All articles tagged with #nvidia gpus
SpaceX disclosed a $30 billion, multi-year contract to lease computing capacity from Google, including access to tens of thousands of Nvidia GPUs, as it aims to boost AI infrastructure revenue ahead of a record IPO; paired with a similar deal with Anthropic, the arrangements could lift SpaceX’s AI-related revenue and underscore Musk’s push into large-scale AI compute.
OpenAI has signed a $38 billion, seven-year deal with Amazon Web Services to access extensive cloud computing resources, including Nvidia GPUs, to support its AI models like ChatGPT, marking a significant shift after restructuring and reducing reliance on Microsoft.
OpenAI has signed a $38 billion seven-year deal with Amazon Web Services to access hundreds of thousands of Nvidia GPUs for training its AI models, marking a shift as Microsoft reduces its exclusive cloud role with OpenAI and the company expands its cloud partnerships.
Supermicro announced new AI-optimized systems at INNOVATE! EMEA 2025, featuring the latest NVIDIA GPUs, Intel processors, and edge solutions designed for high performance and energy efficiency across data centers and edge environments.
Gearbox has released an optimization guide for Borderlands 4 on PC following reports of poor performance and stuttering, especially on lower-end machines. The guide provides detailed settings for Nvidia GPUs to improve gameplay, but some players still experience issues despite the game's recent launch and patches. The game runs on Unreal Engine 5, which has been associated with performance problems in recent titles.
Researchers have demonstrated GPUHammer, a new RowHammer attack variant targeting NVIDIA GPUs that can cause severe degradation of AI model accuracy by inducing bit flips in GPU memory. The attack poses significant risks to AI infrastructure, especially in shared environments, and can be mitigated by enabling ECC, though at some performance cost. This development highlights ongoing vulnerabilities in GPU memory security and the importance of robust defenses for AI systems.
Elon Musk's AI startup, xAI, plans to expand its Colossus supercomputer by over tenfold, aiming to compete with major players like Google, OpenAI, and Anthropic. The expansion will involve over 1 million GPUs and is already underway in Memphis, with support from companies like Nvidia and Dell. This move highlights the intense competition in the AI industry for computing power, as companies race to secure resources for training large-language models. xAI, valued at $45 billion, is rapidly gaining ground in the AI sector.
Elon Musk's startup xAI plans to build a supercomputer by fall 2025 to power a more advanced version of its Grok chatbot. This "gigafactory of compute" will use tens of thousands of NVIDIA H100 GPUs and cost billions of dollars. The new GPU cluster will be at least four times larger than those used by competitors, aiming to support Grok 3.0, which requires 100,000 GPUs. Grok 1.5, released in April, can process visual information and text, and xAI has started rolling out AI-generated news summaries for premium users.
Dell Technologies forecasts upbeat fiscal 2025 revenue and profit above Wall Street estimates, driven by strong demand for its AI servers equipped with Nvidia's GPUs and a recovering PC market. The company's shares surged over 16% in after-hours trading, while rival Hewlett Packard Enterprise's shares dropped after forecasting quarterly revenue below expectations. Dell expects revenue between $91 billion to $95 billion for the current fiscal year, with a strong AI-optimized server momentum and bullish outlook on the PC market's refresh cycle.
Micron has released its latest roadmap, revealing its upcoming memory technologies, including GDDR7 for next-gen NVIDIA GPUs. The GDDR7 memory modules are expected to offer higher clock speeds of 32 Gbps, providing up to 128 GB/s bandwidth per module. The initial GDDR7 modules are anticipated to be available by mid-2024, coinciding with the expected launch of NVIDIA's next-generation lineup. Micron plans to further upgrade GDDR7 with higher speeds and capacities, potentially introducing 32 Gb solutions. Additionally, Micron's roadmap includes plans for HBM4 memory in 2025 and 2027, as well as advancements in LPCAMM2 and DDR memory.
Imbue, a woman-led AI research lab, has raised $200 million in a Series B funding round led by the Astera Institute, bringing its valuation to over $1 billion. Imbue aims to build AI "agents" that can simulate human decision-making and complete complex tasks. The startup has access to 10,000 Nvidia H100 GPUs and has released an open-source training environment called Avalon. While Imbue is still in the early stages and has no public demos yet, its founders believe AI has the potential to bridge the gap between ideas and execution. The funding will support the development of Imbue's agents and its ambitious goal of revolutionizing generative AI.
Linux 6.4-rc7 has been released with various driver updates, networking changes, and a couple of filesystem fixes. The open-source Nouveau driver for NVIDIA GPUs has been fixed to resolve the serious issue that crept into Linux 6.3. Linux 6.4 stable is expected to be released next week.