Yesterday, Qualcomm announced its High Bandwidth Compute (HBC) product, a memory-compute solution hybrid designed to replace traditional High Bandwidth Memory (HBM) and offer greater performance, efficiency, and throughput. HBC utilizes LPDDR memory, stacking them across multiple layers in a 3D vertical space and interconnecting them with through-silicon vias (TSVs). This approach provides better energy efficiency than traditional HBM, which stacks regular DDR in vertical layers, as the low-power DDR (LPDDR) chips consume less power while delivering similar bandwidth and capacity. Qualcomm has successfully stacked multiple LPDDR layers on top of each other, with the base of the HBC being a compute die that performs near-memory computation and offloads some processing from the main processor.

This technique is similar to what HBM4 memory currently does, with the base die being a logic die to better integrate compute solutions underneath for packet tracing and data preparation for entering and exiting HBM. However, Qualcomm aims to achieve higher efficiency and more optimized data movement, with a 6x increase in bandwidth per watt compared to the current HBM specification, likely HBM4. HBC Gen 1 achieved 133 TB/s of bandwidth on the AI250 accelerator card, a 18x increase over the LPDDR5X currently used in the AI200 card from Qualcomm. The company also claims there is a viable roadmap for using HBC across many Qualcomm AI accelerators, with HBC Gen 1 shipping with the AI250 AI accelerator in mid-2027. The next-generation HBC Gen 2 is expected to yield further significant gains in bandwidth.

Colorful today introduced the iGame B850I Mini OC V14, a fairly premium Socket AM5 motherboard in the Mini-ITX form-factor, which aesthetically matches its iGame Mini line of SFF-ready GeForce RTX 50-series graphics cards, including the very capable RTX 5070 Mini OC. The board draws power from a combination of 24-pin ATX and a single 8-pin EPS power connector; and conditions it for the CPU with a 9+2+1 phase VRM solution that uses 60 A DrMOS. The Socket AM5 SoC is wired to two reinforced DDR5 DIMM slots, and the board's sole expansion slot, a reinforced PCI-Express 5.0 x16. Both the M.2-2280 slots on this board—one on the obverse side, and the other reverse—are Gen 5, and wired to the processor in a way that they don't subtract lanes from the PEG slot.

The M.2 slot on the obverse side gets a chunky aluminium heatsink. Besides the two M.2 Gen 5 slots, you get two SATA 6 Gbps ports. USB connectivity includes a 20 Gbps USB 3.2 Gen 2x2 front-panel header, and two additional USB 3.2 Gen 1 ports from a separate front-panel header. The rear panel has one 10 Gbps USB 3.2 Gen 2 type-A, two 5 Gbps USB 3.2 Gen 1, and four USB 2.0. Display outputs include both HDMI and DisplayPort. Network connectivity includes Wi-Fi 7 and Bluetooth 5.4 from a MediaTek MT7925 160 MHz WLAN adapter, and a 5 GbE interface driven by a Realtek RTL8126-CG PHY. The onboard audio solution uses a Realtek ALC1200 HDA codec. The company didn't reveal pricing.

Micron Technology, Inc. (Nasdaq: MU) today announced results for its third quarter of fiscal 2026, which ended May 28, 2026.

Fiscal Q3 2026 highlights

• Revenue of $41.46 billion versus $23.86 billion for the prior quarter and $9.30 billion for the same period last year
• GAAP net income of $28.24 billion, or $24.67 per diluted share
• Non-GAAP net income of $28.86 billion, or $25.11 per diluted share
• Operating cash flow of $25.39 billion versus $11.90 billion for the prior quarter and $4.61 billion for the same period last year

According to the June update of the TOP500 list, which tracks 500 of the world's most powerful supercomputers, China has taken the number one spot with its LineShine supercomputer. This system is capable of running 2.198 ExaFLOPS when measured on HPL, making it the world's first system to exceed 2 ExaFLOPS in dual-precision FP64 data processing for sustained workloads. It has a theoretical peak performance of 2.736 ExaFLOPS, meaning the system can sustain about 80% of its peak performance. This setup is powered by a massive network of traditional CPUs, with no GPUs or other accelerators contributing to the high FP64 figure. While traditional data centers have shifted toward 8-bit FP8 data formats for AI workloads, scientific HPC systems and supercomputers continue to use the traditional FP64 data. The Chinese LineShine is the first-ever system to sustain over 2 ExaFLOPS of compute capability, surpassing U.S.-based El Capitan, Aurora, and Frontier.

