Patriot Memory, a global leader in high-performance memory and storage solutions, will unveil its latest advancements at CES 2026, taking place January 6-9, 2026, in Las Vegas, Nevada. Building on decades of engineering expertise, Patriot's CES 2026 showcase highlights the company's continued focus on next-generation DDR5 memory and PCIe Gen 5 storage technologies designed for enthusiasts, gamers, and professional users. At CES 2026, Patriot Memory will present its newest innovations across both memory and storage under the Patriot andViper Gaming brands, including:

Next-Generation DDR5 Memory: Viper Xtreme 5 Aurum Edition DDR5
The Viper Xtreme 5 Aurum Edition represents the pinnacle of Patriot's DDR5 design philosophy, combining elite performance with refined aesthetics. Engineered for stability at high frequencies, the Aurum Edition delivers DDR5 speeds of up to 8000 MT/s with capacities up to64GB, paired with a premium, precision-crafted heat spreader design. Built for enthusiasts who demand both speed and visual distinction, the Aurum Edition embodies Patriot's commitment to pushing DDR5 performance while maintaining elegance and reliability.

TrendForce's latest investigations indicate that DRAM suppliers in 1Q26 will continue to reallocate advanced process nodes and new capacity toward server and HBM products to support rising AI server demand. This shift has significantly limited supply in other markets, causing conventional DRAM contract prices to increase by about 55-60% QoQ. Meanwhile, disciplined capacity management by NAND Flash suppliers, along with robust server demand that is displacing other applications, is likely to increase contract prices across all NAND Flash product categories by 33-38%.

Despite weaker notebook shipments and slower growth in memory demand due to potential specification downgrades, PC DRAM prices are still set to rise sharply in the first quarter. DRAM suppliers have simultaneously tightened supply to PC OEMs and module makers, forcing some OEMs to procure memory at higher prices through modules makers. This is expected to raise DRAM suppliers' module pricing and significantly drive up PC DRAM prices.

We've already seen the myriad far-reaching effects of the DRAM crisis play out in a number of ways, including bring-your-own-RAM offerings from custom PC builders and rumors of delayed next-gen consoles. However, a new report out of DigiTimes claims that DRAM producers, specifically mentioning Samsung and SK Hynix, are now selectively prioritizing certain hardware makers. Allegedly, large brands like Apple, ASUS, Dell, and Lenovo are receiving priority access to DRAM, which may result in smaller vendors—the likes of Framework—without DRAM supply.

The report also mentions that DRAM suppliers are starting to change their approach to the supply agreements, more frequently re-assessing the situation in order to take advantage of the limited capacity and likely adjust pricing and supply accordingly. Previously, we covered a report that claimed PC hardware vendors may have to adjust pricing, specs, and potentially launch dates of upcoming laptop and smartphone offerings, and these agreements are likely being pursued by mainstream brands, like Apple, ASUS, Dell, and Lenovo, in order to avoid or mitigate that eventuality.

If you're in the market for a high-capacity DDR5 memory kit featuring ECC RDIMMs, this might be your lucky opportunity. NEMIX, a U.S.-based memory and storage supplier that collaborates with major tech firms and the government, has listed a 4 TB DDR5 memory kit. This kit consists of 16x 256 GB ECC RDIMMs operating at 6400 MT with a CAS latency of 52. Notably, the listing comes with a hefty price tag of $70,800, making it accessible only to high-budget hyperscalers or workstation enthusiasts. The 4 TB kit's substantial cost is striking, especially during a time of significant DRAM shortages. This translates to approximately $17.3 per GB, significantly higher than the average $10 per GB for consumer DRAM.

The nearly doubled price of these modules is justified by the advanced technology incorporated into these RDIMMs. They feature off-die error correction chips for ECC, which are not standard in regular DDR5 DIMMs. While DDR5 includes on-die ECC, it does not provide complete error correction on its own. Therefore, additional silicon is necessary for hyperscalers and workstations to achieve full error correction, which is reflected in the price premium. The 4Rx4 rank setup helps maintain smooth performance, and it operates at just 1.1 V, making it quite efficient. With registered buffering and built-in ECC, these modules show why their price point is much higher than traditional DIMMs.

