News Posts matching #GPU

Return to Keyword Browsing

Intel Arc Discrete GPUs Hit 1% Market Share for First Time

Jon Peddie Research has released a new report on the market share of discrete GPUs, noting that Intel achieved its first single-digit percentage breakthrough since the launch of its Arc "Alchemist" GPUs for gamers. This is not a small win, as Intel historically had around half a percent share among dGPUs. Intel's late entry into the overall dGPU market has caused it to gain market share glacially slowly. It hasn't been able to disrupt the already established group that includes AMD and NVIDIA, which are commanding the remaining 99% of the market. However, despite it being only a 1% share, Intel's breakout is notable for the first time, giving Intel dGPU gamers and team blue hope to continue investment in this market.

For the remaining 99%, AMD and NVIDIA split the market a bit differently now. AMD's market share increased by 0.8% from last quarter to 7% in Q3 2025, while NVIDIA remained at a strong top with a decrease of 1.2%, putting it at 92% of the overall AIB market share. Market dynamics such as availability and pricing are all impacting the buyer decision-making process, so a slight inventory adjustment from AIBs could translate to a big increase or decrease in market share from quarter to quarter, as we see with NVIDIA here. Hence, smaller fluctuations and market share "exchange" between AMD and NVIDIA are happening often, while smaller players like Intel seem to be rising.

NVIDIA Reportedly the Only TSMC A16 Node Customer

NVIDIA is reportedly the only customer queued for TSMC's next-generation A16 process, and the company is lining the node up for its upcoming "Feynman" GPUs. The company is lining up for samples in 2026 with volume ramps following in 2027, timing that would slot "Feynman" after "Rubin"-class products built on refined 3 nm variants. If the reports are accurate, NVIDIA's commitment would be notable as it would be the only major company opting for the "stopgap" node between N2 and A14. Customers like Apple are reserving TSMC N2 2 nm capacity and plan to transition to A14 as soon as it becomes available, bypassing the A16 node.

TSMC's A16 node is a nanosheet-focused design featuring enhanced backside power delivery, known as Super Power Rail (SPR). This design separates power routing from signal layers, reducing delivery losses. The process is expected to provide modest single-digit performance improvements, more noticeable power reductions for AI workloads, and slightly higher transistor density compared to the previous generation. For very large dies and high-power cards, these enhancements can lead to cleaner floorplanning, easier thermal management, and increased capacity for memory and interconnect bandwidth. While these improvements are significant for NVIDIA's datacenter chips, Apple's mobile and laptop chips may benefit even more from the less expensive N2 node and its variations.

NVIDIA Refills GeForce RTX 5090 Founders Edition GPU Stock in Europe

NVIDIA has officially restocked its GeForce RTX 5090 graphics card in European stores, as reported by recent updates from online marketplaces. NVIDIA operates its own online market across various countries, allowing gamers to purchase the GPU at the manufacturer's suggested retail price (MSRP). In September, NVIDIA quietly removed the GeForce RTX 5090 and RTX 5080 Founders Edition graphics cards from its online marketplace. This removal affected all international online stores, including those in the US, Germany, the UK, and France. However, the company later confirmed that the GeForce RTX 50 series Founders Editions continue to be in production. They are limited edition products, so they occasionally go out of stock on the website and return when available again.

It seems that today is the day again, as NVIDIA has released a fresh batch of GeForce RTX 5090 Founders Edition SKUs on European marketplaces. This makes the holiday season more appealing for gamers who have saved up to buy the highest-performing GPU currently available. The RTX 5090 FE SKU is currently priced at 2,099 EUR, but stock could run out quickly since it's unclear how many units NVIDIA has allocated for its European markets. With GDDR prices on the rise, there are concerns about the long-term sustainability of the current MSRPs. Therefore, purchasing this GPU at the current price and availability might not be as easy in the future.

