MSI launches scalable AI server solutions with NVIDIA technology
MSI has introduced new AI server solutions using NVIDIA MGX and NVIDIA DGX Station reference architectures designed to support the expanding requirements of enterprise, HPC, and accelerated computing workloads.
The company's new server platforms feature modular and scalable building blocks aimed at addressing increasing AI demands in both enterprise and cloud data centre environments. Danny Hsu, General Manager of Enterprise Platform Solutions at MSI, said, "AI adoption is transforming enterprise data centers as organizations move quickly to integrate advanced AI capabilities. With the explosive growth of generative AI and increasingly diverse workloads, traditional servers can no longer keep pace. MSI's AI solutions, built on the NVIDIA MGX and NVIDIA DGX Station reference architectures, deliver the scalability, flexibility, and performance enterprises need to future-proof their infrastructure and accelerate their AI innovation."
One of the main highlights is a rack solution based on the NVIDIA Enterprise Reference Architecture, comprising a four-node scalable unit constructed on the MSI AI server utilising NVIDIA MGX. Each server in this solution contains eight NVIDIA H200 NVL GPUs, further enhanced by the NVIDIA Spectrum-X networking platform to enable scalable AI workloads. This modular setup provides the capability to expand to a maximum of 32 server systems, meaning up to 256 NVIDIA H200 NVL GPUs can be supported within a single deployment.
MSI states that this architecture is optimised for multi-node AI and hybrid applications and is designed to support complex computational tasks expected in the latest data centre operations. It is built to accommodate a range of use cases, including those leveraging large language models and other demanding AI workloads.
The AI server platforms have been constructed using the NVIDIA MGX modular architecture, establishing a foundation for accelerated computing in AI, HPC, and NVIDIA Omniverse contexts. The MSI 4U AI server provides configuration options using either Intel or AMD CPUs, aimed at large-scale AI projects such as deep learning training and model fine-tuning. The CG480-S5063 platform features dual Intel Xeon 6 processors and eight full-height, full-length dual-width GPU slots that support NVIDIA H200 NVL and NVIDIA RTX PRO 6000 Blackwell Server Edition, with power capacities up to 600W. It offers 32 DDR5 DIMM slots and twenty PCIe 5.0 E1.S NVMe bays for high memory bandwidth and rapid data access, with its modular design supporting both storage needs and scalability.
Another server, the CG290-S3063, is a 2U AI platform also constructed on NVIDIA MGX architecture. It includes a single-socket Intel Xeon 6 processor, 16 DDR5 DIMM slots, and four GPU slots with up to 600W capacity. The CG290-S3063 incorporates PCIe 5.0 expansion, four rear 2.5-inch NVMe bays, and two M.2 NVMe slots to provide support for various AI tasks, from smaller-scale inference to extensive AI training workloads.
MSI's server platforms have been designed for deployment within enterprise-grade AI environments, offering support for the NVIDIA Enterprise AI Factory validated design. This structure provides enterprises with guidance in developing, deploying, and managing AI—including agentic AI and physical AI—as well as high-performance computing tasks on the NVIDIA Blackwell platform using their own infrastructure. The validated design combines accelerated computing, networking, storage, and software components for faster deployment and risk mitigation in AI factory roll-outs.
MSI is also presenting the AI Station CT60-S8060, a workstation built on the NVIDIA DGX Station reference, with components designed to enable data centre-grade AI performance from a desktop environment. This includes the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip and up to 784GB of coherent memory, intended to boost large-scale training and inference. The solution is targeted at teams requiring a high-performance desktop AI development environment and integrates the NVIDIA AI Enterprise software stack for system capability management.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Techday NZ
5 days ago
- Techday NZ
Vultr launches early access to AMD Instinct MI355X GPU for AI
Vultr has announced the availability of the AMD Instinct MI355X GPU as part of its cloud infrastructure services. As one of the first cloud providers to integrate the new AMD Instinct MI355X GPU, Vultr is now taking pre-orders for early access, with global availability scheduled for the third quarter of the year. The GPU forms part of AMD's latest focus on high-capacity computational demands, catering to artificial intelligence (AI) workloads as well as enterprise-scale applications. Product features The AMD Instinct MI355X GPU is based on AMD's 4th Generation CDNA architecture. According to Vultr, this GPU features 288 GB of HBM3E memory, delivers up to 8 TB/s of memory bandwidth, and supports expanded datatypes such as FP6 and FP4. These improvements are designed to address complex tasks ranging from AI training and inference to scientific simulations within high-performance computing (HPC) environments. For customers operating within higher-density data environments, the Instinct MI355X supports direct liquid cooling (DLC). This enhancement offers increased thermal efficiency, which is intended to unlock greater computing performance per rack and facilitate advanced, scalable cooling strategies. The GPU is also supported by the latest version of AMD's ROCm software, which further optimises tasks related to AI inference, training, and compatibility with various frameworks. This results in improved throughput and reduced latency for critical operations. AMD and Vultr partnership Vultr's portfolio already includes other AMD offerings, such as the AMD EPYC 9004 Series and EPYC 7003 Series central processing units (CPUs), as well as previous GPU models like the Instinct MI325X and MI300X. Customers using the MI355X in combination with AMD EPYC 4005 Series CPUs will benefit from a fully supported computing stack across both processing and acceleration functions, streamlining high-powered workloads from end to end. Negin Oliver, Corporate Vice President of Business Development, Data Centre GPU Business at AMD, stated: "AMD is the trusted AI solutions provider of choice, enabling customers to tackle the most ambitious AI initiatives, from building large-scale AI cloud deployments to accelerating AI-powered scientific discovery. AMD Instinct MI350 series GPUs paired with AMD ROCm software provide the performance, flexibility, and