NVIDIA just released fresh GeForce graphics drivers. Their latest version, 572.83 WHQL, supports the new laptop versions of GeForce RTX 5090, RTX 5080, and RTX 5070 Ti graphics cards. These drivers work perfectly with the new Half-Life 2 RTX Demo, bringing RTX Neural Shaders and DLSS 4 with Multi Frame Generation. They also run great with inZOI, the life simulator game that uses NVIDIA Ace characters.
The update helps Warhammer 40,000: Darktide run better with its DLSS 4 MFG update. NVIDIA expanded DLSS 4 MFG overrides to support 100 different games. You can change these settings for any compatible game through the NVIDIA App. According to the notes, they added 61 new titles to the supported list.
This release stamps out several bugs...
MSI's newest AI servers were on display at booth 909 at GTC 2025 from March 18 to 21. These servers include 4U and 2U platforms built with NVIDIA MGX reference architecture. They handle AI, HPC, and data-heavy tasks with excellent performance and reliability. Enterprise and cloud data centers can benefit from these modular systems, which boost AI workloads and improve computing power.
Danny Hsu runs Enterprise Platform Solutions at MSI. He said companies need AI to stay competitive. MSI created servers that grow with increasing AI demands. They offer flexibility and stay running when needed most. The NVIDIA MGX design helps businesses build lasting systems that work better without extra complications or downtime.
MSI built its new AI...
ASUS joined GTC 2025 as a Diamond sponsor to showcase its latest AI POD with the NVIDIA GB300 NVL72 platform. The company has already received large orders, marking a major milestone. It is also showing its newest AI servers in the Blackwell and HGX lineup, which include several models with different NVIDIA technologies.
ASUS aims to help industries adopt AI by providing complete infrastructure solutions with NVIDIA platforms, helping clients reach markets faster. The ASUS AI POD uses the NVIDIA GB300 NVL72 server platform, giving businesses amazing processing power to handle huge AI challenges easily. Built with NVIDIA Blackwell Ultra, GB300 NVL72 leads AI innovation with better computing, more memory, and faster networking.
It has...
QNAP made their AMIZcloud system better by adding remote switch checking. This helps companies manage networks more easily and cuts down on repair costs. Right away, you can use this with these switches: QSW-M3224-24T, QSW-IM3216-8S8T, QSW-M3216R-8S8T, QSW-M3212R-8S4T, and QSW-M7308R-4X. QNAP plans to add more switch types later.
AMIZcloud works through the internet and lets tech staff control QNAP NAS, switches, and other network devices from just one screen. Companies don't need extra servers to run it, which saves money on hardware care. The system groups things by "Organizations" and "Sites" so users see devices clearly across different departments. This works great for businesses with many locations and lots of equipment because...
Every AI system needs fast memory to work right. The better the memory, the better the AI runs. Micron leads the pack as the only company making both HBM3E and SOCAMM products for data centers. They stand out by creating these special low-power DDR memory chips that help AI servers perform their best.
Micron created its SOCAMM memory, working side by side with NVIDIA. This little powerhouse supports the fancy NVIDIA GB300 Grace Blackwell Ultra Superchip. Their HBM3E 12H 36 GB memory fits perfectly into NVIDIA HGX B300 NVL16 and GB300 NVL72 platforms. The HBM3E 8H 24 GB works great with NVIDIA HGX B200 and GB200 NVL72 systems. These memory chips help NVIDIA Hopper and Blackwell systems run AI tasks much faster.
AI keeps growing thanks...
ASUS just showed off its new AI supercomputer, Ascent GX10. This machine brings incredible computing power right to your desk—think of it as a mini supercomputer that fits where you work. The system helps AI developers, researchers, and data scientists run powerful AI programs without needing huge computer rooms.
Big AI models need lots of memory and processing power to run properly. The Ascent GX10 solves this problem by giving you massive computing strength in a desktop-sized box. You can test and adjust large AI models right at your workspace. This makes AI development much faster and cheaper than using remote systems.
The new ASUS machine delivers up to 1,000 TOPS of AI power - enough for huge AI tasks. With its 128 GB memory...
Solidigm showed off one of the first liquid-cooled business storage drives at the GTC AI Conference today. These new drives help companies remove the fans that storage devices usually need. This makes it possible to build future AI servers that cool everything with liquid instead of air. Companies have long wanted completely liquid-cooled training servers.
Servers need total liquid cooling because it lets designers make smaller machines and spend less money on cooling systems. Old liquid cooling plans for storage drives cooled just one side of each drive and made it impossible to replace drives without shutting down. Solidigm teamed up with NVIDIA to fix these problems. They solved both the hot-swap issue and the single-side cooling...
Phison Electronics just made big changes to aiDAPTIV+, their budget-friendly AI system you can run at home or work. They teamed up with Maingear to create the first-ever AI laptop that lets you train language models directly on your computer. People can test this new setup and sign up at NVIDIA GTC 2025 this March. The system helps users teach large language models with up to 8 billion parameters using their personal information.
The company expanded aiDAPTIV+ to work with NVIDIA Jetson devices for AI tasks at remote locations and robotics projects. They also created aiDAPTIVLink 3.0 software that makes the AI respond faster and understand longer questions for better answers. These improvements help many different people - from college...
Kioxia plans to show people how important fast storage becomes for AI systems at NVIDIA GTC 2025. They want to explain ways flash memory helps companies make the most of their AI investments through better performance and higher efficiency. Anyone can visit them at booth #1811 inside the San Jose McEnry Convention Center between March 17 and 21 to see what they offer.
Visitors checking out the Kioxia booth will find several cool things on display. They recently made a huge 122.88 terabyte NVMe SSD that fits in a standard 2.5-inch case—their first business-grade QLC drive using eighth-generation BiCS FLASH 3D memory with 2 terabit QLC chips. People can watch live demonstrations of their new All-in-Storage ANNS with Product Quantization...