Fuxi A0 GPU Demoed at Chinese CAD Expo - Features Imagination Technologies DXD Architecture

Earlier in the year, Imagination Technologies its newest GPU IP: DXTP. This efficiency-oriented architecture was designed with smartphone and (other) power-constrained platforms in mind. Nine months later, a derivative of this tech has turned up housed in discrete formats—targeting workstation applications. At the recently concluded International Conference on Computer-Aided Design (ICCAD) 2025 expo, Xiang Dixian exhibited their Fuxi A0 graphics card. Demo units were on display at this late November event; handleable and hooked up within a working system. A localized version of the Lenovo New Vision Z-engine 3D platform was up and running on demo hardware.

According to reports from the region, Xiang Dixian's Fuxi A0 card—based on a 5 nm process node—has already entered into a mass production phase. Apparently, this model is optimized for rendering and gaming—company representatives claim that their GPU is capable of running Black Myth: Wukong at pleasing frame rates. A past weekend Wccftech article alludes to another variant: B0. Marketed as a training and inference-centric alternative to the A0 model. An onboard NPU offers "FP8 capabilities." It is claimed that the latest Fuxi GPUs can compete with familiar mainstream cards in terms of ray tracing and upscaling (super resolution) performance.

Ayaneo Next II Gaming Handheld Will Feature AMD Strix Halo and 9-inch OLED Display

The Ayaneo Next II was recently revealed as a new gaming handheld powered by AMD's Ryzen AI Max+ 395 APU and its Radeon 8060S iGPU. Aside from those details and a few tidbits about the controller specifications, however, not much had been revealed about the new gaming handheld until now. Ayaneo recently posted a few additional details about the new plus-size Strix Halo gaming handheld on X, mostly centering on screen specs. Ayaneo also teased a launch event for November 28, 2025 at 23:30 PT (7:00 UTC).

According to the post, the Ayaneo Next II will have a 9-inch OLED display with a 165 Hz refresh rate, 2400×1504 resolution, and a native landscape orientation. This works out to a screen ratio of just under 3:2 (3:1.88, to be exact), and Ayaneo claims that this display is a custom, exclusive design, although it seems somewhat unlikely that it will stay this way for very long, given how competitive the high-performance gaming handheld market is getting these days—at the very least, we can likely expect other devices from Ayaneo to feature the same display at some point. Curiously, Ayaneo makes no mention of whether the display will feature VRR, although it seems unlikely that such a premium gaming handheld wouldn't.

AMD Plans to Release Radeon AI Pro R9700S and R9600D Graphics Cards

AMD is preparing two new professional GPUs: Radeon AI Pro R9700S and Radeon AI Pro R9600D, built off the Radeon AI Pro R9700 that launched on October 27. References to the R9600D recently appeared in AMD's 25.30.1 Linux driver, and reports suggest both cards are planned for announcement around CES 2026. The R9700S is described as a server-focused variant of the standard R9700. Instead of the blower cooler used on the original model, the R9700S is expected to switch to passive cooling for use in data-center systems that rely on chassis airflow. The R9600D is positioned as a cut-down version of the same GPU, similar in naming strategy to AMD's earlier Pro W7900D.

Since both cards are rumored to be based on the existing R9700, their general configuration should be similar. The R9700 uses a 4 nm Navi 48 GPU with 64 RDNA 4 compute units, totaling 4,096 stream processors and 128 AI accelerators. It ships with 32 GB of 20 Gbps GDDR6 on a 256-bit interface for 640 GB/s of bandwidth, backed by 64 MB of 3rd-generation Infinity Cache. AMD rates the card at up to 191 TFLOPS of FP16 dense compute or 1531 TOPS of INT4 performance, with a 300 W power target. The R9700 launched as a dual-slot, blower-style card designed for multi-GPU inference systems. The upcoming R9700S and R9600D would expand this lineup, offering a passive server option and a lower-tier configuration built from the same RDNA 4 foundation.

NVIDIA May Stop Bundling Memory with GPU Kits Amid GDDR Shortage

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.

Qualcomm Unveils Snapdragon 8 Gen 5 Mobile Platform

Qualcomm Technologies, Inc. today announced the Snapdragon 8 Gen 5 Mobile Platform, a premium offering that combines outstanding performance with cutting-edge technologies, raising the bar for flagship mobile experiences. Snapdragon 8 Gen 5 joins our premium tier family, offering consumers greater choice and flexibility. It is premium at its core and meets the rising demand for cutting-edge capabilities - from AI and camera innovations to immersive gaming experiences.

