AMD's Madhu Rangarajan, corporate VP of Compute and Enterprise AI, says the company never set out to build its Zen 6 EPYC 9006 "Venice" chips around agentic AI, a not-so-subtle dig at NVIDIA and Arm, which both pitch their latest server CPUs as made for that exact job. Rangarajan still called EPYC 9006 the best CPU for agentic workloads during a media session covered by Tom's Hardware. Chief Zen architect Mike Clark said much the same in a separate talk, admitting AMD didn't see agentic computing coming, though he added that the company always expected the CPU to keep a role in AI software. Here's what Rangarajan said:

We think that the EPYC 9006 CPU is the best CPU for the agentic era, and spoiler alert, we didn't magically know four years ago that agentic was going to be a thing and purpose-built something for agentic. The reality is an agentic workflow is a diverse set of tasks… and in order to execute those tasks well, you need compute that is able to handle a diverse set of workloads with a full portfolio.

Qualcomm has officially validated its High Bandwidth Compute (HBC) concept, which the company showcased this summer. Instead of using traditional HBM and compute dies, Qualcomm has developed a technology that stacks multiple LPDDR layers and embeds a base die that serves as a compute engine for AI inference. In Qualcomm's San Diego lab, the first-generation HBC has achieved silicon validation with full read-write and near-memory computation capabilities. HBC utilizes LPDDR memory, stacking it across multiple layers in a 3D vertical space and interconnecting them with through-silicon vias (TSVs). This approach provides better energy efficiency than traditional HBM, which stacks regular DDR in vertical layers, as the low-power DDR (LPDDR) chips consume less power while delivering similar bandwidth and capacity.

Qualcomm has successfully stacked multiple LPDDR layers on top of each other, with the base of the HBC being a compute die that performs near-memory computation and offloads processing to near-memory dies. This technique is similar to what HBM4 memory currently does, with the base die being a logic die to better integrate compute solutions for packet tracing and data preparation for entering and exiting HBM. However, Qualcomm aims to achieve higher efficiency and more optimized data movement, with a 6x increase in bandwidth per watt compared to the current HBM specification. The goal is to have compute directly under these memory stacks so that data travels only a few microns instead of a few centimeters.

The LattePanda Team today launched LattePanda Sigma 2, an x86 single-board server designed to bring high-performance local AI computing directly into intelligent products and systems. Built around the Intel Core Ultra X7 358H processor with Intel Arc B390 graphics, LattePanda Sigma 2 delivers up to 180 TOPS of total AI performance at INT8 in a compact form factor.

As "The Embeddable AI Single Board Server," LattePanda Sigma 2 combines high-performance computing, up to 64 GB of memory, high-speed networking, flexible expansion, and real-time hardware control in a single board. Rather than sitting alongside a product, it is designed to be integrated directly into the product itself, supporting applications from autonomous robots and vision systems to professional instruments and private AI appliances.

GPD is no stranger to powerful gaming handhelds, having previously launched the very expensive GPD Win Max 3 in August 2026, and the company has recently confirmed via a Weibo post that it is working on an Intel Arc G3 Extreme-powered version of its Win Mini gaming handheld.

Prior versions of the GPD Win Mini were available with up to the AMD Ryzen AI 9 HX 370 with its Radeon 890M iGPU, which put the system more or less on the same level as other popular handhelds, like the Lenovo Legion Go 2, while a Win Mini Ultra with the Intel Arc G3 Extreme CPU could perform significantly faster—about as well as an entry-level dGPU, thanks to the onboard Intel Arc B390 iGPU with 12 full Xe3 cores.

SK hynix reported its latest Q3 2026 quarterly results with a record $78.129 billion in net revenue. However, the company also shared a market outlook, highlighting its long-term supply agreements and plans for expansion with new fab capacity and facilities under construction. For the consumer market, the outlook is grim as SK hynix expects memory demand to continue outpacing supply through 2030. The main supply constraint is the complexity of leading-edge semiconductor nodes and the lengthy lead times for fab construction, which take years to translate into regular mass production volumes.