Framework updated its original DRAM price hike blog post on December 24th, 2025, explaining that the company must respond to the price volatility its suppliers are experiencing. Initially, Framework announced a 50% price increase for its DDR5 memory upgrade configurations. To maintain transparency with customers, the company stated its intention to raise the price of DDR5 memory in Framework Laptop DIY Edition orders by 50%. This decision directly responds to the significantly higher costs from DRAM suppliers and distributors. As a result, the company cannot absorb all these cost increases and needs to pass some on to consumers. Therefore, another increase today is a natural response, as the company originally suggested.

FrameworkWith costs from our suppliers continuing to increase, we've had to make a further price adjustment on DDR5 memory modules. During this period of extreme memory shortages and price volatility, our priority is to make sure you can still buy a computer when you need one. With that in mind, we're setting our memory configuration prices as close as possible to the actual purchase prices we have with our suppliers and distributors. Since we're constantly sourcing additional memory and each purchase comes in at different (and often higher) pricing, we're using the Weighted Average Cost (WAC) of inventory, which currently comes to $10/GB for 8 GB, 16 GB, and 32 GB modules, and slightly higher for 48 GB modules.

As regular desktop memory prices have skyrocketed over the past year, more gamers are turning to repurposing laptop SODIMM memory using SODIMM-to-DIMM adapters. This allows them to use the more affordable memory in their builds. According to 3DCenter.org, these adapters are relatively inexpensive. When paired with SODIMM memory, which has experienced a smaller price increase—an average of 136% compared to 245% for desktop DIMMs—gamers find that buying laptop memory makes more sense. They can achieve good enough memory performance at a much lower cost and finally perform a system upgrade or build a new PC.

These adapters serve more as a workaround than an ideal solution, mainly designed for repurposing spare laptop memory rather than achieving peak performance. Users must be prepared for significantly reduced memory clock speeds and looser timings, as mobile SODIMMs inherently lack the performance capabilities of desktop modules. Additionally, the adapter's signal path often requires further down clocking to maintain system stability.

AMD's Extended Profiles for Overclocking (EXPO) technology is receiving an update with version 1.2. According to the latest HWiNFO v8.35-5890 Beta, this tool has added support for AMD EXPO 1.2, even though AMD has not officially released this version yet. Currently, AMD employs EXPO 1.0 with its Ryzen processors to provide optimal overclocked memory profiles for each configuration. This approach enhances overall system performance and gives the Infinity Fabric, which interconnects the CCDs and IODs, perfect balance of latency and speed.

What will come with EXPO 1.2 remains a mystery. Recently, as part of the "Granite Ridge" Ryzen 9000 X3D series, AMD's upcoming Ryzen 7 9850X3D processor was seen running with 9,800 MT/s DDR5 memory, which is unusually high for its class in this late-cycle "Zen 5" SKU. AMD could bring some late-stage "Zen 5" memory profile optimizations that we haven't seen so far, just in time for the late cycle X3D SKUs.

Update 22:15 UTC: According to AMD insider and author of HYDRA, CTR, DRAM calculator for AMD Ryzen 1usmus on X, the EXPO 1.2 will finally include support for CUDIMMs. These are DDR5 modules with integrated clock driver circuits, which improve signal integrity at higher frequencies, allowing the DDR5 modules to reach frequencies beyond JEDEC-provided range. We still have no idea how AMD plans to implement this. The slide below is just for illustrative purposes and is AMD's earlier mention of EXPO. It does not reflect the latest 1.2 version.

DRAM prices are experiencing significant industry-wide volatility, due to severe global supply constraints and shortages, driven by unprecedented high demand from the AI industry. As a result, G.SKILL procurement and sourcing costs have substantially increased. G.SKILL pricing reflect industry-wide component cost increases from IC suppliers and is subject to change without notice based on market conditions. Purchasers should be mindful of the pricing before purchasing. Thank you.