With the new 8-series Mobile Platform, users can instantly wake their AI assistant up simply by picking up their device, thanks to the Qualcomm Sensing Hub, which intelligently combines microphone and sensor inputs to detect users' intent to speak. Powered by the Qualcomm AI Engine, Snapdragon 8 Gen 5 also enables agentic AI assistants to deliver context-aware interactions and personalized suggestions - elevating mobile experiences with seamless, on-device intelligent experiences. At the core of these capabilities is the Qualcomm Hexagon NPU, which achieves up to 46% improved performance.

Tactical Multiplayer FPS Bodycam Gets Major Update With Switch To Unreal Engine 5.5

We're really happy to be back with this new update. Thanks again for your patience, we know it took some time, but we wanted to push as much as possible to deliver something truly solid and meaningful for the game.

Quick Recap on the Update!
For those who may not have followed all our recent communications: since the launch, the team has grown a lot and we're now over 100 passionate people working on Bodycam! We recently released a devlog showing a lot of upcoming features. Some of them are included in this update, others aren't here yet, and that's completely normal. We're now working on several updates in parallel, so some features will arrive in the next major update. If you want to get a preview of what's coming, you can rewatch the devlog, especially Part 2. We'll also release a new devlog soon focused entirely on the next major update. Thanks again for all the support.

AI Boom Drives Demand for Ultra-Large Packaging as ASICs Expected to Shift from CoWoS to EMIB

TrendForce's latest investigations reveal that the rapid expansion of AI and HPC is increasing the need for heterogeneous integration, positioning advanced packaging as a strategic priority. TSMC's CoWoS platform is currently the leading solution in this area. However, as CSPs accelerate their in-house ASIC development to accommodate more complex functions, their packaging size requirements are growing substantially. As a result, some CSPs are contemplating a switch from TSMC's CoWoS to Intel's EMIB.

TrendForce notes that CoWoS connects compute logic, memory, and I/O dies via an interposer, which mounts them on a substrate. The technology has expanded to include CoWoS-S, CoWoS-R, and CoWoS-L. Demand has been strongly shifting toward CoWoS-L, which incorporates a silicon interposer into the package, as NVIDIA's Blackwell platform approaches mass production in 2025. This trend is expected to continue with NVIDIA's upcoming Rubin architecture, which will boast even larger reticle sizes.

Chieftec Launches The Cube mATX Case for HTPC builds

Chieftec has officially announced The Cube, a new compact microATX case that seems almost specifically designed to blend into a living room gaming setup or serve as a home-theater PC build. The Cube CI-03B-OP, as it is formally known, has a front intake grill and a smooth matte black finish that makes it look like a subwoofer in a speaker setup instead of like a loud, garish PC setup. The whole chassis measures in at just 385 × 265 × 355 mm, and features a healthy helping of front I/O, including 1× USB 3.2 Gen 2 Type-C port, 2× USB 3.0 Type-A ports, and a 3.5 mm audio combo jack. The motherboard is curiously positioned horizontally inside the case, even when the case is upright, a design that might result in a starved GPU in some situations, since there is no ventilation on the GPU fan side.

[Editor's note: Our in-depth review of The Cube is now live]

The Cube supports up to a 240 mm radiator and comes with a single 200 mm PWM fan pre-installed. Some other neat features include a "fold-open mechanism," which basically allows you to easily remove the front and top panels of the case quickly for improved access to build in the case. Internally, things are tight, with CPU cooler clearance limited to just 160 mm and video cards similarly limited to 355 mm. Aside from the pre-installed 200 mm front intake fan, The Cube supports up to a single 120 mm rear exhaust fan and dual 120 mm fans, a single 200 mm fan, or a 240 mm radiator on the top panel. It also supports a full size ATX PSU up to 160 mm in length and two 3.5-inch HDDs and two 2.5-inch SSDs, providing reasonable expansion options. It's unclear if The Cube will be available in the US at any point, but it is currently priced at €69.90 in the EU.