The LineShine comprises 13.79 million ARMv9 cores, spread across 20,480 nodes in a non-uniform memory access (NUMA) configuration. Each node contains two LX2 CPUs, with 304 cores and 32 GB of HBM memory. The CPU uses HBM for fast memory access, then offloads to system DDR memory, which has about 256 GB per CPU. The CPU is split across two dies, each divided into four NUMA domains. Each NUMA domain contains 38 cores running at 1.55 GHz and has 4 GB of HBM assigned to it. Four of these NUMA domains make up a die, and each die is assigned 128 GB of DDR memory. This is likely DDR5, but we haven't confirmed those details yet.

TerraMaster, a leading global private cloud storage brand, today announced the global debut of its latest 4-bay flagship NAS, the F4-425 Pro, scheduled for June 23 to coincide with Amazon Prime Day. Powered by TOS 7, the world's first AI-native operating system, the F4-425 Pro redefines the private cloud storage eperience with natural language interaction, localized intelligent computing, and enterprise-grade performance.

To celebrate the launch, the F4-425 Pro will be offered at a limited-time 20% discount during Prime Day, along with several other best-selling NAS and DAS products.

[Editor's note: Our in-depth review of the TerraMaster F4-425 Pro is now live]

Chinese companies like Gloway and KingBank, which provide DDR5 memory kits for consumer platforms, have started integrating domestically produced CXMT DDR5 memory into their RAM kits, effectively ending their reliance on DRAM from Micron, SK hynix, and Samsung after years of external dependence. It has been reported that Gloway and KingBank have begun incorporating CXMT's 24 Gb DDR5 modules, which are being shipped as part of 48 GB RAM kits. With 3 GB per module, this results in eight modules per DDR5 DIMM for a total of 24 GB of capacity. In the regular dual DIMM configuration, Gloway and KingBank are offering consumers 48 GB of capacity.

Gloway has prepared the Longwuyi Yi Special Edition kit, which operates at 6,000 MT/s with latencies of CL36-38-38-80. This kit runs at a voltage of 1.2 V and is specialized for AMD CPUs. Reportedly, it features custom heat spreaders to maintain cooling and control heat dissipation, with 5 W/mK thermal pads included. Meanwhile, KingBank DIMMs use 2 mm heat spreaders and have a custom thermal interface material applied over the PMIC, along with dual-side RGB LED lighting that can be controlled via software.

AMD, through its official documentation portal, has confirmed the arrival of the next-generation Ryzen Threadripper TR6 "Mustang Peak" platform, along with some intriguing details about the upcoming CPU lineup. Based on the "Zen 6" CPU architecture, AMD is incorporating the latest I/O standards for these workstation powerhouses, including PCIe 6.0 and DDR5. As the "Zen 6" CPU cores are manufactured on TSMC's 2 nm semiconductor node, we expect AMD to increase the core count over its previous "Zen 5" Ryzen Threadripper workstation platform and deliver a significant performance boost due to advancements in node and microarchitecture. While specific details like TDP, clock speed, core counts, and segmentation are still lacking, we have learned that AMD is retiring its TR5 socket in favor of the new TR6 socket that will accommodate the platform.

With the introduction of PCIe 6.0, which doubles the bandwidth of PCIe 5.0 per lane configuration, AMD needs to add more pins to handle the increased bandwidth. Additionally, with varying core counts and clock speeds, the CPU socket likely requires more electrical capacity to support the new platform, necessitating an overhaul. If readers recall, the previous TR5 socket used 4,844 pins. For the TR6 update, it is possible that the pin count could exceed that number.

