#GPU

Nvidia is expanding its NVLink Fusion technology to seamlessly integrate with Arm-based Neoverse CPUs, enabling custom infrastructure solutions for hyperscalers. This move allows tighter integration between Arm CPUs and Nvidia GPUs, crucial for AI workloads requiring high-speed data transfer. Companies like Microsoft, Amazon, and Google are increasingly using Arm CPUs in their cloud infrastructure. This partnership highlights the increasing importance of Arm technology in AI server infrastructure and signifies Nvidia’s commitment to open innovation and diverse partnerships.

With tech giants investing heavily in AI data centers, depreciation of AI GPUs is a crucial accounting concern. Unlike traditional servers, the lifespan of these rapidly evolving components is uncertain, impacting profitability and investment decisions. While some, like CoreWeave, see long-term value, short seller Michael Burry suggests companies may be inflating useful life for earnings. Factors like new chip releases and wear-and-tear could accelerate depreciation. Companies are adopting varied strategies, and auditors are scrutinizing depreciation claims closely.

Samsung is investing in a 50,000 Nvidia GPU cluster, dubbed the “AI Megafactory,” to enhance chip manufacturing for mobile devices and robotics. This project signifies the escalating demand for AI acceleration and strengthens Nvidia’s position in the AI revolution. In addition to being a customer, Samsung will leverage Nvidia’s Omniverse and optimize chipmaking lithography. Samsung will also collaborate with Nvidia on HBM4 memory technology for future AI chips. These partnerships support Korea’s ambition to become a global AI leader.

CoreWeave (CRWV) expanded its partnership with OpenAI, securing a $6.5 billion deal. This brings their total contracted revenue from OpenAI to $22.5 billion. CoreWeave provides specialized compute infrastructure for AI and machine learning, utilizing high-density Nvidia GPUs. CEO Michael Intrator emphasizes CoreWeave’s ability to power demanding AI workloads. The deal highlights the escalating demand for computational resources in AI and CoreWeave’s position as a key infrastructure provider, alongside players like Microsoft. This signals both growth for CoreWeave and increasing concentration of resources in the AI sector.

Nebius Group shares surged in premarket trading after announcing a $19.4 billion agreement with Microsoft. Nebius will provide cloud computing power for AI workloads, generating an expected $17.4 billion in revenue through 2031. The deal highlights the increasing demand for specialized AI infrastructure. Nebius, which spun off from Yandex in 2023, specializes in GPUs for AI model training and boasts investors like Nvidia and Accel. The partnership also boosted shares of competitor CoreWeave, reflecting broader optimism in the AI infrastructure market.

NVIDIA’s GTC 2026, a key event for GPU technology and data center innovation, will be held in San Jose, California, from March 16-19. The conference is expected to showcase the debut of the Rubin GPU and Vera CPU, focusing on their performance and capabilities. GTC 2025 unveiled future GPU hardware plans, including Blackwell Ultra and the Vera Rubin architecture, along with software initiatives like NVIDIA Dynamo and Isaac GR00T N1. The industry anticipates insights into how these advancements will shape NVIDIA’s competitive edge.

Baidu announced significant advancements at the 2025 Smart Economy Forum. Baidu Smart Cloud partnered with 65% of China’s central SOEs to boost the LLM industry. They launched the Qianfan Huijin Financial LLM and introduced industry-specific intelligent agents. Baidu also deployed “Ten Thousand Card” and “Thirty Thousand Card” clusters on the Baige GPU platform, designed for efficient LLM training and inference and compatible with various models.