NVIDIA's DLSS 4.5 was the one of the most interesting CES-centric announcements from the company's 2026 keynote. Today, the company confirmed to the community that DLSS 4.5 will be available for every GeForce RTX GPU from day one. This includes even the older RTX 20, RTX 30, and RTX 40 series, besides the latest RTX 50 series "Blackwell." To easily enable the latest DLSS 4.5, you can use the NVIDIA App with just a few clicks. Go to the Graphics tab and select "DLSS Override - Model Preset." You can apply this setting globally or on a per-program basis. Choose "Recommended," which sets Preset M to DLSS Performance mode, Preset L to DLSS Ultra Performance mode, and Preset K to the other available modes. Interestingly, this will work in more than 400 titles that support DLSS Super Resolution, and over 250 titles that work with the multi-frame generation of DLSS 4.

The most notable feature of the regular DLSS 4 was the shift from convolutional neural network (CNN) based AI models to a newer Transformer-based architecture, which significantly improved image quality. NVIDIA has since advanced this technology further with DLSS 4.5, updating the algorithm with an even more advanced Transformer model trained on a much larger dataset. This second-generation Transformer was developed with five times the computational investment of the first DLSS 4 Transformer upscaler and utilizes the FP8 data format, allowing for more processing with minimal precision loss. The new algorithm is more context-aware, and NVIDIA confirmed that the input for upscaling is used more intelligently, resulting in much more detailed and precise video output. For GeForce RTX 50 series GPUs, NVIDIA dedicated a dynamic multi-frame generation with 6x increase in rendered frames as an exclusive feature, while older RTX series are left on 2x.

Last year, curious-looking custom GeForce RTX 5090D 32 GB card designs emerged from China—at the time, these unofficial designs were based on the region specific GeForce RTX 5090D GPU. Apparently, a mix of AI research firms, DIY enthusiasts, and bootleggers had created blower-style cooling solutions for the best available "Blackwell" hardware. Many months later, a surprisingly comprehensive stack of similarly outfitted GeForce RTX 50-series products have popped up for sale on the Taobao shopping platform. A recent VideoCardz news piece highlights that these offerings are not produced locally by mainstream AIBs; instead unnamed AI companies have reportedly managed to produce blower-style products based on the entire range of "Blackwell" gaming GPUs, including the regionally banned GeForce RTX 5090 32 GB flagship model. An even more nerfed "V2" variant—making do with only 24 GB of GDDR7 VRAM—was released late last summer as a Chinese market exclusive, replacing its older 32 GB sibling.

In early 2026, Taobao listings advertise available stock of all GeForce RTX 5090 variants: standard, D V1, and D V2. Additionally, GeForce RTX 5080, RTX 5070 Ti, and RTX 5060 Ti (all) 16 GB options were included. According to VideoCardz's past weekend article, the cheapest model is listed at 3999 RMB (~$573 USD), with the range-topper (RTX 5090 standard) coming in at a whopping 28,999 RMB (~$4158 USD). Product photos likely show off higher-end cards. Smaller cooling solutions could be present on the less potent GB206 GPU-based model. Further rumors point to involved AI hardware manufacturers readying greater VRAM capacity blower-style custom cards—VideoCardz has picked up on whispers of GeForce RTX 5090 96 GB and RTX 5080 32 GB variants. The former could be enabled by a use of 3 GB GDDR7 memory modules, reflecting work done on NVIDIA's very official RTX PRO 6000 Blackwell range.

The non-appearance of GeForce RTX 50-series "Blackwell" SUPER graphics cards at CES 2026 came as a major shock for portions of the PC hardware community. Normally, NVIDIA sticks to a fairly regular release cadence—barring a few hiccups here and there—with the launches of new gaming products and mid-gen refreshes. Unfortunately, current market conditions have affected a wide swath of manufacturers—prompting price hikes, supply chain adjustments, launch postponements, and rethinks of next-gen technology strategies. As laid out in TechPowerUp's recent top five "biggest no-shows of CES 2026," Team Green's much-anticipated current-gen GeForce product refresh gets top billing.

Last Thursday (January 8), members of the Board Channel community engaged in speculative talk of NVIDIA's alleged indefinite delay of GeForce RTX 50-series SUPER GPUs. One participant outlined three main reasons behind this disappointing postponement, reportedly extracted from a brief issued to board partners. According to this leak, the first factor is production capacity being adjusted to favor compute GPU lines—current demand, especially overseas, needs to be satisfied. Most importantly, greater profit margins are being pursued in the extremely lucrative AI server segment. The second bullet point focuses on a long-term forecast for escalating memory-related product costs—a serious problem for modern graphics cards that utilize fancy GDDR7 VRAM modules.

Infortrend Technology, Inc., a leading provider of enterprise storage and AI solutions, today announced the launch of the KS 3000U edge AI server. The platform enables mid-sized organizations and distributed sites to deploy AI inference at the edge with high availability and simplified setup.

While AI workloads move closer to where data is generated, organizations that continue to rely on cloud services face challenges, including latency, high operational costs, and data privacy. According to IDC, by 2030, 50% of enterprise AI inference workloads will be processed locally at the edge. For companies with limited on-site IT resources, this shift highlights the need for reliable, easy-to-deploy on-site edge AI infrastructure.

GIGABYTE, the world's leading computer brand, showcased its AI TOP product line at CES 2026, highlighting that the rapid adoption of AI inferencing is accelerating the shift of AI from cloud to local. AI has become more accessible, responsive, and adaptable for everyday use, accomplishing the concept of "AI for everyone" in the AI era. Recognizing this transition, GIGABYTE introduced the AI TOP system in 2024 to enable local AI development with household-standard power, and has since expanded the lineup to support diverse usage scenarios ranging from personal projects to business deployments.

At the top of the product lineup is the AI TOP system, a fully customizable AI computing solution built with GIGABYTE AI TOP series hardware to match specific AI workload requirements. Designed to empower desktop users for local AI development, the AI TOP 500 system supports AI models up to 405B parameters and is tailored for medium-sized businesses seeking scalable AI computing power locally. For smaller businesses, startups, and individuals, the AI TOP 100 system supports LLM fine-tuning up to 110B+ parameters. To expand performance and accelerate the computing process, multiple AI TOP systems can be clustered through Ethernet and Thunderbolt ports, increasing training speeds and providing scalable power for more workloads.

As the Consumer Electronics Show of 2026 comes to an end, our field crew has done of a lot of work to keep readers updated. However, pre-CES expectations were high for several product launches, that sadly didn't happen. In today's piece we are ranking the top five things we didn't get at CES 2026, even with hopes of a product launch occurring during the show.

1) First on our non-weighted ranking is NVIDIA's GeForce RTX 50-series "Blackwell" SUPER refresh. This mid-cycle was originally planned for late Q1 to early Q2 2026, with a possible announcement at CES 2026. The "SUPER" treatment was planned with denser GDDR7 memory modules with 3 GB of capacity per chip, increasing the memory configuration of the standard GeForce RTX 5070, RTX 5070 Ti, and RTX 5080. Initially, the RTX 5070 SUPER was planned with an upgrade to offer 18 GB, while the RTX 5070 Ti SUPER and RTX 5080 SUPER would each provide 24 GB of GDDR7 memory.

At CES 2026, the Alphacool booth was brimming with a wide array of water cooling products. Among their staple offerings was the Core GPU AIO, part of Alphacool's Core series, providing a comprehensive cooling solution for various GPU models, including those from AMD and NVIDIA. This system is built around the Core GPU cold plate and features two Core DC-LT 2 pumps. A 360 mm radiator is included, enabling significant overclocking potential even for high-performance cards like the RTX 5090. It is planned for Q2 release of ths year. In addition to this, Alphacool showcased single-slot water blocks for NVIDIA's RTX 6000 Pro "Blackwell" GPU. These are available with either a transparent acrylic finish or a sleek opaque design. The single-slot configuration allows for multiple GPUs to be stacked together in a custom loop, making them ideal for dense AI workstation deployments.

At CES 2026 in Las Vegas, Digital Storm unveiled its latest pre-built PCs featuring custom chassis that rethink system layouts. One of the new models, the "Vector," redesigns the traditional setup by positioning the GPU at the top of the PC. The chassis measures 18x13x4 inches and accommodates an NVIDIA RTX 6000 Pro Blackwell GPU within its slim four-inch thickness. Perforations on both sides ensure adequate airflow for the GPU and radiator. The system is built on a Mini-ITX motherboard, compatible with either an AMD Ryzen 9 9950X or an Intel Core Ultra 9 285K processor. The Vector PC's exterior is adorned with vertical perforated stripes, and a tempered glass panel elegantly showcases the AIO cooler's ring that manages the CPU's temperature. Vector will start at $2,000 with planned availability in Q2 of 2026.

ASUS Republic of Gamers (ROG) today announced the arrival of a new generation of Zephyrus G14 and G16 gaming laptops, powered by cutting edge next-gen chips. Both the Zephyrus G14 and Zephyrus G16 feature models with Intel, AMD, and NVIDIA silicon, perfect for creatives and professionals on the go who need extra power. Both laptops also now feature models with a new ROG Nebula HDR OLED display capable of 1100 nits peak brightness, and all models feature a timeless CNC-milled aluminium chassis to ensure a precise, premium experience from end to end.

Next-gen power
Both the Zephyrus G14 and G16 are available with the latest Intel Core Ultra Series 3 processors. This family employs a modular, tile-based architecture to enable broad flexibility, from low power operations to peak performance that users would expect from an ROG Zephyrus gaming laptop. With up to 50 TOPS of NPU performance available, local execution of AI tasks and potential for in-game AI workflows make the Core Ultra Series 3 family a perfect choice for modern computing. The Zephyrus G14 GU405 features up to an NVIDIA GeForce RTX 5080 Laptop GPU, while the Zephyrus G16 GU606 features up to an NVIDIA GeForce RTX 5090 Laptop GPU. Featuring the latest NVIDIA Blackwell architecture and killer software features like DLSS 4 and Frame Generation, these laptops are built for next-gen gaming.

GIGABYTE, the world's leading computer brand, unveils the latest AORUS GeForce RTX 5090 INFINITY graphics card at CES 2026, showcasing a breakthrough design that combines compact dimensions with a next-generation cooling solution. Featuring the revolutionary WINDFORCE HYPERBURST Cooling System and GIGABYTE's first separated PCB layout, this latest flagship model redefines airflow performance and form factor footprint for elite gaming setups.

A well-designed cooling system is the key to unleashing the full potential of a GPU, and it has always been the top priority for every GIGABYTE graphics card. Powered by NVIDIA Blackwell, GeForce RTX 50 Series GPUs bring game-changing capabilities to gamers and creators. Equipped with a massive level of AI horsepower, the RTX 50 Series enables new experiences and next-level graphics fidelity. Multiply performance with NVIDIA DLSS 4, generate images at unprecedented speed, and unleash creativity with NVIDIA Studio.

ASUS today announced the showcase of ExpertCenter Pro ET900N G3 at CES 2026, underscoring its commitment to delivering enterprise-grade AI infrastructure in accessible, deskside form factors. The all-new ET900N G3—accelerated by the NVIDIA GB300 Grace Blackwell Ultra Desktop Platform—sets a new standard for deskside AI supercomputers, combining extreme performance, unified memory architecture, and data-center-class connectivity.

Solving the last-mile challenge of modern AI development
AI development has reached extraordinary heights through massive cloud and data-center infrastructure, yet many researchers, developers, and data scientists still face a 'last-mile' challenge. While cloud services offer immense scale, countless workflows require local, private, and immediate compute power—especially for training, fine-tuning, or running large AI models that exceed the limits of traditional workstations. What these professionals need is a system that delivers data-center-class performance in a familiar, easily deployable deskside form factor.

GIGABYTE, the world's leading computer brand, unveils an expanded AI PC vision with a redesigned GiMATE Interface and a new generation of AI gaming laptops at CES 2026. Built for clearer navigation and faster decision-making, the updated GiMATE experience brings real-time system insights and intuitive, voice-driven control into one streamlined hub, advancing GIGABYTE's commitment to truly adaptive, user-centric AI computing across its laptop lineup.

As GiMATE, GIGABYTE's exclusive AI agent for seamless hardware and software control, evolves, so does the ecosystem behind it, delivering smarter control for power, productivity, and creativity. The newly introduced AI Power Gear III enhances efficiency and flexibility with direct MUX switching between Discrete Mode and MSHybrid through GiMATE. It also brings smarter power behavior, the Auto Mode disables the GPU on battery to maximize runtime, the Optimus Mode delivers smoother, balanced performance, while the Eco Mode provides extreme power saving on AC power and battery. For creators and builders, GiMATE Creator expands its generative capabilities with the addition of Qwen-Image, supporting both English and Chinese prompts to bring powerful visual AI to users worldwide. Meanwhile, GiMATE Coder lowers the barrier to development by enabling natural-language code generation, correction, and optimization, making coding more accessible for beginners while accelerating iteration for experienced users.

NVIDIA today announced the new DLSS 4.5 technology at its 2026 International CES media event, to improve gaming performance and image quality, bringing higher resolutions and refresh rates within reach of PC gamers. DLSS 4.5 is a major update over DLSS 4 that debuted in 2024 with the GeForce RTX 50-series "Blackwell" graphics cards. It is designed to bring games with performance-intensive path tracing effects to a broader audience, with the fastest RTX 50-series GPU models being able to play path-traced games at 4K Ultra HD with 240 Hz.

At the heart of the DLSS 4.5 update are two key announcements—2nd Generation Super Resolution Transformer, and Dynamic Multi Frame Generation. One of the defining features of DLSS 4 is its high-quality upscaler, which replaced convolutional neural network (CNN) based AI models with a newer Transformer-based model, resulting in a significant improvement in image quality. Now, at CES, NVIDIA announced that they have updated their algorithm with an even newer Transformer-based model that has been trained on a much larger dataset. This 2nd Gen Transformer was developed using five times the compute power that went into the first-gen DLSS 4 Transformer upscaler, and leverages the FP8 data format that's supported by the Tensor cores of GeForce RTX 40-series "Ada" and 50-series "Blackwell" GPUs. DLSS 4.5 is still supported on older 30-series "Ampere" and 20-series "Turing" GPUs, however these generations lack FP8 acceleration, so the performance uplift over native wouldn't be as significant as with newer GPUs.

MSI has officially unveiled the specifications of its new AI-centric mini PC line-up that is being debuted at CES 2026. The new line-up offers something for everyone—as long as you're looking to run local LLMs or similar AI workloads—whether you're looking for a hyper-compact x86 system powered by AMD's RDNA 3.5, a more powerful NVIDIA RTX 5070, or something more exotic with NVIDIA's new Arm-based DGX Spark platform.

Perhaps the most interesting mini PC that MSI is showing off at CES is the EdgeXpert, which is targeted at AI developers and is built on NVIDIA's DGX Spark platform. In the EdgeXpert, which was announced earlier this year, that means 20 Arm CPU cores and an AI Blackwell GPU as well as NVIDIA's NVLink C2C tech, which claims to offer "five times the bandwidth of PCIe 5.0," for improved data transfer and sharing between the CPU and GPU. According to MSI, the EdgeXpert is capable of a petaFLOP of AI performance. It also features 128 GB of LPDDR5x unified system memory and a vapor chamber cooling system.

Digital Storm today unveiled two new desktop platforms at CES 2026, Vector and Aventum 5, which showcase two distinct approaches to modern high-end computing: an unapologetically extreme liquid-cooled tower and a dramatically slimmer performance desktop that condenses serious hardware into a compact footprint.

Rethinking High-End Desktops
For the last decade, many high-end desktops have grown bigger, louder, and more power hungry in the race for peak numbers. Vector and Aventum 5 take a different path, focusing on how a PC performs and behaves over hours of rendering, creation, or AI workloads, not just what it hits in a 30-second benchmark run. Both platforms are engineered around modern CPUs from Intel and AMD and graphics options up to NVIDIA RTX 6000 Pro Blackwell.

Inno3D's Chinese website has listed a brand-new iChill "Super Ice Dragon" GeForce RTX 5060 Ti 16 GB graphics card model. On initial inspection, this oversized triple-fan/triple-slot custom design appears to be visually "louder" when compared to a nearby traditionally-shaped sibling. The more angular design seems to house an "optimized NVIDIA reference PCB." According to a preview outlet's report, this board—reportedly bearing a "Model 00-A1" moniker—features traces for 8-pin and 16-pin (12V-2x6) power connectors. Currently, Inno3D's latest card sports a traditional 8-pin input. VideoCardz surmises that a "higher-power variant" could be in the works; enabled via an alternative avenue present on the hybrid PCB.

Traces, pads, and extra room on the board's layout indicate provisions for a futuristic revision. So far, Inno3D has opted into outfitting its GeForce RTX 5060 Ti custom cards with 8-pin power connectors, but a couple of NVIDIA AIBs have gone down the 12V-2x6 route for this class of GPU. Industry observers predict an uptick in adoptions of 16-pin designs within the mid-range segment. It is possible that the freshest reference PCB layout is paving the way for speculated GeForce RTX 5060 Ti and non-Ti SUPER variants. On the flip side, Inno3D could be readying a new wave of easy-to-prepare factory overclocked iChill products.

Back in December, AMD's current RDNA 4 GPU lineup apparently showed little impact on Steam's Hardware and Software Survey results. However, as we enter 2026, data from December 2025 has been processed, and the first Radeon RX 9000 series GPU has appeared in the survey rankings. At 100+ ranking, the first AMD Radeon RX 9000 series GPU based on the RDNA 4 IP is the Radeon RX 9070 SKU, commanding a 0.22% share of all gamers participating in the survey. This is also the only RDNA 4 GPU present on the list, meaning that all the remaining SKUs are below 0.15% of the total share, hence not shown in this ranking.

For other AMD GPU SKUs, there is a slight change in the overall structure, where the RDNA 3-based Radeon RX 7800 XT is the most popular AMD gamer GPU with a 1.57% share this December. This dethrones the previous RDNA 2-based Radeon RX 6600, which now stands at 1.51% of the overall group. This includes the market share of AMD GPUs on both Windows and Linux-based PCs. Additionally, this is the first time RDNA 4-based GPUs have appeared on the Steam Survey list since the initial launch in March, finally showcasing the share of the latest AMD graphics.

Eurocom is proud to announce the launch of the EUROCOM Raptor X18 - a mobile server platform that brings enterprise-grade, rack-class performance into a single, portable chassis. Built to deliver full server functionality without the constraints of traditional data center infrastructure, the EUROCOM Raptor X18 gives organizations the freedom to deploy high-performance computing, private cloud hosting, edge-computing nodes or temporary network infrastructures - anywhere, anytime.

EUROCOM Raptor X18 is powered by Intel Core Ultra 9 Processor 275HX with 24C/24T 36M cache and NVIDIA GeForce RTX 5090 24 GB DDR7 with Blackwell 328 Tensor AI cores architecture. It supports up to 256 GB of DDR5 memory, up to 32 TB of NVMe SSD storage in RAID 0/1/5, two LAN ports and two additional LAN ports via TB5 ports.

The NVIDIA RTX PRO 5000 72 GB Blackwell GPU is now generally available, bringing robust agentic and generative AI capabilities powered by the NVIDIA Blackwell architecture to more desktops and professionals across the world.

The new GPU configuration offers AI developers, data scientists and creative professionals the hardware for modern, memory-hungry workflows—and arrives at a time when demand for NVIDIA Blackwell-class compute is higher than ever.

Reports from Asian supply chain sources suggest that NVIDIA might cut back on making GeForce RTX 50 series GPUs in the first six months of 2026, Videocardz reports. According to Board Channels, overall supply during H1 2026 could be down by roughly 30-40% compared to the same period in 2025, although the information remains unconfirmed. The report points out that the issue is not entirely related to GPU VRAM alone, but also to tight availability across GDDR6, GDDR7, and other memory components, including motherboard-related memory such as DDR5/DDR4. Board Channels also claims NVIDIA may adjust allocation strategies for add-in board (AIC) partners in Mainland China to better align supply with changing DIY market conditions.

Benchlife appears to support parts of the rumor, though without specifying a percentage reduction. Instead, it reports that two models (best value models in the Blackwell RTX 50 lineup), the GeForce RTX 5070 Ti and RTX 5060 Ti 16 GB are likely to see the earliest supply adjustments. While both Board Channels and Benchlife have long and good track records, the reports are based on regional sources and may not reflect global supply conditions. We will have to wait for NVIDIA or its partners to provide more information. However, it's worth noting that just a month ago, reports from Korean and Taiwanese outlets said AMD and NVIDIA were already considering GPU output cuts, while ASUS and other vendors were slowing motherboard plans due to memory shortages. In the end, this could end up being a broader, global production adjustment by NVIDIA rather than a region specific move.

ASUS has expanded its DUAL graphics card family with a smattering of lower-end models—based on NVIDIA's GB206 "Blackwell" GPU—that offer a slimmer profile. Brand-new listings, added in a low-key manner—are tagged with "EVO" designations. As highlighted by Vortez, the DUAL EVO design comes in at "2.1-slot" thickness (42 mm). This slight shrinkage—mostly in one dimension: width—distinguishes new entries from standard DUAL options. The latter examples exist as 2.5-slot thick (50 mm) offerings. There are negligible differences—between twin axial tech-fan cooled EVO and non-EVO SKUs—in terms of card length and height.

At a glance, ASUS has only introduced this slimmer shroud design across overclocked (OC) edition and non-overclocked GeForce RTX 5060 Ti 16 GB and RTX 5060 8 GB models. Confusingly, this generation's EVO aesthetic actually debuted mid-way through 2025, under the guise of vanilla DUAL GeForce RTX 5050 graphics card options—all truly 2.0-slot (40 mm) thick. This brand-new wave of cards has lead to an adjusted placement of single 8-pin power connectors—DUAL GeForce RTX 5060 Ti and RTX 5060 EVO designs have this aspect positioned closer to their I/O brackets. The TechPowerUp readership can familiarize itself with a myriad of DUAL EVO and DUAL options present within the site's well-maintained GPU database. Fresh entries popped up before the publication of this news piece.

NVIDIA's recent generations of GPUs, such as "Hopper" and "Blackwell," have been known for stalling their high-performance computing (HPC) oriented FP64 double-precision performance. However, the company is not inherently abandoning the HPC space and has confirmed for HPCWire that 64-bit floating point data is not going away. According to Dion Harris, the senior director of HPC and AI hyperscale infrastructure solutions at NVIDIA, the company is "definitely looking to bring some additional [FP64] capabilities in our future gen architectures. We are very serious about making sure that we can deliver the required performance to power those simulation workloads."

The acceleration of 64-bit floating-point data paths is crucial for the HPC community, particularly in the life sciences. When a workload demands sustained high-precision support, NVIDIA's recent generations have not met expectations. For comparison, NVIDIA's current most powerful B300 "Blackwell Ultra" accelerator achieves only 1.2 TeraFLOPS of FP64 performance. In contrast, the older H200 "Hopper" reaches an impressive 34 TeraFLOPS of FP64 compute at its peak. For FP8 low-precision, the B300 delivers 9 PetaFLOPS, while the H200 provides 3.958 PetaFLOPS.

NVIDIA has reportedly developed a chip-tracking technology that will pinpoint the exact locations its GPUs are deployed in. According to a report from Reuters, this technology is a software tool that detects exactly where and how its chips are used to fight GPU smuggling networks from escaping the official U.S. sanctions. For example, NVIDIA's latest "Blackwell" GPUs have been banned from selling to Chinese entities, and GPU smugglers in other countries have been able to stock up on these and export them, bypassing the official U.S. sanctions. Just recently, NVIDIA got a green light from the Trump administration to sell its H200 "Hopper" GPUs to Chinese customers, but that excludes the more powerful "Blackwell" generation.

NVIDIA for ReutersWe're in the process of implementing a new software service that empowers data center operators to monitor the health and inventory of their entire AI GPU fleet. This customer-installed software agent leverages GPU telemetry to monitor fleet health, integrity and inventory.

The Trump administration has approved a framework allowing NVIDIA to resume exports of its H200 data center accelerators to China, subject to a revenue-sharing condition that allocates 25% of sales proceeds to the U.S. government. Rather than lifting export restrictions outright, the arrangement requires that recipients be vetted through Commerce Department licensing, so shipments will be controlled and limited to approved commercial customers. The decision reopens a major market for high-end accelerators while maintaining a regulatory channel to monitor distribution.

The H200 chip remains a powerful solution for large-scale AI workloads even today. Manufactured using TSMC's N4 process, it combines NVIDIA's "Hopper"-derived architecture with 141 GB of HBM3E memory and 4.8 TB/s memory bandwidth, making it ideal for large model training and dense inference tasks. Although it doesn't match the performance of NVIDIA's current "Blackwell" family and other upcoming accelerators, the H200 is a mature product that has been shipping since spring 2024 and benefits from a well-developed driver and software ecosystem. Its raw compute power is roughly double that of solutions like the Huawei Ascend 910C, making it very competitive even today.

Interesting developments are unfolding in the world of DRAM and NAND flash, as one of the largest players, SK Hynix, has announced product updates for 2026. The company appears to be following its competitor Samsung by slowing the ramp-up of its next-generation HBM4. Initially, SK Hynix planned to increase its HBM4 capacity in the second quarter of 2026. However, this has been postponed to the third quarter of 2026, as the company plans to expand its HBM4 manufacturing capacity. The delay is due to the high demand for its current HBM3E capacity, necessitating the continued operation of HBM3E manufacturing sites longer than expected. SK Hynix's largest customer for HBM chips, NVIDIA, is reportedly facing challenges with the volume production of "Rubin," delaying the need for HBM4, while the HBM3E-equipped "Blackwell" remains in strong demand.