All articles tagged with #gpu
Meta expands its multiyear partnership with NVIDIA to build hyperscale AI data centers, deploying NVIDIA Grace and Vera CPUs, millions of Blackwell and Rubin GPUs, Spectrum-X networking, and NVIDIA Confidential Computing to power training and inference across on‑prem and cloud deployments, including privacy-preserving AI for WhatsApp.
Nvidia (NVDA) trades around $191 after a roughly 7.6% drop over the last three months, contending with margin pressure from rising high-bandwidth memory costs and competition from AMD and hyperscalers, even as AI-driven demand and data-center expansion support a longer‑term rally. NVDA posted 62% revenue growth to $57.01B in the October quarter with net income of $31.91B; analysts expect fiscal 2026 revenue above $212B and AI spending to rise about 30% in 2026–2027. TipRanks’ consensus remains Strong Buy with an average target of $262.79 (about 37.5% upside) driven by AI data-center demand and a China rebound for GPU orders, despite valuation concerns.
Nvidia shares rose after reports it increased GPU/VRAM prices for graphics-card makers; the price hikes are unconfirmed by Nvidia and may not affect MSRP yet, with potential consumer changes in 2026. The RAMpocalypse-driven demand for RAM and AI data centers supports pricing discussions. NVDA was up about 1.5% intraday, with a 12‑month gain around 21%. Analysts rate Nvidia as a Strong Buy with an average target of about $263.44, implying roughly 45% upside.
Mesa 26.0 introduces a dedicated transfer-only queue support in the RADV Vulkan driver using the SDMA engine, benefiting resource streaming and memory management on GFX9 and newer AMD GPUs, with activation via an environment variable, expected to improve Vulkan app performance.
Nvidia announced at CES 2026 that it will accelerate Siemens' electronic design automation (EDA) software using its GPUs to speed up chip design processes and create digital twins for testing chips and systems before manufacturing.
MSI unveiled its most powerful GPU to date, the GeForce RTX 5090 Lightning Z, featuring a 40-phase VRM, up to 3,742 MHz overclocked speed, and advanced cooling with a full-copper cold plate and 360 mm radiator, designed for extreme overclocking with a limited production run of 1,300 units.
MSI has unveiled the GeForce RTX 5090 Lightning Z at CES 2026, featuring a liquid-cooled system, a large front display, reinforced PCB, and high power delivery capabilities up to 2,500 W for extreme overclocking, with benchmarks showing record-high clock speeds of 3,742 MHz.
The article explores the significance of ASML's $400 million EUV lithography machine, essential for GPU manufacturing and AI development, highlighting its complex technology, geopolitical implications, and the inevitable progression towards more advanced and expensive machines, reflecting on the fleeting nature of technological dominance.
Nvidia made a $20 billion acquisition of AI inferencing startup Groq to strengthen its position in the growing AI inferencing market, which could help it maintain its leadership in AI technology and continue its explosive growth.
Shares of AMD and Nvidia declined amid rumors of a GPU price increase in 2026, impacting investor sentiment in the tech stock market.
China's Lisuan G100 GPUs have begun shipping, initially targeting professional workloads, with plans to enter the retail market by Q1 2026. The GPUs feature advanced architecture, support for Windows on ARM, and aim to compete with NVIDIA and AMD, especially in the Chinese market.
Samsung is developing its own proprietary GPU architecture for future Exynos processors, potentially starting with the Exynos 2800 in 2027, aiming to improve AI functions and software optimization, and possibly extending this technology to other devices beyond smartphones.
The article explains how AI works by breaking down the process into steps involving energy generation, data center infrastructure, hardware like GPUs, and software, illustrating that AI's 'magic' relies on a complex stack of energy, hardware, and software working together to produce outputs like images or text.
Samsung is developing the Exynos 2800, scheduled for launch in 2027, which may feature the company's first in-house GPU, expanding its application beyond smartphones to areas like autonomous vehicles and robotics. The company is investing heavily in its semiconductor division, recruiting top talent and advancing its manufacturing processes, aiming to create a unified ecosystem with its custom silicon.
AMD may benefit from a potential deal with Alibaba to supply GPUs in China, which could significantly boost revenue, but regulatory approval remains uncertain. Despite geopolitical and regulatory challenges, AMD's strong recent stock performance and consistent revenue growth make it an attractive investment option for 2026.