We previously covered the announcement of the GPD Box, a new Panther Lake mini PC that features an MCIO 8i port and dual USB4 v2.0 ports for expanded connectivity with eGPUs and other high-bandwidth applications. Now, the GPD Box has finally launched on Indiegogo with a starting price of HK$11,377 ($1,452.25 converted). The base configuration is powered by an Intel Core Ultra 7 356H with 32 GB LPDDR5-8533 RAM and a 1 TB M.2 2280 NVMe SSD. The next level up upgrades the CPU to the Core Ultra X7 358H for HK$12,020 ($1,534 converted), and the bundles with the Box mini PC and the G2 MCIO eGPU Dock come in at HK$14,394 ($1,837 converted) for the Core Ultra 7 356H version and HK$15,036 ($1,919 converted) for the X7 358H variant. The biggest difference between the Core Ultra X7 358H and the Ultra 7 356H is the significantly more powerful Arc B390 iGPU in the former—importantly, there is no MCIO connection on the Core Ultra X7 358H version of the Box, leaving it with only USB4 for eGPU connections.

Despite its compact chassis, at just 175 × 134 × 39.5 mm, the GPD Box has decent I/O capabilities, with the highlight of the show being the MCIO 8i port that allows for eGPU expansion with minimal bandwidth loss compared to USB4 and Thunderbolt connections. According to GPD, the maximum bidirectional bandwidth of the GPD Box's MCIO port is 512 Gbps, which effectively matches a native PCIe 5.0 x8 cable connection. Notably, while the G2 eGPU dock does include a PSU with an 800 W ATX 3.1 PSU with two GPU power cables, you will need to bring your own GPU if you buy the bundle with the dock. The G2 dock can also be purchased separately via Indiegogo at a starting price of $385. The GPD Box itself also features dual 2.5 GbE RJ45 ports with link aggregation support. For wireless support, there is Bluetooth 5.3 and Wi-Fi 6E, while the physical I/O includes 2× USB4 V2 Type-C ports, 2× USB 3.2 Gen 2 Type-A, and a 3.5 mm audio combo jack on the front. The rear I/O has a DP 2.1 UHBR20 port, 1× HDMI 2.1 FRL ports, 2× RJ45 ports with 2.5 Gbps speeds, 2× USB 3.2 Gen 2 Type-A ports, and the MCIO 8i port.

Goodram expands its portfolio with a new gaming brand - Goodram RIVAL. Its first product is the Goodram RIVAL DDR5 SODIMM series: memory modules designed for gaming laptops and compact mobile workstations built to perform without hardware limitations.

The RIVAL brand was created for gamers, PC enthusiasts, and users who expect stability, speed, and predictable performance under load. DDR5 SODIMM is its mobile-focused lineup - an upgrade for users who want to get more from their laptop without replacing the entire platform.

AMD's upcoming Ryzen 10000 series of desktop processors, codenamed "Olympic Ridge" and based on the "Zen 6" CPU architecture, are rumored to introduce some interesting platform changes. According to the latest leak from X account "Gotou_3rd," AMD's upcoming AM5 platform is likely to replace its integrated graphics processing unit (GPU) with an integrated Neural Processing Unit (NPU), marking a significant shift from the current setup. Since the "Zen 4" based "Raphael" processors, part of the Ryzen 7000 series, AMD has included integrated graphics solutions with its desktop CPUs. This trend continued with the "Zen 5" based "Granite Ridge" Ryzen 9000 series, which also featured an integrated GPU for basic graphics output and boot functionality.