Despite sustained demand, pricing is also on an upward trajectory, with the average selling price of memory modules in Q3 increasing by over 20% compared to Q2. This is concerning, as it suggests that the prices of consumer electronics devices might jump by more than 20% once memory inventory is depleted. Although SK hynix expects a decrease in PC and smartphone shipments, this will be offset by a strong rollout of AI data centers. About 14 major customers, including most hyperscalers, have signed 5-year long-term agreements, which will account for over 50% of total capacity sold for 2027 and 2028.

Intel recently announced its D-series "Crescent Island" inference graphics cards, based on the Xe3P GPU microarchitecture, with the D270 being the highest-end SKU. The company's website has added a GPU bare die shot, revealing that the card is not the massive chip one might expect, but rather something seemingly smaller. The D270 GPU SKU features 32 Xe3P cores, 256 XMX engines, and 32 MB of unified L2 cache. The XMX engines have been redesigned from the four-deep systolic design of the regular Xe3 to 16-deep processing on the Xe3P. This card is intended for AI inferencing rather than training, so silicon requirements are secondary, and the card is rated for up to 480 GB of LPDDR5X.

Additional details include support for the FP4 data format alongside FP8, as well as large register files of 1 MB per core and an L1 cache of 512 KB per core. In total, there are 256 Vector engines and 256 of the new XMX engines, each divided into groups of eight per Xe3P core. Furthermore, the Xe3P lacks graphics processing elements for outputting imagery, as the silicon is designed purely for AI inference workloads that require long context and agentic tasks. A media engine with four encoders and four decoders is also included, which could assist with multimodal workloads where video generation and understanding are essential.

YMTC, known as a NAND Flash manufacturer, is reportedly preparing to start mass production of LPDDR5X memory. While it has been known for some time that YMTC plans to enter DRAM production alongside its main NAND Flash business, it was not widely known that the company had already begun low-scale manufacturing. According to Counterpoint Research, YMTC has been producing about 5,000 wafers per month of 16 Gb LPDDR5X DRAM running at 7,500 MT/s at its Fab 2 since October 2025. The company is now preparing to begin volume production of LPDDR5X memory with speeds of 9,600 MT/s at its Fab 3, using Vertical Channel Transistor (VCT) 4F² cell structure technology.

YMTC has a significant advantage with 3D silicon, which represents the next phase of memory evolution. With its Xtacking 4.0 technology, YMTC can separate memory cell wafers from peripheral logic circuits and combine them using hybrid bonding wafer-to-wafer stacking. Hybrid bonding creates a permanent bond between two silicon chips using embedded metal bumps that are fused together to form a single chip. Multiple dies or entire silicon wafers can be connected by stacking the back ends of two chips together.

Xiaomi has officially spilled the beans on the 2027 update to its Redmi Book laptop series, detailing both a launch date and what to expect from the laptops in terms of performance and battery life. The information comes courtesy of the Xiaomi Redmi account on Weibo, and the announcement reveals an October 27 release date for the Redmi Book 14 and likely the 16, alongside it. It's so far unclear if and when the new Redmi Book laptops will launch internationally or what the pricing will be, but the outgoing Redmi Book 14 starts at 6,499 CNY with 16 GB of RAM and 512 GB of storage.

With the post, Xiaomi also confirms that the Redmi Book 14 and 16 will be using Intel's Wildcat Lake Core Series 3 processors, with previous leaks having tipped the Core 5 320 for the Redmi Book 16, and the fine print on the marketing images now confirming the Core 5 320 for the Redmi Book 14. The Redmi Book 14 2027 has also been confirmed to ship with a 72 Wh battery, which Xiaomi claims will deliver over 20 hours of office use, with a claimed 1.26 days of local video playback. It will also use LPDDR5X-7467 RAM, although capacities are unclear at the time of writing, aside from the confirmation that there will be a 16 GB version available. The fine print also confirms that the display will be 2.8K, however the panel tech isn't mentioned.

Intel's senior vice president, Cho Min-sung, announced at the Semiconductors Powering Humanoids summit in South Korea that the next-generation "Razor Lake-AX" platform will be a computing powerhouse, featuring 700 TOPS of AI compute on die. This means that at INT8/FP8 precision, "Razor Lake-AX" will outperform the current "Panther Lake" mobile chips by 3.88 times, which have a total compute power of 180 TOPS. This represents a significant boost in compute power for the SoC, primarily derived from a large integrated GPU alongside a substantial CPU complex. Earlier reports indicated that the "Razor Lake-AX" iGPU will be available in two versions: one with a 16-core iGPU and a flagship SKU with a 32-core iGPU.

