa, Apple Siri, and Google Assistant—allowing for voice-activated operation in addition to control via Beatbot’s mobile app. The AquaSense X will be available on January 5, 2026 at a price of $4250. The first 500 customers who preorder a unit will receive a $250 discount. If you’re attending CES in person and would like to see the AquaSense X in action, the robot will be on display at Beatbot’s booth: Venetian #53852. This news story is part of TechHive’s in-depth coverage of the best robotic pool cleaners Read...

Qualcomm’s Snapdragon X2 Plus processor has the same goal as its predecessor: to provide a slightly stripped-down, leaner alternative processor for midrange Windows on Arm laptops. But there’s a slight difference. While last year’s X Plus chip came in two flavors: a 10-core and an 8-core offering, Qualcomm is offering customers either a 10-core or 6-core Snapdragon X2 Plus chip instead. Qualcomm is eyeing Intel’s Core Ultra 200-series chips as well as the AMD Ryzen AI 300 series as potential targets. Qualcomm’s Plus series of processors essentially fill the same role as a Core i5: Qualcomm uses the same CPU architecture, plus the same GPU (adjusted for power, which lowers the speed) and the same NPU. But Qualcomm is being way more aggressive in terms of the performance comparisons. Eighteen months ago, Qualcomm was claiming that its X Plus chip offered 28 percent faster performance at the same power. This time, it claims that the top 10-core chip in the X2 Plus family delivers over three times the performance of the competition, the Core Ultra 7 265U. Qualcomm’s first-generation Plus processors appeared in the Acer Swift 14 AI, the Dell Inspiron 14 Plus, and the Dell Latitude 7455, among others. The company will disclose more customers here at the CES show in Las Vegas. Executives said to expect the X2 Plus to land in the same price range as the X Plus, or devices costing about $799. Essentially, the mission of the X2 Plus is the same as Qualcomm’s first generation, “to really get the same technology advantages to more and more users,” said Mandar Deshpande, senior director of product management for the compute business at Qualcomm, in a press briefing with reporters. How fast is the Snapdragon X2 Plus? Officially, the two Snapdragon X2 Plus chips are known as the 10-core X2P-64-100 and the 6-core X2P-42-100, reflecting Qualcomm’s impenetrable naming scheme. Typically, you’ll see both chips referred to as the 10-core X2 Plus and the 6-core X2 Plus. The 10-core X2 Plus will run at speeds up to 4.0GHz, with an X2-45 GPU that Qualcomm hasn’t disclosed the speeds of. The 6-core X2 Plus also runs at that same clock speed and features that same GPU as well. Both chips also include identical NPUs with 80 TOPS. Why six cores? Deshpande shrugged off the question. “Fundamentally, we’re not really tied to core counts, frequencies, or [power],” he said. Qualcomm’s current list of processors does not reflect the new additions, but you can see how they’ll fit in: probably a touch below the older Snapdragon X Elite, and with similar specifications to the first-gen Snapdragon X Plus chips, at least on paper. Those chips also included an additional two processors with some variations in clock speed; Deshpande declined to comment when asked if Qualcomm would do the same with the X2 Plus. We also don’t know the clock speeds of the GPUs. Though the Snapdragon X2 Plus and the Snapdragon X1 Plus look similar on paper, the numbers may be deceiving. Like the Snapdragon X2 Elite, the X2 Plus is built on updated third-generation Oryon CPU cores, which increase performance. While the Snapdragon X Elite included a pair of cores that could run at a faster “turbo speed” to accomplish high-priority tasks quicker, the X Plus did not, and the X2 Plus does not either. On the other hand, the X2 Plus base clock speed is as high as the turbo speed was in the X1 Plus, a Qualcomm representative noted. One of the key improvements for Qualcomm’s Snapdragon X2 Elite was its inclusion of “performance” and “prime” cores, also tuned for high-performance work. Those, the X2 Plus has. The ten-core X2 Plus has six prime cores and four performance cores, while