However, with the next-generation Ryzen 10000 series, AMD seems to be reallocating the silicon area for different purposes, as the role of integrated GPUs in desktop systems diminishes. Most PC builders pair a dedicated GPU with their DIY or OEM PC builds, using integrated graphics mainly for debugging when the discrete GPU encounters issues. Since the silicon space in the CPU can be repurposed, AMD might switch to an NPU to enhance local AI capabilities and offload some AI tasks from the GPU to the CPU's integrated NPU. This could mean the platform might become Copilot+ AI PC certified if the NPU achieves over 40 TOPS of compute power.

Since the 13th Gen Core "Raptor Lake" microarchitecture already got its refresh with the 14th Gen Core "Raptor Lake Refresh," the company decided to go with "Raptor Lake Next" as codename for its next line of Socket LGA1700 desktop processors. These chips are born out of necessity, to provide options to PC buyers stuck with the decade-old DDR4 memory standard in wake of the DDR5 memory pricing crisis, given that the current Core Ultra "Arrow Lake" processors lack DDR4 backwards compatibility.

An inside report by JayKihn suggests that Intel could switch to the newer Core Series 3 processor model numbering scheme for its new processors, without the "Ultra" brand extension, denoting the lack of an integrated NPU. Intel could use the opportunity to carve out new combinations of CPU P-core counts, E-core counts, and L3 cache sizes, with the top part in the series expected to be 8P+8E, and not 8P+16E that the "Raptor Lake-S" silicon physically has. These chips could occupy price points in the entry-thru-performance segments, and Intel will ensure they don't tread on the market space of its next-generation Core Ultra "Nova Lake-S"

Apacer has introduced GraTherX, an industrial-grade memory cooling technology aimed at addressing thermal challenges in high-speed DDR5 modules, particularly in fanless and space-constrained systems.

As demand for DDR5 continues to grow alongside AI applications, increasing thermal density and power consumption have raised concerns over system stability. According to the company, testing under high-load conditions shows that DDR5 modules equipped with GraTherX can achieve temperature reductions of up to 23.4°C, compared with the typical 3-5°C improvement seen in conventional cooling solutions. The design also improves heat dissipation across both sides of the module, helping maintain stable operation over extended periods.

Montage Technology today announced the sampling of its Gen 6 DDR5 Registering Clock Driver chip (RCD06) with key customers. Designed for next-generation DDR5 Registered Dual In-line Memory Modules (RDIMMs), the chip is set to accelerate performance and reliability for upcoming server platforms.

The RCD06 supports data transfer rates of up to 9200 MT/s, a 15% increase over the previous generation, addressing the stringent high-bandwidth demands of next-generation server platforms. The chip features a dual-channel independent architecture in which both sub-channels share clock logic while operating independently, enabling separate parity checking without cross-channel interference. The device also integrates Continuous Time Linear Equalization (CTLE) and a low-jitter Phase-Locked Loop (PLL), improving signal integrity while ensuring precise and stable clock distribution.

Cooler Master has been on a roll at Computex 2026, showing off a host of new air cooling and case hardware, but the hardware company has also debuted some interesting water cooling options that it plans to both sell as standalone hardware and in new pre-built workstation PCs. The new cooling line-up consists of both a CPU and GPU water block designed for workstation CPUs and a 360 × 360 mm radiator that Cooler Master calls the 360² Radiator, with accompanying 180 mm fans. The new liquid cooling system is allegedly capable of cooling up to 2,000 W of thermal load. The new radiator and CPU water block come with new Y-shaped hoses, which disperse liquid through the radiator. The example Cooler Master had at Computex had hot water entering two inlets in the center of the radiator and two outlets on opposite ends of the same edge of the radiator. Alternatively, it looks as though the cooling can be routed so that the CPU and GPU inlets and outlets are on opposite sides of the radiator. The radiator was also mounted in a separate chamber behind the motherboard tray, which appears to be basically the only feasible option, given its size.

At Computex, Cooler Master showed off an early prototype of what it is calling the MasterWorkStation: Project MasterFrame 800, a pre-built PC built around its aluminium MasterFrame 800 PC case. The exact specs of the workstation PC are being kept under wraps, but it was clear to see that the PC contained four ASRock AMD Radeon AI Pro R9700 GPUs and what looks to be an ASRock WRX90 WS EVO DDR5 motherboard, which means it will support AMD's latest Threadripper Pro 9000 and 7000 WX-series processors, up to eight DDR5 DIMMs, seven PCIe 5.0 ×16 expansion cards, SlimSAS SFF-8654 SSDs, and 10 Gb networking. ASRock is pitching the new MasterWorkStation workstations as a powerful option for everything from AI training to data processing, creative work, and general compute-heavy workloads. It remains to be seen when these workstations will launch or at what price point.

Gigabyte, the world's leading computer brand, continues to advance motherboard innovation by pushing its performance to a new industry standard. From Ultra Durable, performance-focused Super Overclocking technology to the latest AI-enhanced D5 Bionic Corsa and X3D Turbo Mode 2.0, Gigabyte continues to transform engineering challenges into leading innovations for AI computing and gaming.

With the rise of AI technology, Gigabyte integrates AI-driven tuning, hardware design, and advanced BIOS optimization for a smarter, faster, and more adaptive motherboard platform. At the core of Gigabyte's AI-enhanced innovation is X3D Turbo Mode 2.0, designed to maximize AMD Ryzen X3D processor performance through a dynamic AI overclocking engine and a dedicated onboard hardware chip. The exclusive OC engine is trained on massive datasets and optimized for per-processor tuning, while the onboard chip continuously monitors platform conditions and workload behavior in real time. This intelligent hardware-software integration enables adaptive optimization for AI computing, multitasking, and gaming scenarios.

GIGABYTE, the world's leading computer brand, achieved several milestones in overclocking, including a DDR5-13556 MT/s world record with Corsair VENGEANCE DDR5 memory and 10 global first overclocking places on G.SKILL 12th Annual OC World Record Stage at Computex 2026. GIGABYTE's renowned overclocker team, led by Hicookie, achieved the DDR5 record-breaking performances with the OC-enthusiast Z890 AORUS TACHYON DUO X ICE motherboard. The team was also invited by G.SKILL to the overclocking event and topped the competition with new accomplishments in several categories achieved by the newly launched X870 AORUS INFINITY.

The highly anticipated G.SKILL event drew elite overclockers from across the globe, including Hicookie, Sergmann, Saltycroissant, Madness777, and Exaberries from Team GIGABYTE, who boldly pushed the boundaries of hardware performance under extreme conditions. During the competition, Team GIGABYTE and the X870 AORUS INFINITY motherboard became the highlight on stage by delivering the unmatched performance of the G.SKILL Trident Z5 RGB memory frequency and 10 global first places in CPU frequency and other benchmarks with AMD Ryzen 9 9950X3D2 processors.

Netac showed up to Computex 2026 with DDR5 memory kits, portable and enterprise SSDs and a couple of accessories. The Shadow II RGB DDR5 memory modules uses premium ICs and can achieve speeds up to 7600 MHz while supporting both XMP 3.0 and EXPO profiles. Aluminium heatsinks and dynamic RGB complete the features. Pricing and availability remain to be announced as this is the norm nowadays with the Rampocalypse. Netac ZX20 V is a portable SSD with a compact format (71.5 × 43.5 × 8 mm, 23 g). For shock protection it uses an aluminium case with a reinforced internal frame. The series uses 3D TLC NAND flash and came in 512 GB, 1 TB, 2 TB and 4 TB variants featuring USB 3.2 Gen 2x2 at 20 Gbps connection, up to 2,000 MB/s reads and up to 1,700 MB/s writes.

On the enterprise side Netac showcased two U.3 PCIe 4.0 x4 SSD series. The ENSP401 comes with capacities from 1.92 TB to 30.72 TB with sequential reads up to 7,000 MB/s, writes up to 3,800 MB/s, 850K random read IOPS and 130K random write IOPS. The ENSP403 offers a bit more performance with sequential write speeds up to 6,800 MB/s while keeping the same 7,000 MB/s sequential read. It can reach random reads up to 1.6 million IOPS and writes up to 380K IOPS. Netac is offering the ENSP403 with capacities ranging from 1.6 TB to 25.6 TB. Both enterprise SSD series use 3D TLC NAND and include power loss protection.

At Computex 2026, we've covered a number of well-designed and unique CPU coolers, from the new Cryorig semi-passive cooling case to AIO coolers with swiveling displays on them, and Cooler Master is expanding the list of innovative products at the show with its new hybrid AIO CPU cooler, a new GPU exhaust system, and new active RAM cooling developed in conjunction with. Arguably the most interesting of the lot, and winner of a TechPowerUp "Best of Computex" award, the G11M is a concept cooler that combines air and liquid cooling in an attempt to reach 400 W of cooling capacity. The G11M uses liquid channels, cooling fins, and a downward-firing fan mounted directly on the CPU block to start the heat dissipation process before the warm liquid inside the AIO has even left the CPU block. The rest of the cooler is effectively just a regular 360 mm AIO, so it would theoretically deliver the same performance as a regular AIO with the added benefit of the CPU block cooling and additional airflow over the VRMs and memory.

Team Group brought a large selection of new products to Computex 2026 across its main brand, the T-Force brand targeting PC enthusiasts, and the T-Create brand targeting creative professionals and AI developers. We begin our tour with the new Team Group T-Force Vulcan DDR5 CAMM2 module. Intended to be installed along the plane of the motherboard, CAMM2 modules conserve Z-Height around the CPU socket, and are dual-channel memory modules. The T-Force Vulcan CAMM2 somes with a thick heat spreader with a carbon fiber decal on top. Next up, is the T-Force Delta CAMM2, which is among the first CAMM2 modules we've seen with an ARGB LED lighting setup. Both these CAMM2 modules offer speeds of DDR5-8000. The T-Force DARK Low-Profile RGB are a line of DDR5 UDIMMs with reduced Z-Height, yet featuring ARGB LEDs. These come in mid-range speeds of DDR5-6400.

Next up, is the T-Create Expert AI 4R CUDIMM, a quad-rank memory module with CKD. A single module offers 128 GB of density across four ranks, along with speeds of DDR5-8000 and timings of 64-64-64-128, at 1.1 V. Team Group demoed this module on an MSI MEG Z890 Unify-X motherboard that has a 1 DIMM per channel topology. The new T-Create Master AI DDR5 RDIMM series is targeted at HEDTs and workstations that support ECC RDIMMs. The module shown offers 64 GB of density (per module), with DDR5-7600 speed at CL38. Winding things up on the memory front is the T-Create Expert AI LPCAMM2 for notebooks and mini-PCs that support LPDDR5X.More pictures follow.

Apacer at the 2026 Computex showed off its latest PC memory and storage products under its own brand, and under its ZADAK brand targeting PC enthusiasts and gamers. We begin our tour with its enthusiast-segment PC memory. The company showcased the NOX series of DDR5 memory kits under its main marquee. The new Apacer NOX CUDIMM comes in out-of-the-box frequencies ranging between DDR5-8400 and DDR5-9200, with capacities of 16 GB, 24 GB, and 32 GB per module. The modules are crowned by silicone ARGB diffusers.

The ZADAK PRO RGB DDR5 series of memory are conventional DDR5 DIMMs that come in speeds ranging between DDR5-6000 and DDR5-8000. The ZADAK PRO ROG Certified memory comes in DDR5-6000 with CAS latencies as low as 28. We're not sure if these feature EXPO-ULL. We've also seen the Apacer NOX RGB DDR5 series from last year, which come in DDR5-5200 to DDR5-8000 speeds. Winding things up on the memory front are the Apacer Panther RGB DDR5 series of mainstream memory kits that range between DDR5-5200 and DDR5-6800.

From June 2 to 5, COMPUTEX 2026 in Taipei runs under the theme "AI Together." Longsys showcases its full-stack edge AI storage portfolio under "Edge AI Storage - Integrated Implementation," together with its high-end consumer storage brand Lexar, to optimize local LLM performance and accelerate AI ecosystem collaboration.

Global Launch: AIDIMM & AILPBGA for Edge AI Inference
AIDIMM delivers up to 128 GB capacity, 256-bit bus, and 307.2 GB/s bandwidth in a compact form factor. It resolves memory, thermal, and upgrade pain points of AI agent hosts—a single module reliably and stably runs edge LLMs with over 70 billion parameters. Dynamic voltage scaling (0.9 V-1.05 V) and FDVFS power tuning enable high-performance, low-power mass deployment.

We visited the ZBOX exclusive area of the ZOTAC booth, where we were greeted by the Magnus One Ultra (EU27508OC), the world's most compact prebuilt desktop with a desktop NVIDIA GeForce RTX 5080 graphics card. This thing is about the size of a shoebox, with a total volume of just 11.46 liters, but crams some serious kit. There is a desktop GeForce RTX 5080 graphics card that runs at NVIDIA-reference clock speeds. At its heart is an Intel Core Ultra 7 265 processor (8P+12E, 65 W).

There are two DDR5 SO-DIMM slots capable of CSODIMMs (you add your own memory). There are also two M.2 NVMe slots, one is Gen 5, the other Gen 4 (you also add your own SSDs). Connectivity includes Wi-Fi 7 160 MHz, Bluetooth 5.4, 5 GbE wired LAN, and two Thunderbolt 4 connectors. One of these has DisplayPort passthrough from the graphics card, the other from the processor's iGPU. Powering it all is an 850 W PSU that gives the RTX 5080 a 600 W-capable 12V-2x6 connection.

MSI in its Computex 2026 booth showed us some of their latest mini PCs. The Cubi NUC WCG is designed to be an affordable desktop for client edge deployments. It features a Core 7 360 "Wildcat Lake" processor, and comes in black and white color options. The box is rich in connectivity, featuring up to three display outputs (two HDMI, one USB-C DisplayPort passthrough), two wired Ethernet interfaces (1 GbE + 2.5 GbE), and feature USB-C based PowerLink. You also get one 40 Gbps Thunderbolt 4 port. It uses DDR5 SO-DIMM memory. Next up, is the Cubi NUC AI+ 3MG, a powerful and compact Copilot+ ready mini PC powered by Core Ultra Series 3 "Panther Lake" processors, going all the way up to the top Core Ultra 9 386H. You get Thunderbolt 5, four display interfaces, dual 2.5 GbE, and Wi-Fi 7.

Origin Code today announced that select configurations within its Vortex DDR5 memory lineup now officially support AMD's newly released AMD EXPO Technology: Featuring Ultra Low Latency, placing the company among the earliest memory brands ready for the next evolution of AMD DDR5 memory tuning technology.

Introduced by AMD on June 1, the AMD EXPO ULL standard further refines memory overclocking profiles, enabling improved compatibility, and optimized training behavior for next-generation AMD platforms. Through extensive validation and tuning, Origin Code has optimized select Vortex DDR5 configurations to leverage the capabilities of the new specification.

NVIDIA today announced what is arguably its most important product of the year, one that—if done right—could help it break into the consumer PC processor market. NVIDIA envisions a change in the PC user interface paradigm. For over 40 years, the PC human interface contract involved you interacting with software through their toolbars, dialog boxes, and work areas, to get work done. NVIDIA sees AI agents fundamentally change the way consumers interact with PCs. Imagine you opened up Photoshop, and instead of manually running a complex workflow, simply prompted it your agent, and it did the rest. NVIDIA thinks the processor of the future should be able to not just run your applications, perform its core function as a general purpose CPU, but also run large AI models and agents entirely locally, while also fulfilling consumer-segment needs as an entertainment and gaming powerhouse. Enter, the NVIDIA RTX Spark.