Both of these integrated graphics solutions are expected to be based on Intel's Xe3 graphics IP, currently found in "Panther Lake." This would make "Razor Lake-AX" nearly three times as powerful as the current Arc B390, which powers Intel's PTL lineup. This suggests that "Razor Lake-AX" will be a large package, as the GPU die is reportedly around 162.84 mm². We also learned that Intel plans to reintroduce on-package memory with the upcoming "AX" chip. Although the company initially stated that the Core Ultra 200V "Lunar Lake" was the last design with on-package memory, Intel is reportedly bringing back on-package LPDDR5X/LPDDR6 memory for "Razor Lake-AX." Do note that the image below is simply an illustration.

A leaker going by Haze on X claims that some partner-built versions of Intel's new GPU D-Series accelerator will come with as much as 512 GB of memory. The October 13 post, spotted by VideoCardz, names no board partners and gives no other specs, so take it with a pinch of salt for now. Intel hasn't confirmed anything like this either. Officially, Intel's own D-Series card, formerly known as "Crescent Island", ships with 160 GB of LPDDR5X, and the company allows partners to go as high as 480 GB. The 512 GB figure is a bit odd when you look at the hardware. The leaked "Crescent Island" PCB from May showed 20 LPDDR5X sites, 12 on the front and eight on the back, each with an 8 GB package for 160 GB in total. Moving to 24 GB packages gets you neatly to 480 GB, while 512 GB doesn't divide evenly across 20 sites.

The D-Series itself is an inference-focused accelerator built on Xe3P with 32 cores, 1.5 TB/s of memory bandwidth and up to 1.3 PFLOPS of FP8 compute. It's a 350 W, air-cooled PCIe 5.0 x16 card with no display or 3D graphics hardware, so gamers shouldn't get their hopes up, and earlier leaks already suggested Xe3P would skip Arc gaming cards altogether. Customer sampling is underway, but pricing and availability are still unknown. A 512 GB partner card, if it exists, would likely cost a small fortune. It would probably take a different board layout or a mix of package densities, and there's no word on whether any partner is actually doing that.

ASUS today announced the ExpertBook A7 (14" AMD) laptop, designed to set the new measure for the modern workday through refined aesthetics, effortless agility, and everyday assurance. Its seamless all-metal construction, clean lines, beautiful finish, and precision details create a sophisticated look that feels polished from every angle. With a compact design measuring just 1 cm thin and starting at 1.2 kg, ExpertBook A7 moves effortlessly from one workspace to the next, while smooth performance, advanced security, and dependable durability help professionals stay productive wherever work takes them.

Designed for professionals whose work moves with them, ExpertBook A7 brings together style and mobility without compromising the essentials of everyday productivity. From its precision-milled aluminium alloy construction and immersive 3K OLED display to its intuitive haptic touchpad, responsive performance, and layered protection, every element is carefully considered creating an experience that feels refined, effortless, and dependable throughout the day.

AMD showed off its next-generation EPYC "Verano" server CPUs during CEO Dr. Lisa Su opening keynote at the 2026 OCP Global Summit. The new CPU is the first EPYC processor to support LPDDR5X memory via SOCAMM2 modules. Now AMD is mirroring rival NVIDIA's Vera CPU that already uses SOCAMM2 modules. The "Verano" processor sits within the EPYC 9006 "LP" family and is built on the same TSMC 2 nm process as the already-launched EPYC "Venice". While AMD "Venice" is the first EPYC processor based on Zen 6 architecture, "Verano" is optimized for AI tasks in agentic AI racks. For this, AMD plans to pair them with its own Instinct MI500 GPUs via xGMI connectivity at up to 112 Gbps. It will run on the new SP8 platform rather than Venice's SP7. When we first heard about "Verano", AMD indicated that the upcoming CPUs will feature up to 72 cores on a single-socket platform. However, that has now been updated to up to 192 cores.