the six-core X2 Plus includes six prime cores, the representative added. Still, comparing the X1 Plus to the X2 Plus, performance improves significantly: 35 percent more single-threaded CPU performance, and up to 39 percent more GPU performance, using 43 percent less power. (Qualcomm used Geekbench and 3DMark’s Steel Nomad tests for the comparisons.) Qualcomm is projecting significant gains over the other mobile processor platforms competing against it, which it classifies as some of Intel’s Core Ultra 200 7-class processors as well as the AMD Ryzen AI 7 350. Like the Snapdragon X2 Elite, the X2 Plus chips run at full speed on battery power as well when the laptop is plugged in, a key differentiator from the competition. Naturally, Qualcomm is comparing itself to the existing “Lunar Lake” and “Strix Point” platforms from Intel and AMD; Intel has already announced its next-gen “Panther Lake” mobile platform, and AMD is expected to debut its “Gorgon Point” mobile platform at CES. None of these chips have shipped, however, so Qualcomm can’t test against them. Qualcomm didn’t quote any battery-life estimates for Snapdragon Plus notebooks; the company said that it is still working with customers to optimize those products. A big question mark continues to be the utility of the NPU, and how receptive applications partners and customers will be to using local AI on the PC. Since the X2 Plus includes the same 80 TOPS NPU as the X2 Elite, that will allow more NPU-powered features from Microsoft Windows, Da Vinci Resolve, and other apps to run simultaneously, much in the same way that a powerful CPU or GPU allows multiple apps to be run simultaneously. The chip’s memory bandwidth is also up to 152Gbps, which gives enough headroom for LLMs. If and when an NPU-powered AI future comes to pass, Qualcomm will be ready. As far as games, Deshpande said that the company has tested 1,400 games on the Snapdragon platform, and that they’re running “really good” — where that means at 60Hz on a 1080p resolution. Microsoft published an update to its Prism emulator in mid-December, which provides some additional support. While Qualcomm is encouraging its partners to write native Arm code that can boost performance, the updated Prism emulator will allow a wider range of applications to simply run on Windows on Arm, without specifying exactly what. Qualcomm is exploring in other ways, too. “One thing now that we’re working closely with the OEMs is really now innovating on the form factor front, right?” Deshpande said. “So we’re trying to make the PCs thinner and lighter. We’re looking at panelist devices. We’re looking at desktops, which can be made more portable. So all of that form factor innovation is very exciting, and you should expect those types of devices coming in the market very soon.” Read...

AMD is launching the Ryzen AI 400 at CES 2026 as the next chip in the company’s CPU roadmap, with what appears to be a similar goal as the current Ryzen AI 300: Aim high on CPU performance, but with sufficient AI TOPS and battery life to attract mainstream laptop buyers, too. AMD said that the Ryzen AI 400 chips will power both Copilot+ laptops as well as “socketed desktops,” bringing their AI capabilities to desktop PCs as well. AMD also announced “Pro” configurations of most of the chips, designed to power enterprise PCs. AMD executives didn’t refer to the AI 400 by its expected codename, Gorgon Point, but the chip’s specs matched up with a leak inadvertently published last year: up to 12 cores and 24 threads using the Zen 5 architecture, with a boost clock that can hit 5.2GHz. The Ryzen AI 400 will achieve 60 AI TOPS, AMD promises, with 16 RDNA 3.5 GPU cores. On paper, that’s very similar to the current Ryzen AI 300 chip, which in PCWorld testing of the Ryzen AI 300 emerged as a surprisingly powerful competitor to the Intel Core Ultra 200 “Lunar Lake” as well as the Qualcomm Snapdragon X Elite: somewhat comparable in battery life, but at the top of the heap in CPU benchmarks. Rahul Tikoo, senior vice president and general manager of AMD’s client business, said in a call with reporters that AMD aimed for “leadership performance across the CPU, the