security needed to deliver tailored AI solutions that meet the diverse demands of the modern AI landscape." The collaboration builds on Vultr's efforts to support a range of AMD solutions tailored for enterprise, HPC, and AI sectors, reinforcing the company's capacity to cater to evolving customer workloads. Cloud market implications J.J. Kardwell, Chief Executive Officer of Vultr, highlighted the alignment of the new GPU with market requirements. Kardwell commented: "AMD MI355X GPUs are designed to meet the diverse and complex demands of today's AI workloads, delivering exceptional value and flexibility. As AI development continues to accelerate, the scalability, security, and efficiency these GPUs deliver are more essential than ever. We are proud to be among the first cloud providers worldwide to offer AMD MI355X GPUs, empowering our customers with next-generation AI infrastructure." AMD is recognised as a member of the Vultr Cloud Alliance, which supports a collaborative ecosystem of technology providers focused on offering integrated cloud computing solutions. The introduction of the MI355X GPU follows a period of upgrades across AMD's GPU lineup, including a greater emphasis on catering to both inferencing and enterprise-scale workloads. Vultr's offering is aimed at organisations seeking advanced compute resources for AI-driven applications and scientific tasks requiring significant computational capacity. Vultr's global network reportedly serves hundreds of thousands of customers across 185 countries, supplying services in cloud compute, GPU, bare metal infrastructure and cloud storage. The addition of AMD's latest GPU to its infrastructure underlines Vultr's commitment to providing a variety of options for businesses and developers pursuing AI and HPC advancements.
Techday NZ
13-06-2025
- Techday NZ
Oracle unveils AMD-powered zettascale AI cluster for OCI cloud
Oracle has announced it will be one of the first hyperscale cloud providers to offer artificial intelligence (AI) supercomputing powered by AMD's Instinct MI355X GPUs on Oracle Cloud Infrastructure (OCI). The forthcoming zettascale AI cluster is designed to scale up to 131,072 MI355X GPUs, specifically architected to support high-performance, production-grade AI training, inference, and new agentic workloads. The cluster is expected to offer over double the price-performance compared to the previous generation of hardware. Expanded AI capabilities The new announcement highlights several key hardware and performance enhancements. The MI355X-powered cluster provides 2.8 times higher throughput for AI workloads. Each GPU features 288 GB of high-bandwidth memory (HBM3) and eight terabytes per second (TB/s) of memory bandwidth, allowing for the execution of larger models entirely in memory and boosting both inference and training speeds. The GPUs also support the FP4 compute standard, a four-bit floating point format that enables more efficient and high-speed inference for large language and generative AI models. The cluster's infrastructure includes dense, liquid-cooled racks, each housing 64 GPUs and consuming up to 125 kilowatts per rack to maximise performance density for demanding AI workloads. This marks the first deployment of AMD's Pollara AI NICs to enhance RDMA networking, offering next-generation high-performance and low-latency connectivity. Mahesh Thiagarajan, Executive Vice President, Oracle Cloud Infrastructure, said: "To support customers that are running the most demanding AI workloads in the cloud, we are dedicated to providing the broadest AI infrastructure offerings. AMD Instinct GPUs, paired with OCI's performance, advanced networking, flexibility, security, and scale, will help our customers meet their inference and training needs for AI workloads and new agentic applications." The zettascale OCI Supercluster with AMD Instinct MI355X GPUs delivers a high-throughput, ultra-low latency RDMA cluster network architecture for up to 131,072 MI355X GPUs. AMD claims the MI355X provides almost three times the compute power and a 50 percent increase in high-bandwidth memory over its predecessor. Performance and flexibility Forrest Norrod, Executive Vice President and General Manager, Data Center Solutions Business Group, AMD, commented on the partnership, stating: "AMD and Oracle have a shared history of providing customers with open solutions to accommodate high performance, efficiency, and greater system design flexibility. The latest generation of AMD Instinct GPUs and Pollara NICs on OCI will help support new use cases in inference, fine-tuning, and training, offering more choice to customers as AI adoption grows." The Oracle platform aims to support customers running the largest language models and diverse AI workloads. OCI users leveraging the MI355X-powered shapes can expect significant performance increases—up to 2.8 times greater throughput—resulting in faster results, lower latency, and the capability to run larger models. AMD's Instinct MI355X provides customers with substantial memory and bandwidth enhancements, which are designed to enable both fast training and efficient inference for demanding AI applications. The new support for the FP4 format allows for cost-effective deployment of modern AI models, enhancing speed and reducing hardware requirements. The dense, liquid-cooled infrastructure supports 64 GPUs per rack, each operating at up to 1,400 watts, and is engineered to optimise training times and throughput while reducing latency. A powerful head node, equipped with an AMD Turin high-frequency CPU and up to 3 TB of system memory, is included to help users maximise GPU performance via efficient job orchestration and data processing. Open-source and network advances AMD emphasises broad compatibility and customer flexibility through the inclusion of its open-source ROCm stack. This allows customers to use flexible architectures and reuse existing code without vendor lock-in, with ROCm encompassing popular programming models, tools, compilers, libraries, and runtimes for AI and high-performance computing development on AMD hardware. Network infrastructure for the new supercluster will feature AMD's Pollara AI NICs that provide advanced RDMA over Converged Ethernet (RoCE) features, programmable congestion control, and support for open standards from the Ultra Ethernet Consortium to facilitate low-latency, high-performance connectivity among large numbers of GPUs. The new Oracle-AMD collaboration is expected to provide organisations with enhanced capacity to run complex AI models, speed up inference times, and scale up production-grade AI workloads economically and efficiently.