AMD, Cisco and HUMAIN to Form Joint Venture to Deliver World-Leading AI Infrastructure

AMD, Cisco and HUMAIN, a PIF company delivering global full-stack AI solutions, announced today plans to establish a joint venture aimed to support the development of Saudi Arabia's position as a leading provider of world-class AI solutions for regional and global customers. AMD, Cisco and HUMAIN will serve as founding investors in the joint venture, which is expected to begin operations in 2026 with plans to combine HUMAIN's state-of-the-art data centers with AMD and Cisco technology, delivering modern data center capacity with efficient power and lower capital expenditures. AMD and Cisco will act as exclusive technology partners to the joint venture, contributing their portfolio of products and services to its development of up to 1 GW of AI infrastructure by 2030.

The companies also announced the first phase of the project as a buildout of 100 MW AI infrastructure, with the intent to include HUMAIN modern data center capacity, AMD Instinct MI450 Series GPUs and Cisco's industry-leading critical infrastructure.

Qualcomm Says 90% of Games Run on Snapdragon X2 Elite, Outperforming Intel and AMD

Qualcomm is ready to push its Snapdragon X2 Elite and Elite Extreme SoCs into a sector that was previously dominated by x86 SoC makers, and that is gaming. The company now claims that its SoC can run more than 90% of the most played games available on the market on the initial launch. However, this is not the only striking claim, as Qualcomm has benchmarks that prove that its Adreno X2 iGPU is 50% faster than the iGPU inside Intel Core Ultra 9 288V "Lunar Lake" and 29% faster than the iGPU on AMD Ryzen AI 9 HX 370 "Strix Point," under the same graphics settings. This is of course, no small win, and Qualcomm expects to win hearts of gamers with its next-generation platform.

The new chip delivers frame rates up to 2.3x higher than its predecessor while supporting many popular Windows games. Its architecture uses four processing blocks that give the flagship X2 Elite Extreme an edge over competing integrated graphics. Each block efficiently processes graphics work before passing it to shader processors organized around two micro-units. These units each contain 128 KB of memory feeding 128 ALUs. The design supports multiple precision formats, employs a more efficient execution model, and can process two instructions at once to maximize its 128-ALU groups. In total, the GPU has eight shader processors and 2,048 ALUs running at up to 1.85 GHz, backed by 21 MB of on-chip memory, 2 MB of cache, and 228 GB/s of memory bandwidth on the flagship model.

GPU Price Hikes Coming in 2026, Warns PowerColor

A series of shocking news about price hikes continues to unfold, as we now receive official "warnings" from GPU AIB partners about the situation. On the official PowerColor subreddit, a representative noted that Holiday 2025 discounts are about to commence, but also claimed that prices will rise in 2026, urging gamers to buy before the increase. While this could be a marketing tactic, the recurring theme of warnings is becoming more prominent in the supply chain. Previously, we reported that DRAM prices have risen significantly—171.8% year-over-year—driving costs to new levels for this essential component found in every GPU, smartphone, PC, and console.

South Korean memory giants, such as Samsung and SK Hynix, are currently unable to fulfill all orders, completing only 70% of them. Smaller OEMs and channel distributors have been informed to expect only 35-40% fulfillment through the first quarter of 2026. Even some larger AIBs, like PowerColor, may face shortages in DRAM supply, potentially leaving their GPUs without installed DRAM modules. This situation not only delays their planned product rollouts but also jeopardizes their expected revenue streams if it persists. They are faced with the choice of either gambling on the spot market for a significant markup or leaving their production lines idle. Additionally, the shortage could ultimately increase the price of GPUs that are shipped, as they now contain a rarer commodity—DRAM.

TOP500: El Capitan Retains #1 as JUPITER Becomes Europe's First Exascale System

The 66th edition of the TOP500 list of the world's most powerful supercomputers was announced today at the SC25 Conference in St Louis, Missouri. The new list reflects continued U.S. leadership in high-performance computing (HPC), historic European milestones, and growing global diversity across architectures and energy-efficient design.