GPU, and the NPU,” plus “multi-day mobility” as well as “AI performance to enable the next wave of experiences.” The new Ryzen AI 400 series has higher CPU and GPU boost clocks, a higher supported memory speed, and extra TOPS. Still, it’s close enough to the Ryzen AI 300 that reporters asked if it was just “rebadged” silicon. It’s not, according to Rakesh Anigundi, director of product management at AMD, though improved performance arrives via improved firmware as well as manufacturing changes. The process technology used in the AI 400 is 4nm, or basically the same process technology used in the Ryzen AI 300. AMD is launching a total of seven Ryzen AI 400-series chips, ranging from a specialized Ryzen AI 9 HX 475 processor at the top of the stack, designed for gaming, down to the Ryzen AI 5 430 at the low end. The cores will be a mix of the full-fledged Zen 5 cores as well as the more efficient Zen 5c cores, in various configurations differentiated by core count, clock speed, and the number of graphics CUs as well as their speed. All of the Ryzen AI 400 chips tolerate anywhere from 15 to 54 watts of thermal design power, or TDP. AMD’s new Ryzen AI 400 chips include substantially more offerings than the original AI 300 series, which didn’t reveal the base clock speed at launch. It’s also interesting that all of the processors run at 2.0 GHz, but boost to different speeds. And yes, the Ryzen AI 9 HX 475 and 470 are nearly identical, save for the difference in NPU TOPS: 60 versus 55 TOPS. Ryzen AI 9 HX 475: 12 cores/ 24 threads, 2.0GHZ base clock/5.2GHz boost clock; Radeon 890M/16 CUs/3.1GHz Ryzen AI 9 HX 470: 12 cores/ 24 threads, 2.0GHZ base clock/5.2GHz boost clock; Radeon 890M/16 CUs/3.1GHz Ryzen AI 9 465: 10 cores/ 20 threads, 2.0GHZ base clock/5.0GHz boost clock; Radeon 880M/12 CUs/2.9GHz Ryzen AI 7 450: 8 cores/ 16 threads, 2.0GHZ base clock/5.1GHz boost clock; Radeon 860M/8 CUs/3.1GHz Ryzen AI 7 445: 6 cores/ 12 threads, 2.0GHZ base clock/4.6GHz boost clock; Radeon 840M/4 CUs/2.9GHz Ryzen AI 5 435: 6 cores/ 12 threads, 2.0GHZ base clock/4.5GHz boost clock; Radeon 840M/4 CUs/2.8GHz Ryzen AI 5 430: 4 cores/ 8 threads, 2.0GHZ base clock/4.5GHz boost clock; Radeon 840M/4 CUs/2.8GHz The question is whether the Ryzen AI 400 will remain at the top of the heap in terms of performance. In this, AMD was somewhat vague, claiming that in “responsive multitasking,” such as running a Microsoft Teams call with background blur enabled, the Ryzen AI 400 was 1.3X faster than the competition, or 1.7X faster in content creation. In this, AMD can only compare to the silicon its competitors have shipped; in this case, it refers to Intel’s Lunar Lake or Core Ultra 200 silicon. Head-to-head comparisons will have to wait until both companies ship their silicon in early 2026; AMD is claiming that Asus, Acer, Dell, HP and Lenovo, among others, have signed up. AMD said laptops from its customers would be available beginning in the first quarter, from thin-and-light laptops to gaming and content-creation PCs to, yes, desktops. Still, AMD’s benchmarks are impressive, both in content creation and in gaming. In gaming, however, AMD isn’t saying whether the games listed are technically “playable” (over 60 frames per second, generally) or whether any frame enhancement technologies were used. In a sense, however, its a win for gamers just to be able play some of these games on integrated graphics. (AMD’s configuration notes say that the Ryzen AI 9 HX 470 was used, with its integrated Radeon 890M GPU.) Clearly, the increased clock speed and NPU TOPS will be of benefit to consumers. However, the support for faster DRAM — 8533 MT/s — will be dependent on PC makers actually finding and buying that high-speed DRAM to add to customer devices. What’s not clear is whether AMD will be able to increase its market share in mobile, as it has done in desktops with its superb Ryzen X3D parts. Traditionally, AMD has held on to about 20 percent of the mobile market, according to analysts. “With this range of processors, OEMs can deliver AI PCs that are tailored to your specific need, while offering the best performance and robust on-device AI,” AMD’s Tikoo said. Read...