Techday NZ
12-06-2025
- Techday NZ
Fake booking sites push malware as HP warns of click fatigue
HP Wolf Security has reported an increase in cyberattacks targeting people booking holidays, with attackers using fake websites to distribute malicious software. The company's latest Threat Insights Report highlights a series of campaigns in which users visiting spoofed travel booking websites are presented with a deceptive cookie banner, prompting them to click "Accept" to access the content. This action inadvertently downloads a malicious JavaScript file, resulting in an XWorm infection that allows attackers full control over the victim's device. Spoofed booking sites The report describes how these counterfeit websites closely imitate including branding and blurred content that appears legitimate at first glance. When users click to accept the cookies, a malicious process begins in the background. "Since the introduction of privacy regulations such as GDPR, cookie prompts have become so normalized that most users have fallen into a habit of 'click-first, think later.' By mimicking the look and feel of a booking site at a time when holiday-goers are rushing to make travel plans, attackers don't need advanced techniques - just a well-timed prompt and the user's instinct to click," said Patrick Schläpfer, Principal Threat Researcher in the HP Security Lab. The first signs of this campaign were detected in the first quarter of 2025, coinciding with the busy summer holiday booking season. The campaign remains active, with threat actors continuing to register new domains imitating booking services to target users during the peak period for travel arrangements. Threat techniques The report also covers a variety of other malware delivery methods identified through HP Wolf Security's research. One such technique involves the use of Windows Library files to disguise malware as seemingly harmless PDFs, placed in familiar local folders such as "Documents" or "Downloads." Victims may see a Windows Explorer pop-up displaying what appears to be a standard file, but clicking this shortcut initiates a malware download. Another observed tactic uses malicious PowerPoint files. When opened in full-screen mode, the PowerPoint deck appears to replicate a normal folder window. If users attempt to close or escape the presentation, they trigger the download of a compressed archive containing a VBScript and an executable file, which connects to GitHub to download additional malware. The report notes that MSI (Microsoft Installer) files are now frequently leveraged for malware delivery. Much of this activity has been linked to ChromeLoader campaigns, with MSI installers distributed through deceptive software sites and malicious advertising. These installers often use valid and recently generated code-signing certificates, which help them bypass Windows security warnings and appear legitimate to prospective victims. Exploiting click fatigue According to the report, attackers across all these campaigns are taking advantage of so-called "click fatigue" and routine user behaviours to bypass security measures. The normalisation of prompts such as cookie banners and other pop-ups has led users to respond reflexively, opening new avenues for cybercriminals to deceive even cautious individuals. Dr. Ian Pratt, Global Head of Security for Personal Systems at HP, commented, "Users are growing desensitized to pop-ups and permission requests, making it easier for attackers to slip through. Often, it's not sophisticated techniques, but moments of routine that catch users out. The more exposed those interactions are, the greater the risk. Isolating high-risk moments, like clicking on untrusted content, helps businesses reduce their attack surface without needing to predict every attack." Active campaign and user impact The report states that HP Wolf Security customers have encountered over 50 billion email attachments, web pages, and downloaded files with no reported breaches, thanks to the product's use of virtualised containers that allow malware to detonate safely without impacting user devices. The data used in the report was collected from millions of endpoints running HP Wolf Security between January and March 2025, and includes findings from an independent investigation by the HP Threat Research Team. The research offers insights into the most recent techniques criminals are using to evade traditional detection tools and compromise PCs. The threat campaigns identified in the report remain active, especially those focusing on intercepting holiday bookings through spoofed travel sites. The findings underline the importance of continued vigilance among users, particularly during periods of heightened activity such as the busy summer travel season.