El Capitan Extends Leadership Across Benchmarks
The El Capitan system at Lawrence Livermore National Laboratory (LLNL) remains the undisputed leader of global HPC. Based on the HPE Cray EX255a architecture, powered by AMD 4th Gen EPYC CPUs and AMD Instinct MI300A accelerators, El Capitan achieved 1.809 Exaflop/s on the HPL benchmark—a significant remeasurement that further strengthens its position as the fastest computer in the world. El Capitan also dominates the HPCG benchmark, delivering 17.41 HPCG-Petaflop/s, making it the #1 for real-world application performance. On the HPL-MxP mixed-precision benchmark, it recorded 16.7 Exaflop/s, confirming its versatility in AI and data-driven workloads. With 11.34 million cores and an energy efficiency of 60.9 GFlops/Watt, El Capitan exemplifies the current generation of exascale computing.

AMD Details AI Innovations for Windows 11 PCs at Microsoft Ignite

Since Microsoft Build 2025, AMD has accelerated AI innovation across the Windows ecosystem through deep collaboration with Microsoft. Together, we are unlocking the full potential of Windows ML and Microsoft Foundry on Windows, an evolution of Windows AI Foundry. With the latest AMD Ryzen AI 300 Series processors and Radeon GPUs, AMD delivers fast, more efficient, and intelligent on-device AI performance. This partnership empowers developers to create the next generation of AI-powered Windows applications, delivering tangible benefits to customers and businesses without requiring cloud connectivity.

Accelerating AI on Windows with AMD Hardware
Our ongoing collaboration with Microsoft centers on optimizing Windows ML for AMD hardware accelerators -- GPUs and NPUs. These optimizations deliver high performance, improved efficiency, and broader support for the most advanced model architectures.

VKD3D-Proton Adds FSR 4, Even for Older Radeon GPUs

While AMD's shiny new FSR 4 is ostensibly exclusive to Radeon RX 9000 series, there have been a number of successful attempts to get it running on older AMD Radeon GPUs after a rather sizeable mistake on AMD's part led to the FSR 4 libraries leaking online. Now, however, a Linux developer working for Valve has implemented a fix to get FSR 4 running on older GPUs right into the Proton translation layer responsible for running Direct3D 12 via the Vulkan API on Linux. Support for FSR 4 is only one of a bunch of updates to D3VK-Proton shipping in the latest 3.0 update.

As of version 3.0 of VKD3D-Proton, the translation layer responsible FSR 4 is present for RX 9000 series Radeon GPUs using the official method, but the Valve developer behind the project, Hans-Kristian Arntzen, has also implemented a "hacky emulation path" that allows FSR 4 to work on older Radeon GPUs. The developer notes that this implementation uses int8 and float16 cooperative matrix support, and that it may come at "significant performance cost." It is, as such, not officially supported by the default build of VKD3D-Proton, and the current official implementation is also mostly a first step in the process of getting FSR 4 implemented "in a more proper way" into Valve's Proton itself. Support for FSR 4 may not be relevant to devices like the Steam Deck, due to the increased performance hit, but it may become more important around the release of the Steam Machine, which still relies on RDNA 3 GPU cores but is expected to provide around the same performance as an AMD Radeon RX 7600M.

ASUS GeForce RTX 5090 ROG MATRIX Priced at $3,999, Only 1,000 Units Available

ASUS has officially listed its GeForce RTX 5090 ROG MATRIX 30th anniversary limited-edition GPU for $3,999 on its U.S. store. Limited to 1,000 units, this is ASUS's most powerful and expensive card for the RTX 50-series "Blackwell" generation. The build quality justifies the price, as ASUS uses 3 oz copper PCB layers to enhance heat dissipation and minimize voltage drop across components, resulting in improved overall performance and potentially higher overclocking capabilities. The GPU features four large fans, similar to ASUS's RTX 5090 Astral GPU, with three fans in the front and one in the back, utilizing a push-pull air cooling design. The GPU shroud has a prominent ROG red design with a distinct ring on one fan, featuring an acrylic plate with the ROG MATRIX logo.