Kwikset has unveiled the Aura Reach at CES, a smart lock designed for homeowners seeking a flexible, connected security solution that’s easy to install and use. The Aura Reach supports both Matter and Bluetooth, enabling broad compatibility with smart home ecosystems including Amazon Alexa, Google Home, and Apple Home (lacking NFC technology, however, it does not support Apple Home Key). As with other locks in Kwikset’s portfolio, the Aura Reach features the company’s patented SmartKey technology, which allows users rekey the lock in seconds using an existing Kwikset key and a simple tool. Once that’s accomplished, the same key can be used to unlock every other Kwikset SmartKey lock in the home. It also protects against advanced break-in techniques like lock bumping and picking. The smart lock includes several practical features aimed at simplifying daily access. The LED-backlit keypad lights up when users approach, while geofencing and Bluetooth technology enable automatic unlocking for a hands-free experience. Users can also manage up to 250 access codes, review entry history, and receive alerts via the Kwikset app. The Aura Reach uses Kwikset’s patented SmartKey Security system. Kwikset The lock’s Matter compatibility includes support for Matter-over-Thread, the low-power mesh networking protocol that helps extend battery life and improve responsiveness. You’ll need a Matter-to-Wi-Fi bridge or a Thread border router to bridge this lock to your Wi-Fi network, but most modern smart speakers (Amazon Echo, Apple HomePod, and Google Nest models, as well as the Apple TV 4K) that fulfill that role. Installation is designed for DIYers, requiring only a screwdriver and guided by step-by-step instructions within the Kwikset app. Should you experience range issues with the lock’s Thread radio, Kwikset says a Signal Range Boost setting in its mobile app will boost the radio’s output by a factor of two. The Aura Reach is ANSI/BHMA Grade 2 certified and is available in Satin Nickel and Matte Black finishes. Kwikset has set an MSRP of $189 for the lock, and is available now at Kwikset, Amazon, Home Depot, Lowe’s, and other retailers. This news story is part of TechHive’s in-depth coverage of the best smart locks. Read...

How many new gadgets did you get this past holiday season? I bet most of them didn’t even come with charger blocks (also known as power bricks or power adapters), right? Well, no worries. They did you a favor. Now you can get a proper charger block that’s actually fast, compact, and power-efficient! Like this 65W Anker Nano charger that’s only $26 right now on Amazon. That’s a lovely 35% off! View this Amazon deal This is a super-small wall plug that’s built with GaN II technology, which means it’s able to be a lot smaller and produces less heat and charges faster and packs easier for travel, all compared to older power adapters. With three USB ports, you can fast-charge up to three devices at once—your phone, tablet, and earbuds, for example. How great is that? Delivering up to 65W of charging power, this Anker Nano charger block is even suitable for use with laptops, including some MacBook devices. Of course, if you’re using multiple USB ports at once, that total power will be distributed across the ports in use, with the top USB-C port taking priority and delivering the most charging power. You can now leave your massive power adapters at home and take this one with you on trips, and the cherry on top is that the prongs are foldable, meaning they won’t scratch the other stuff in your bag nor will the prongs risk getting damaged themselves. Hurry up and grab this powerful Anker Nano USB charger for $26 while you can! Save 35% on this fast-charging 3-port USB wall plug chargerBuy now at Amazon Read...

Lockly has introduced the Affirm smart lock series at CES, its first line of smart deadbolts and latches with native Matter-over-Thread support, Wi-Fi connectivity, and an integrated doorbell. Available in both deadbolt and latch configurations, the Affirm line is designed for homeowners and rental operators seeking platform-agnostic smart locks that don’t necessarily require a dedicated hub. The Affirm series can be integrated with Amazon Alexa, Google Home, Apple Home, and Samsung SmartThings. Thread’s low-power design helps extend battery life versus Wi-Fi–only connections. While a Thread border router is needed for full functionality, Affirm smart locks can also connect via their built-in 2.4GHz Wi-Fi adapters using the Lockly Home app. For access, users can enter PIN codes on a backlit keypad (featuring Lockly’s unique PIN Genie technology) or unlock via NFC cards or fobs, smartphones, or wearables. Physical keys are also available as a fallback option. The system can store hundreds of codes, suitable for families or short-term rentals. Lockly An integrated doorbell in the lock’s exterior housing allows visitors to announce their presence without additional hardware. When paired with Lockly’s TapCom platform, the NFC tag on the lock can initiate audio or video calls, depending on configuration. Software features include scheduled auto-locking, offline-accessible codes, and a “welcome mode” for unlocked periods. Real-time updates, activity logs, and remote controls are available via the Lockly app. LocklyOS supports property-wide lock management for multi-unit installations. Lockly’s Affrim smart lock comes in deadbolt and latch configurations. Lockly The Affirm series is rated IP65 for weather resistance, runs on eight AA batteries (rated for up to 8 months), and includes a USB-C port for backup power. Designed for DIY installation, it fits most standard doors and backsets. Both versions of Affirm will launch in matte black, with pricing starting at $179.99 and availability expected in late Q2 2026. If you’re attending CES in person, Lockly is exhibiting its wares at Venetian Suite 29-318, where it is also showing its new Smart Safe XL, OwlGuard IPC100 home security camera, and TapCon access-control system for short-term rentals. This story is part of TechHive’s in-depth coverage of the best smart locks. Read...

The DIY market will take the largest hit from the shortages and price increases in the DRAM and SSD market, a leading analyst firm predicts. But there will be no escape for those buying a prebuilt PC in 2026, as analyst firm IDC says they’ve confirmed that major PC makers will raise prices by up to 20 percent. In a recent press note, analyst firm IDC said that “Lenovo, Dell, HP, Acer and Asus have warned clients of tougher conditions ahead, confirming 15-20 percent hikes and contract resets as an industry-wide response.” The apocalyptic rise in DRAM and SSD prices began to be felt a few months ago, with SanDisk, Samsung, and Micron warning of increased prices in both components. Black Friday became the focal point for the DRAM and SSD price increases, with companies like Framework, Minisforum, and Raspberry Pi warning that prices will increase. A memo obtained by TrendForce says that Asus will be the most recent PC maker to follow suit, with price increases from between 10 percent to 30 percent. Even though the largest vendors have begun quietly telling customers of the price increases, IDC said that the largest vendors will have the most clout to negotiate prices. That, in turn, will leave smaller boutique vendors and DIY companies vulnerable, presumably making them scrabble for whatever remaining components that they can get their hands upon. Bigger vendors will win “White box as well as lower tier (often local) vendors, on the other hand, will bear the greatest burden of the shortage, and that would include DIY systems, oftentimes built by gamers,” IDC wrote. “That in turn represents an opportunity for large OEMs to gain share from smaller assemblers in the gaming space by positioning pre-built systems as offering higher value.” The culprit remains large AI hyperscalers, who are snapping up whatever components that they can to build out data centers for training and inferencing new AI models. The difference between the RAM ending in servers and PCs is relatively slight, meaning that memory makers can and are shifting production to higher-margin server products. SSDs are being pulled in, too. All this means that is that commodity memory makers are reaping the benefits of astronomical demand, and the economics of the market mean that prices rise as a result. Ironically, that means further pressure on AI PCs. Intel-based AI PCs typically allocate half of the available DRAM to VRAM. AMD’s mobile Ryzen platforms used a fixed amount of VRAM, though a technology called Variable Graphics Memory allows you to fine-tune this manually. VRAM is basically where AI applications currently run on PCs, so maxing out VRAM is seen as the key to improved overall AI performance. Copilot+ PCs, for example, require at least 16GB of DRAM. The issue, though, is an obvious one: with more DRAM being shipped off to datacenters, the less that AI PC makers have to populate their own systems. “As more small language models and large language models move on device, memory becomes even more important, with many higher-end systems shifting toward 32GB or higher,” IDC wrote. “Just as the industry is seeing a need to add more RAM, it has become prohibitively expensive to do so, even if they can get supply. This will result in higher prices, lower margins, or a potential downmix in the amount of RAM in new systems at the worst possible time for this to occur.” IDC hasn’t officially changed its forecasts for the PC market, though it said that it’s now offering several “scenarios” for how the market could play out, depending upon the supply constraints. At worst, the PC market could drop 8.9 percent. IDC also predicted that the prices of phones, especially Android phones, could be affected. Memory can make up as much as 20 percent of the build cost of a cheaper smartphone, IDC noted. Again, larger vendors like Apple will have an advantage, though flagships might not see the continued increases in RAM and storage that customers have come to expect. “For consumers and enterprises alike, this signals the end of an era of cheap, abundant memory and storage, at least in the medium term,” IDC wrote. Read...