SK Hynix held an internal company meeting, where the company reportedly presented some tough pills to swallow for many gamers. The company now forecasts the tight memory supply to last through 2028 for the commodity DRAM, which includes DDR5/DDR4, GDDR6/GDDR7, and LPDDR5x/LPDDR6. All the aforementioned DRAM variants are essential for PC and console components, making millions of gamers at risk of absorbing a massive price increase as a result. As memory supplier inventories deplete, production capacity will not increase to meet demand as it has in the past. This marks a departure from the usual response by memory manufacturers, who typically ramp up capacity in response to demand.

However, an interesting thing that SK Hynix has noted is that this situation will exclude the company's most advanced memory solutions—HBM and SOCAMM. These products are expected to get an additional capacity expansion, without tight supply impacting this part of the DRAM supply chain. These products are even in high demand as they are mostly consumed in higher volume than the regular commodity DRAM, due to their integration into AI products such as GPUs and servers.

JEDEC Solid State Technology Association, the global leader in standards development for the microelectronics industry, today announced it is nearing completion of a new standard for Standard Package High Bandwidth Memory (SPHBM4). SPHBM4 devices are similar to the HBM4 devices commonly used in artificial intelligence accelerators, using the same DRAM dies on a new interface base die which can be mounted on standard organic substrates. In contrast, HBM4 is typically mounted on silicon substrates.

As planned, SPHBM operates at the same aggregate data throughput as HBM4 using fewer pins by operating at a higher frequency. Where the HBM4 interface has 2048 data signals, when published, SPHBM4 will define 512 data signals with 4:1 serialization to achieve the same bandwidth. This change allows the relaxed bump pitch required for connection to organic substrates.

Fresh JD.com store listings have aptly reaffirmed ever-increasing memory pricing trends—over the past weekend, ITHome noticed extortionate tags applied to Asgard's newest high-capacity DDR5 kits. Given current circumstances, certain PC hardware enthusiasts will begrudgingly meet the challenge of taking on daunting costs of ownership. The Chinese brand's latest premium-tier 256 GB (4 x 64 GB) and 192 GB (4 x 48 GB) kits are not particularly cutting-edge in terms of basic specifications: 6000 MT/s memory transfer rates, on SK Hynix M-die ICs. Over a year ago, the company launched a better-binned Thor DDR5-9600 product line, albeit with a much lower capacity: 48 GB (2 x 24 GB).

The Valkyrie II RGB 256 GB limited edition kit—marked with CL32-45-45-90 timing credentials—reportedly launched at 14,599 Yuan (~$2067 USD), but an updated JD.com page tacks on an additional 2400 Yuan. This (roughly) 16.4 % increase leads to a ~$2407 USD total. The Valkyrie's "racing orange" color scheme is an acquired taste; thankfully the manufacturer's cheaper alternative offers a stealthier approach. At the time of writing, the Asgard Thor Armor 192 GB limited edition kit's launch price remains intact: at 8599 Yuan (~$1217 USD). Granting access to respectable CL28-36-36-72 timings, and understated "carbon fiber heat spreaders." Recent press coverage has compared Asgard high-capacity memory prices to current-gen NVIDIA desktop graphics products—namely the Valkyrie II 256 GB costing $400 more than a GeForce RTX 5090 Founders Edition card (at MSRP). Tom's Hardware points out that similarly specced G. Skill and Corsair kits sport greater price tags, with competing Asgard SKUs matching or exceeding performance (specs-wise).

Penguin Solutions, Inc. (Nasdaq: PENG), a leading provider of high-performance computing and AI infrastructure solutions, today announced the release of its SMART 64 GB DDR5-6400 ECC CSODIMM memory modules, the newest addition to its broad DDR5 SODIMM lineup.

Designed for high-performance computing (HPC) in industrial, edge, telecom, and networking systems, the new module supports both commercial-temperature (C-temp) and industrial-temperature (I-temp) options, engineered to provide reliable operation in harsh environments. The compact CSODIMM form factor delivers higher density and faster bandwidth with error correction capability (ECC) to meet the demanding requirements of next-generation platforms.

As part of the "Granite Ridge" Ryzen 9000 X3D series, AMD's upcoming Ryzen 7 9850X3D processor has been spotted running with 9,800 MT/s DDR5 memory, placing this late-cycle "Zen 5" SKU unusually high for its class. The AMD Ryzen 7 9850X3D features 8 cores and 16 threads, 96 MB of L3 cache, and a boost clock of up to 5.6 GHz, with a 120 W TDP. What sets this processor apart is not just the 400 MHz increase over the Ryzen 7 9800X3D model, which shares the same cache and core configuration. Officially, AMD's Ryzen 9000 series supports DDR5 memory speeds up to 5600 MT/s, but this model could push that number to 9,800 MT/s if the reports are accurate.

AMD might be using higher-binned IODs it has accumulated over the past few months to offer a competitive alternative to Intel's upcoming "Arrow Lake Refresh," set for early 2026, and "Nova Lake," scheduled for late 2026. Since the overall "Granite Ridge" family officially supports much lower DDR5 memory speeds, in line with JEDEC specifications, a new or better-tested IOD could enable these 9,800 MT/s speeds. In the BIOS image below, you can see a 9,800 MT/s figure. We can disregard the incorrect date, as it could have been intentionally altered by the leaker to prevent AMD from tracing the source. The CPU picture below serves as a mockup of what the actual design might look like.

Artificial intelligence works fast, but its energy consumption is growing rapidly. A German-Taiwanese research team is now developing a solution: new memory for leading chip technologies smaller than 3 nm. These innovative nanosheet devices enable computing operations directly in memory, thereby drastically reducing energy consumption. They are based on ferroelectric field-effect transistors (FeMFETs) made from hafnium oxide, which are particularly efficient. With a joint research program, Fraunhofer IPMS, Fraunhofer IMWS, and the Taiwanese research institute TSRI are laying the foundation for the next generation of energy-efficient AI chips - from smartphones and automobiles to medical devices.

Given the rapidly growing demand for artificial intelligence (AI) and neuromorphic computing, the energy consumption of data centers and edge systems is increasing dramatically. A key bottleneck is the transfer of data between main memory and the computing unit. A joint German-Taiwanese project aims to address precisely this issue: innovative memory technology will enable computing "directly in memory", with significantly lower latency and energy consumption.

XPG, a fast-growing provider of systems, components, and peripherals for Gamers, Esports Pros, and Tech Enthusiasts, today announces the launch of the new ARMAX DDR5 gaming memory series. Inspired by the aggressive silhouette of stealth fighter jet defending the skies, the series features a hardcore exterior and delivers high-performance speeds of up to 6,400 MT/s. The lineup includes the ARMAX RGB DDR5 with dazzling lighting effects and the understated, low-key ARMAX DDR5, aiming to be the premier DDR5 upgrade choice for next-gen gamers to conquer all gaming campaigns.

Targeting the DDR5 Performance Sweet Spot with SFF Compatibility
In response to the growing trend of SFF (Small Form Factor) mini-PC modifications, the ARMAX DDR5 features a heatsink height of just 39.5 mm. This design avoids interference with large CPU coolers, cables, or other equipment, allowing for easy installation in various small chassis. Performance-wise, the series locks into the "performance sweet spot" of next-generation platforms with speeds ranging from 6,000 to 6,400 MT/s. Built with strictly selected high-quality ICs and featuring built-in PMIC (Power Management Integrated Circuit) and On-Die ECC (Error Correcting Code), it supports Intel XMP 3.0 and AMD EXPO for easy overclocking. The ARMAX series provides gamers with a perfect experience that balances performance, style, and excellent compatibility. The XPG ARMAX DDR5 memory series is available now through global channels and comes with a limited lifetime warranty.

China's semiconductor industry is looking to bridge its technology gap with NVIDIA by stacking mature logic nodes with high-bandwidth memory and new compute architectures, according to remarks from Wei Shaojun, vice president of the China Semiconductor Industry Association. Speaking at an event in Shenzhen (via ESM China), Wei said that 14 nm logic chips, far behind the 4 nm-class silicon used in NVIDIA's current AI GPUs, could reach comparable performance if paired with 18 nm DRAM using 3D hybrid bonding and a near-memory computing architecture. The approach reduces data movement and improves energy efficiency by placing compute elements directly beside memory. The concept mirrors Huawei's recent strategy of "stacking and clustering" chips as a workaround to US export controls, which block China's access to advanced nodes such as 5 nm and below. Huawei has argued that performance can still scale by increasing chip integration density instead of relying on cutting-edge lithography.

Wei emphasized that his proposed design uses a fully domestic supply chain, 14 nm logic and 18 nm DRAM are both covered by current US export restrictions, limiting access to foreign production. He also noted that the global AI ecosystem remains heavily dependent on NVIDIA hardware and CUDA software, calling it a "triple dependence" across models, architectures, and development tools. Chinese GPU alternatives are emerging, including Zhonghao Xinying, a startup founded by Google engineer Yanggong Yifa. The company claims its custom ASIC TPU accelerator can reach "up to 1.5x" the compute performance of NVIDIA's older A100 GPU.

TrendForce's latest report states that rising memory prices have significantly increased BOM costs in consumer electronics, leading brands to hike retail prices and dampening market demand. After reducing its 2026 shipment forecasts for smartphones and laptops earlier in November, TrendForce has also downgraded its 2026 forecast for game console shipments—from an initial YoY decline of 3.5% to 4.4%.

TrendForce notes that the swift rise in memory prices is affecting a range of consumer devices, including game consoles. Historically, game console manufacturers earned the majority of their profits from software and subscription services. As such, they depended significantly on incremental retail price reductions and occasional promotions to expand their user base.

Memory markets are currently facing a significant supply shortage, with December contract prices for key DRAM categories increasing by 80 to 100%, according to Team Group general manager Gerry Chen. He described this substantial price hike as the beginning of a multiyear memory upcycle and warned that the most severe impact will be felt in the first half of 2026, once distributors deplete their inventories. This price shock is already affecting the system bills of materials for OEMs, with memory's share in PC and notebook BOMs rising from the mid-teens to the mid-20s and higher, depending on the configuration.

Costs for DDR5 and SSDs have, on average, increased two to three times. For overall DRAM, it is 171.8% year-over-year. Cloud service providers are securing allocations for 2026 and 2027, which reduces the available supply for consumer and mainstream devices. This situation risks significant shipment declines in cost-sensitive segments like mainstream PCs, notebooks, and Chromebooks. However, even the powerful hyperscalers cannot secure all the DRAM allocations, as Tier-1 U.S. and Chinese cloud order books are filled to an effective 70% capacity, eliminating the safety stock that most buyers thought they had secured.

SK Hynix is gearing up to demonstrate its next wave of high-speed DRAM at ISSCC 2026, headlined by a 24 Gb GDDR7 chip rated for 48 Gb/s. According to ZDNet Korea the device uses a symmetric dual-channel design and updated internal interfaces to push past the 32 to 37 Gb/s speeds the industry previously expected from GDDR7. Korean outlet Global Economic News reports the per-pin rate represents more than a 70 percent jump over today's 28 Gb/s parts, raising per-chip bandwidth from 112 GB/s to 192 GB/s. At 24 Gb density, each package provides 3 GB of VRAM. That allows configurations such as 24 GB with eight chips, 36 GB with twelve, or 48 GB with sixteen. A 256-bit GPU using these devices could deliver around 1.5 TB/s of bandwidth, while a 512-bit interface would reach about 3 TB/s.

The GDDR7 paper listed for ISSCC positions the chip for mid-range AI inference hardware. SK Hynix joins Samsung and Micron in pushing GDDR7 speeds, but on paper it now targets the highest data rate of the three. Actual product clocks will likely land lower; GPU vendors routinely tune memory speeds to meet yield and power targets. GDDR7 at 24 Gb is still new to the market. Only NVIDIA's RTX PRO 6000, the 72 GB variant of the RTX PRO 5000, and the RTX 5090 Laptop currently use 3 GB modules. ISSCC papers typically appear a year or more before commercial availability, so this generation may reach data center or AI accelerator boards first. SK hynix is also introducing LPDDR6 on the same stage. The new mobile DRAM reaches 14.4 Gb/s, up sharply from LPDDR5's 9.6 Gb/s. The company is positioning it for high-performance smartphones, AI PCs, and edge devices running on-device generative AI.

NVIDIA is rumored to be changing its approach by supplying only the raw silicon to its AIC partners, rather than bundling its usual GPU and memory kit. Typically, GPU manufacturers like NVIDIA, AMD, and Intel provide their GPU dies with GDDR memory to AIC partners as a kit. These partners then solder the components onto their custom PCBs, often modifying the layout and cooling systems. However, according to Golden Pig Upgrade, NVIDIA might halt this practice and supply only the bare silicon dies due to memory shortages affecting the company's ability to meet orders.

Recently, AMD informed its supply chain of an impending price increase across its entire GPU product line by about 10%. This decision is primarily driven by a significant memory shortage, which has led to soaring DRAM prices and reduced profit margins for GPU manufacturers. AMD's AIB partners have been advised of the planned 10% price hike, and it is expected that AIBs will follow suit in the coming weeks to maintain their margins. However, this now adds to the entire confusion of DRAM shortage, especially for GPUs.

Tachyum today announced details about how its TDIMM is bringing the future of AI and computing, enabling AI models with parameters many orders of magnitude greater than those of any existing solution at a fraction of the cost. TDIMM is key to reducing the estimated cost of OpenAI data center $3 trillion and 250,000 megawatts of power to $27 billion and 540 megawatts.

Prodigy Ultimate provides up to 21x higher AI rack performance than NVIDIA Rubin Ultra NVL576. Tachyum DDR5 DIMM (TDIMM) bandwidth increase is 5.5 times, from 51 GB/s to 281 GB/s. Prodigy Ultimate's 24 channels provide 6.7 TB/s, 11x increase over 12-channel CPUs. The TDIMM supports 256 GB in standard, 512 GB in tall, and 1 TB in extra tall height, and TSV increases capacity up to 8 times to 3 PB per Prodigy node.

Samsung Electronics today announced that TM Roh was officially named Head of Device eXperience (DX) Division and appointed as CEO, joining Young Hyun Jun, Vice Chairman and Head of Device Solutions (DS) Division, as co-CEOs of the Company. President Roh will continue to oversee the Company's mobile business as Head of Mobile eXperience (MX) Business, while Vice Chairman Jun will remain as Head of Memory Business.

To strengthen Samsung's technology leadership, the Company named Janghyun Yoon as President, Chief Technology Officer of DX Division and Head of Samsung Research, and Hongkun Park as Head of Samsung Advanced Institute of Technology (SAIT). President Yoon was previously the CEO of Samsung Venture Investment. Prior to Samsung Venture Investment, he was responsible for software platforms as well as IoT and Tizen development at the MX Business. Park is currently the Mark Hyman Jr. Professor of Chemistry and Professor of Physics at Harvard University, and a renowned researcher in the fields of nanoscience, quantum science and engineering.

Intel's W890 workstation platform, the foundation for its upcoming "Granite Rapids-WS" HEDT workstation Xeon processors, has been leaked. The platform introduces a new Socket E2 design with 4,710 LGA pins built to support CPUs with TDP ratings reaching 350 watts. Intel is taking a two-tier approach with distinct "Expert" and "Mainstream" configurations: Expert-class implementations unlock 112 PCIe lanes (96 PCIe 5.0 lanes plus 16 PCIe 4.0 lanes), while the Mainstream tier delivers 80 PCIe 5.0 lanes. This will support a XCC-based "Granite Rapids-WS" processor with 86 cores and 172 threads, achieving single-core boost speeds near 4.8 GHz with 336 MB of L3 cache.

Memory capabilities mark a significant evolution from previous workstation platforms. The W890 chipset supports registered DDR5 modules, with RDIMMs validated for speeds up to 5,200 MT/s. In a quad-channel configuration with two DIMMs per channel, systems can address up to 2 TB of memory per processor socket. The connectivity suite includes a 2.5 Gigabit Ethernet controller, numerous USB 3.2 and USB 2.0 ports, multiple SATA III interfaces, and dual SlimSAS connectors that each provide PCIe 4.0 x4 bandwidth. Server-grade management infrastructure includes an ASPEED AST2600 BMC and a Nuvoton system supervisor, while power delivery provisions mandate a standard 24-pin ATX connector alongside support for up to four supplementary 8-pin CPU power headers.

AMD has unveiled the Instinct MI430X, a data center accelerator designed for the intersection of high-performance computing and large-scale AI. Deployments are already underway at two of the world's leading research facilities. The GPU is being integrated into the Discovery supercomputer at Oak Ridge National Laboratory and the Alice Recoque exascale system in Europe. Both systems will utilize the chip alongside AMD's upcoming EPYC "Venice" processors. By launching directly into production systems rather than through a preliminary announcement, AMD is making the MI430X as the immediate successor to its previous Instinct lineup and a key element for countries and organizations aiming to develop self-reliant, open AI infrastructure.

The MI430X tackles the memory bottlenecks that challenge modern AI and simulation workloads with an impressive 432 GB of HBM4 memory, delivering approximately 19.6 TB/s of bandwidth. These specifications essentially transform the accelerator into a near-memory-compute device for trillion-parameter models. The chip is powered by AMD's next-generation CDNA architecture, which maintains hardware-level double-precision FP64 support for traditional scientific computing while introducing native FP8 and FP4 precision modes for efficient AI training and inference. The design also focuses on performance-per-watt improvements, catering to environments where power efficiency is as crucial as raw performance for sustained exascale operations.

A series of reports from Korean and Taiwanese outlets suggests growing pressure across the PC hardware industry as DRAM, NAND, and even NOR Flash prices continue to climb. Korea Economic Daily says NVIDIA, AMD, and other GPU makers are considering dropping some mid-to-high-end gaming cards where memory now makes up an unusually large share of total cost. Taiwanese PC brands, including ASUS, are also said to be looking at ways to reduce memory configurations in upcoming products. TrendForce warns that manufacturers may shift more production toward lower-margin models or raise prices across the board to cope with higher component costs. With most modern PCs, laptops, consoles, tablets, and phones now shipping with at least 16 GB of RAM, any further price spikes or supply shortages could force major brands to cut orders and increase retail pricing. Commercial Times reports that memory alone could add nearly NT$3,000 (about US $96) to basic office PCs next year.

According to Commercial Times, due to rising memory prices some motherboard makers and notebook ODMs vendors have paused new motherboard development or mass production. DRAM, NAND, and NOR Flash supply is tightening at the same time. DDR4 is becoming especially scarce as suppliers accelerate phase-outs and move mature-node capacity toward HBM and DDR5. WJ Capital Perspective estimates a DDR4 shortfall of around 70,000 wafers by the end of 2025, with 2026 unlikely to recover fully. NOR Flash demand is rising as well, driven by AI servers. Commercial Times says NVIDIA's GB200 NVL72 system already uses more than US $600 of NOR per rack, possibly reaching $900 within two years. This comes right after earlier reports that AMD is preparing GPU price increases and Samsung is raising memory prices by up to 60% due to tight supply.