ASUS also offers an option for BTF-supplied power delivery, along with a 16-pin 12V-2x6 connector, allowing for a total power setting of up to 800 watts, which significantly exceeds the regular 575 watts of the Founders Edition GPU. ASUS claims this supplies a 10% performance boost on its own, and the boost clock frequency is confirmed at 2,730 MHz regular, and 2,760 MHz in OC BIOS mode. Embedded within the GPU is a miniature Bosch Sensortec BMI323 inertial measurement unit (IMU), which combines an accelerometer and gyroscope in a single component. ASUS integrates this sensor into the PCB to continuously monitor whether the graphics card remains level with the motherboard, triggering an immediate alert if the card shifts even slightly downward. More features are listed on ASUS website.

AMD & NVIDIA Reportedly Consider GPU Cuts, ASUS and Others Slow Motherboard Plans Amid Memory Shortage

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.

AMD Reportedly Planning GPU Price Increase as Memory Costs Spike

AMD is reportedly preparing for a round of GPU price increases, with rising memory costs cited as the reason. Posts on the Chinese Board Channels forum claim AMD told partners that memory prices have climbed enough that its next wave of GPUs will ship at higher prices. The first price increase, back in October, was small and never showed up at retail, but this new increase is said to be larger and apply to all models, from Radeon gaming cards to workstation GPUs and AI parts. There's no confirmed date, and AMD hasn't commented publicly. The memory market has tightened sharply as AI demand consumes available supply. DDR5 prices have jumped as much as 60% since September, and broader DRAM pricing is up roughly 170% year over year. GDDR6, the memory used in most graphics cards, has climbed by around 30% as manufacturers shift more capacity toward server DDR5, HBM, and AI hardware.

For gamers, the timing isn't great. Radeon RX 9070 XT prices had only just started to stabilize; one ASRock Challenger card in the US recently slipped to its official $599 MSRP, and European pricing has finally settled near launch levels after months of inflated street prices. NVIDIA is also rumored to be planning a similar price hike in early 2026. Neither AMD nor NVIDIA is driving the increases, both are caught in the same memory crunch. As usual with reports like this, it can take weeks before any change appears at retail.

NVIDIA Moves to In-House Server Production, Leaving OEMs with a Secondary Role

NVIDIA plans to reshape the industry's approach to building high-performance AI servers with its "Vera Rubin" product stack, which will enter volume production as the VR200 line in late 2026. Rather than supplying CPUs and GPUs to OEMs, NVIDIA plans to ship fully finished L10 compute trays that come pre-populated with "Vera" CPUs, "Rubin" accelerators, memory, 800G NICs, 110kW power delivery, and liquid-cooling infrastructure, all of which are validated before they leave the factory. This, however, will leave many system vendors and OEMs shifting away from designing core server electronics and toward tasks such as rack-level assembly, power configuration, installation of rack cooling sidecars, and final certification.

Rubin silicon is rumored to have a TDP of up to 2.3 kW for highest-end models, leading to rack-level power consumption exceeding 250 kW and making custom cooling designs economically unviable. NVIDIA's strategy reportedly involves standardizing cooling and power across two performance tiers while increasing network I/O density. These choices decrease the feasibility and cost-effectiveness of third-party replication. By centralizing production with select partners, development cycles can be shortened and manufacturing yields improved. However, this approach also limits the options OEMs and hyperscalers have traditionally used to differentiate their hardware.

Eviden Unveils its BullSequana XH3500, a Converged Supercomputing Platform for the AI-HPC Era

Eviden, the Atos Group product brand leading in advanced computing, today announces the launch of BullSequana XH3500, its latest supercomputer built to support convergence of High-Performance Computing (HPC) and Artificial Intelligence (AI) workloads. Designed to scale beyond exascale performance, the BullSequana XH3500 combines cutting-edge hardware with a comprehensive software and application ecosystem, helping organizations accelerate scientific discovery, industrial innovation, and AI-driven breakthroughs.

Building on over two decades of HPC leadership, Eviden introduces the BullSequana XH3500 as a natural progression from the BullSequana XH3000, refining its hybrid computing strategy and harnessing the technological foundations laid by its predecessor.