From the initial information available, this could mean either a single large chip or a dual 96-core configuration. More details came from System360Cheese on X, who reported that the chip shown featured eight Zen 6 CCDs alongside two large IO dies and four LPDDR5X memory controllers. The CCDs appear smaller than those in Venice's Zen 6C design, suggesting these are the standard 12-core Zen 6 dies rather than the denser 32-core Zen 6C variants. The announcement follows AMD's earlier reveal of the Helion agentic AI rack, an Open Compute Project spec-designed system that Verano is expected to be integrated. Availability is targeted for 2027.

AMD's upcoming EPYC "Verano" processors will apparently use a new socket called SB1. The name comes from server cooler maker Dynatron, which now states on its website that its cooling solutions support "AMD EPYC Verano on Socket SB1." Dynatron also has a product page up for a 4U active cooler listed as "SB1-4U-ACTIVE (TBD)," so the final product name is still pending. The "B" in SB1 could point to a BGA package soldered to the board, although that is a guess based on the name and nothing official. The listing doesn't help much either. Maximum power is marked as TBD, and the 128.05 x 106.6 x 130.7 mm dimensions are identical to Dynatron's J26 cooler for SP7, so the specs could simply be placeholders for now.

Verano is the codename for the EPYC 9006 LP, a Zen 6 part that AMD announced alongside the rest of its 6th Gen EPYC lineup. It tops out at 72 cores with boost clocks of up to 5 GHz, and it runs on 24 channels of LPDDR5X memory through replaceable SOCAMM2 modules. AMD is aiming it at AI racks as a host processor for GPU clusters, with a launch planned for the second half of 2027. The regular "Venice" chips use different sockets, namely SP7 for up to 256 cores and SP8 for up to 128. For "Verano," AMD itself hasn't revealed the socket name yet.

ASUS has already shown off its upcoming ProArt 16 creator laptop powered by the NVIDIA RTX Spark N1X Arm SoC at IFA 2026, but there was no exact launch date attached to the new laptops when we published our hands-on coverage on the show floor. Now, thanks to a new Weibo post by the official ASUS account, it has been revealed that ASUS will launch the ProArt P16 H7607BA (or a similar variant) on October 8, 2026—at least in China. It's not clear if this will be a simultaneous international launch, but, even if it isn't, an international launch shouldn't be too far behind.

ASUS on Weibo (machine translation)Smooth local AI agent response, creative workflows no longer reliant on the cloud
World's first AI agent PC | ASUS ProArt 16
Equipped with NVIDIA RTX Spark N1X super chip
Reconstructing the boundaries of professional creation, local computing power accelerates the implementation of the entire workflow
Coming October 8th

Intel has added its highly anticipated "Crescent Island" Xe3P GPU IP as a supported platform to its Media Driver in the 2026Q3 release. This confirms that the company is working on bringing another GPU generation to the masses soon. With the Intel Media Driver 2026Q3, we learn that the GPU will support video acceleration as well. This means that video editing, video decoding and encoding, and processing will all be available on "Crescent Island," which we initially believed to be a GPU dedicated solely to AI inferencing. However, Intel will also power media processing workloads for workstations, and we expect to see similar features in video standards, like AV1 decode, as seen in "Panther Lake" and the upcoming "Nova Lake" with their respective integrated graphics.

If readers recall, Intel unveiled more details at Hot Chips 2026. Based on the Xe3P GPU IP, the graphics card is designed as a 350 W, air-cooled, PCIe format compatible with any configuration. The GPU itself features 32 Xe3P cores, 256 XMX engines, 32 MB of unified L2 cache, and 160 GB of LPDDR5X memory. This memory configuration is exclusive to Intel-branded "Crescent Island" graphics cards, while the company allows its ODMs to equip the card with up to 480 GB of LPDDR5X memory for maximum VRAM capacity.

ASUS today announced ASUS Googlebook 14 (CX9406CAA), a 14-inch ultraportable laptop that's the first ever ASUS laptop designed for Gemini Intelligence. Built for users who want a more connected experience with their Android phone, ASUS Googlebook 14 combines effortless portability, strong performance, and a premium-crafted design that keeps productivity moving seamlessly throughout the day.

ASUS Googlebook 14 comes with 12 months of Google AI Pro. Users can get expanded access to powerful features in Gemini, Flow, and Gemini Notebook as well as Gemini directly in the Google apps they know and use. Plus, they get 5 TB of cloud storage to keep photos and files backed up.

Dell Technologies announced the XPS Googlebook, a 13-inch premium laptop, powered by the Snapdragon X Elite processor, and the first XPS device designed for Gemini Intelligence. Built for Android users and supercharged to power their most advanced workflows, the XPS Googlebook brings the premium materials of the XPS portfolio to the Googlebook platform for the first time.

Why it matters
The XPS Googlebook is the first Dell laptop built for Gemini Intelligence, marking an expansion of the XPS portfolio into a space where premium options have been limited. It brings together ecosystem integration with the design, materials and performance expected from XPS.

Acer today debuted the Acer Googlebook 14, designed for Gemini Intelligence and the first from the company in a brand-new category of flagship laptops bringing together premium materials, powerful performance, intelligence features, and seamless syncing with Android phones.

Designed for the new Googlebook OS, the device works seamlessly with Android-based workflows and supports touch, keyboard, stylus, voice, and traditional laptop interaction methods within a unified user experience and help from Gemini at every step. The Acer Googlebook 14 is the industry's first convertible Googlebook and provides users with groundbreaking new features designed to boost productivity and creativity by enabling context-aware actions for personalized, proactive help and a more intuitive, streamlined workflow.

Honor, the smartphone and computer company formerly owned by Huawei, has officially revealed the Tiangong AXB35, a new AI workstation platform that designed for enterprise use and made to run on a choice of AMD, Intel, or NVIDIA hardware in a tiny SFF desktop chassis measuring just 70 × 192 × 203 mm, or around 2 L volume. Honor says that the Tiangong SFF PC is designed for everything from high-level office work and local knowledge bases, AI inference workloads, and agentic applications, but it would also be useful as a raw computational workstation for everything from 3D modelling and creative tasks to developing, and even gaming.

The Tiangong AXB35 comes in three variants: Standard, Pro, and Ultra, The Standard AXB35 comes equipped with Intel's Core Ultra X7 358H CPU, which features the new Intel Arc B390 iGPU, a total of 16 cores (4× P-Cores, 12× E-Cores, 4× LP E-Cores), and support for up to 96 GB of memory. The Pro version comes with AMD's 16-core Ryzen AI Max+ 395, packing the Radeon 8060S iGPU and up to 128 GB of LPDDR5X memory on a wide 256 GB/s bus. Meanwhile, the Ultra version comes with NVIDIA's new RTX Spark platform, based on the Arm architecture.

Google will open pre-orders for its first Googlebook laptops on September 21 at 9 AM ET. The company confirmed the date in an email to people who signed up for updates, and the same date now appears on Google's official Googlebook website. That's the first firm date tied to the product since Google teased a fall launch back in May. Acer, ASUS, Dell, HP, and Lenovo are all lined up as first-wave Googlebook manufacturers, and several of their machines have already leaked ahead of launch.

The Lenovo Googlebook 15 is probably the most exposed so far, reportedly pairing an Intel Core Ultra 5 325 "Panther Lake" chip with up to 32 GB of LPDDR5X memory, a 15.3-inch 120 Hz OLED touchscreen, and Google's colorful Glowbar light strip. That same Core Ultra 5 325 keeps showing up across leaked Acer and ASUS models too, which is a change from how Chromebooks have typically launched with entry-level Intel silicon. Google still hasn't said which models will actually be available on September 21, and it hasn't shared a shipping date either, only that units will arrive this fall.

Valve has opened reservations for the Steam Frame, its standalone VR headset, with pricing starting at $1,059 for the 256 GB model and going up to $1,299 for the 1 TB version. We had already reported back in July that Qualcomm's own chip price hikes were pushing Steam Frame costs upward well before today's reveal. In an interview with Tom's Hardware, Valve engineer Pierre-Loup Griffais walked through why the final price landed where it did. Here's what he said:

"I think as far as memory prices and those kinds of conditions, we're really in it like everyone else. It's an unfortunate situation all around. We wish that gaming hardware was cheaper, but we're definitely trying to be on the cheapest side that we can get despite all those conditions. Even just things like motherboards, copper, anything, even shipping anything worldwide right now is way more expensive than it was last year and two years ago. So every part of the sticker price is affected there."

Intel's next-generation Nova Lake mobile chips with 12 Xe-core graphics may support DDR5-8000 memory through CSODIMM modules, according to leaker Golden Pig Upgrade Pack on Weibo. If that pans out, it would be the first mobile Nova Lake variant to officially support DDR5, a notable departure from the soldered LPDDR memory that dominates thin-and-light laptops right now. Current Panther Lake systems with 12 Xe cores, like the Core Ultra X7 358H, don't officially list DDR5 support, though ASRock Industrial already ships mini PCs with two DDR5 SO-DIMM slots running at DDR5-7200 on that platform, and Intel itself includes this configuration among its Core Ultra Series 3 developer systems.

Nova Lake-S, the desktop version, has already been tied to DDR5-8000 through leaked B960 chipset details and platform documents, so Intel appears to be pushing that same memory ceiling down into at least part of the mobile lineup too. Unlike soldered LPDDR5X, CSODIMM support would let OEMs offer upgradable RAM again in these systems, an option that's grown rare in slimmer laptop designs built around battery life and board space. Despite the leaker's solid track record, this is still a single leak rather than anything Intel has confirmed, so I'd hold off treating the exact speeds as final. If it checks out, it suggests Intel is willing to give at least some mobile Nova Lake configurations more memory flexibility, even as most of the market keeps drifting toward soldered memory for efficiency.

CXMT has achieved the remarkable feat of officially beginning high-volume manufacturing of LPDDR6 memory. This positions CXMT's LPDDR6 memory modules as some of the first in the industry, with South Korean memory makers on a similar timeline for volume production of this next-generation technology. Reportedly, Xiaomi is the first customer for CXMT's domestic LPDDR6 memory, set to feature in the upcoming Xiaomi 18 Fold smartphone. Since the Xiaomi 18 Fold is scheduled to launch in September, CXMT's LPDDR6 memory will need to be delivered before the end of the month, indicating that customer shipments are already underway. This news follows recent reports that CXMT is conducting risk production of LPDDR6, suggesting a very high yield in production.

CXMT's LPDDR6 memory will operate at speeds of up to 12.8 Gbps, with the base speed of LPDDR6 being 10.7 Gbps. LPDDR6 builds on where LPDDR5X left off, which could reach speeds of up to 10.7 Gbps when highly overclocked. However, the new standard aims for significantly higher speeds, with the first generation of LPDDR6 from CXMT's fabs reaching 12.8 Gbps. Additionally, CXMT is producing these chips with a 16 Gb capacity, equivalent to 2 GB, and the memory IC is housed in a package-on-package (POP) casing, featuring about 1,295 pins in a ball grid array.Update, 14:55 UTC, September 7th : CXMT's official website confirmed mass production, with Alibaba Cloud, ByteDance, Tencent, Lenovo, Transsion, Honor, Vivo, and Oppo as customers.

TrendForce's latest findings on the memory industry reveal that sharp rises in traditional DRAM contract prices have led to a 59.5% QoQ increase in total DRAM industry revenue, reaching nearly US$154.73 billion in 2Q26.

Growing demand for AI servers, fueled by expanding LLM training and AI inference, has driven higher shipments of HBM3e, LPDDR5X, and high-capacity RDIMMS. The surge in agentic AI applications is further boosting RDIMM demand across various capacities.

At the IFA 2026 booth tour of Acemagic, we saw that the company was offering its F2A mini PC in both Intel and AMD configurations, providing buyers with two distinct hardware platforms in essentially the same compact form factor. The Intel version is powered by the Core Ultra X7 358H, based on Intel's "Panther Lake," paired with Arc B390 integrated graphics. This 16-core, 16-thread processor can reach 4.8 GHz, and the system is rated for up to 180 TOPS of combined AI performance. The AMD variant, on the other hand, uses the Ryzen AI 9 HX 470 "Gorgon Point" APU, featuring 12 cores and 24 threads, a 5.2 GHz boost clock, and Radeon 890M graphics based on RDNA 3.5. Its XDNA 2 NPU provides up to 55 TOPS, while the complete platform is rated at up to 86 TOPS. Both versions come with 32 GB of onboard LPDDR5X memory. However, the Intel model pairs it with a PCIe 5.0 x4 M.2 slot alongside a PCIe 4.0 slot, while the AMD version uses two PCIe 4.0 x4 interfaces.