Valve: Steam Deck 2 not Until 50%+ Performance Uplift

Almost immediately after Valve launched the Steam Deck, fans were already starting to ponder what the next version of the gaming handheld would look like. Now, following the launch of the Valve Steam Machine, Steam Frame, and the new Steam Controller, Valve has confirmed that there will still be a considerable wait for the next-gen gaming handheld. In an interview with IGN, Valve engineer, Pierre-Loup Griffais, commented that the company has "a pretty good idea of what the next version of Steam Deck is going to be," but that current-gen hardware doesn't yet meet Valve's expectations when it comes to battery life or efficiency. This is in keeping with Valve's previous statements regarding a Steam Deck 2—in 2023, Valve designer Lawrence Yang and Griffais himself said that the Steam Deck 2 would not launch for "a few years," again citing a lack of meaningful CPU and GPU performance improvements.

As it turns out, the hang-up for the Steam Deck 2 is battery life, not so much performance, with Griffais saying "We're not interested in getting to a point where it's 20 or 30 or even 50% more performance at the same battery life. We want something a little bit more demarcated than that." As we've seen with recent devices like the OneXPlayer OneXFly Apex, it is possible to achieve desktop RTX 4060-tier performance in a handheld form factor, but even optimistic manufacturer-supplied battery life estimates on those devices are not particularly impressive. On the other hand, Valve's current-gen Steam Deck boasts impressive efficiency. Even if it doesn't deliver high-resolution AAA gaming performance, the Steam Deck manages to deliver halfway decent gaming performance, even at just 15 W TDP.

NVIDIA GeForce RTX 5070 Ti Dips Below MSRP for the First Time

GPU prices are finally coming down. After we saw AMD's Radeon RX 9070 and Radeon RX 9070 XT graphics cards have finally dip below their MSRPs, it seems that NVIDIA cards are now in order. NVIDIA GeForce RTX 5070 Ti slipped below its MSRP this week after Newegg applied a rebate that brought widely available listings to $729.99, under the $749 MSRP tied to this "Blackwell" SKU. The new pricing may not instantly alter the narrative around the 5070 Ti, but rather give this GPU some price breathing room for card that carried persistent price premiums, turning some buyers away.

The SKU in question is MSI Ventus 3X OC BLACK edition, which apparently has not been listed on MSI's official site, meaning that retailers are now the exclusive place to get this GPU. We had a pleasure of reviewing the non-BLACK MSI RTX 5070 Ti Ventus 3X OC GPU, which stood out as a sensible and well-rounded offering. Its factory overclock, strong headroom for further tuning and efficient gaming power draw make it especially attractive to 1440p high refresh gamers, compact-system builders and overclocking enthusiasts. Taking performance, features and price into account we awarded the Ventus 3X OC a "Recommended" badge. It is a very good value for its audience, but it missed our Editor's Choice due to the thermal and noise trade-offs. However, at the time the actual retail price was a bit higher than MSRP at launch. Now, PC builders can get this GPU at prices below MSRP from the Newegg listing here.

AMD Claims AI Fuels Soaring Growth in $60 Billion Server CPU Market

The global data center expansion is not only accelerating GPU deployments, but also CPUs. At AMD's financial analyst day yesterday, the company presented an interesting chart that shows massive growth rate for the regular CPUs amid the AI boom. As a universal base platform for every system, CPUs have been somewhat overlooked, with GPUs taking much of the attention. AMD's internal projection shows that AI inflection will drive an additional $30 billion of revenue for CPU demand exclusively, by 2030. This number is supposed to stack on the already existing regular server CPU demand—sitting at about $30 billion—making the 2030 projections stand at a staggering $60 billion of revenue by 2030, for CPUs alone.

The company's analysis shows that the $19 billion CPU sector from 2022 is rapidly evolving, with AI-driven CPU workloads fueling an 18% annual growth rate that will shape this decade. This is more than just incremental expansion. It shows a shift in enterprise computing priorities and a continued rise in demand for traditional hardware. While the conventional general compute segment is growing at a modest 4% CAGR, the AI-driven portion is advancing rapidly, reaching an inflection point that will allow it to capture the majority of data center CPU value by the end of the decade.
Return to Keyword Browsing
Nov 22nd, 2026 17:28 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts