1. Samsung has received approval for its HBM3 memory to be used in NVIDIA's AI GPUs. 2. The approval for Samsung's next-gen HBM3E memory is expected in the next 2-4 months. 3. The HBM market is projected to grow significantly, reaching $71 billion by 2027.

1. NVIDIA CEO Jensen Huang identifies the next wave of AI as 'physical AI', requiring three computer systems: AI, robotics, and Omniverse. 2. Huang mentions the 'three-body problem' in reference to the complexity of integrating these systems. 3. NVIDIA aims to empower organizations with AI capabilities for better customer service and operational efficiency.

1. NVIDIA CEO Jensen Huang predicts that AI assistants will be ubiquitous in every company and job. 2. At SIGGRAPH 2024, NVIDIA showcased the integration of AI in various fields including 3D modeling, physics, materials, and robotics. 3. Huang believes that AI will transform industries by enhancing human capabilities and reducing energy consumption in scientific computing.

1. NVIDIA has reduced the production of high-end RTX 40 series GPUs by up to 50% in preparation for the launch of the RTX 50 series. 2. Supply shortages and price hikes are expected for the RTX 40 series, especially for models like the RTX 4070 and above. 3. AIB manufacturers like ASUS, GIGABYTE, MSI, and ASRock may benefit from the price increases due to the AI server boom.

1. Lenovo launches new YOGA Portal Mini-PC with up to Intel Core i7-14700 and NVIDIA GeForce RTX 4060. 2. The Chinese version features a less powerful CPU but a more powerful GPU compared to the global market version. 3. The system allows for upgrades including CPU, RAM, SSD, and potentially GPU.

1. NVIDIA plans to release a new version of the GeForce RTX 3050 Laptop GPU. 2. The new GPU will feature Ada Lovelace architecture and use a larger chip than the GeForce RTX 4050 Laptop GPU. 3. The AD106M GPU, listed in NVIDIA's latest driver release, confirms the upcoming launch of this new GPU.

1. The Elder Scrolls V: Skyrim is now available on NVIDIA's GeForce NOW cloud gaming service. 2. GeForce NOW Ultimate members can stream Skyrim at up to 4K and 120 frames per second. 3. Seven additional games, including Skyrim, have been added to GeForce NOW this week.

1. NVIDIA's GeForce RTX 5090 is rumored to be 48% faster than the RTX 4090. 2. The TITAN AI variant is reported to be 63% faster than the RTX 4090. 3. Both cards are based on the new GB202 architecture and feature next-gen GDDR7 memory.

1. NVIDIA introduces AI Foundry, a service to help companies build custom AI 'Supermodels' using NVIDIA software and Meta's Llama 3.1 model. 2. The service leverages NVIDIA's DGX Cloud AI platform for advanced hardware and technology access. 3. NVIDIA and Meta's partnership aims to democratize access to generative AI for enterprises.

1. TSMC declined NVIDIA's request for a dedicated packaging line for its GPUs. 2. The request was made during a meeting between NVIDIA CEO Jensen Huang and TSMC executives. 3. TSMC is struggling to meet the high demand for advanced packaging capacity due to the booming AI market.

1. NVIDIA releases GeForce Game Ready 560.70 WHQL driver with day-one support for three new games: Dungeonborne, Flintlock: The Siege of Dawn, and Stormgate. 2. The driver also adds support for 14 new G-SYNC Compatible gaming displays. 3. The driver includes fixes for NVENC, OBS, Unity, and Adobe Premiere.

1. NVIDIA's unreleased GeForce RTX 4070 prototype with 10GB VRAM is revealed; 2. The actual released version features 12GB VRAM with a 20% faster memory bandwidth; 3. The prototype has more CUDA Cores and faster clock speeds than the retail version.

1. NVIDIA plans to release the GeForce RTX 5090D in January 2025, tailored for the Chinese market due to U.S. export restrictions on high-performance GPUs. 2. The RTX 5090D is expected to be a powerhouse for PC gaming but with limited AI capabilities. 3. There is speculation about a global launch of the GeForce RTX 50 Series in January 2025, including the RTX 5090 and RTX 5080.

1. The semiconductor industry is showing signs of recovery, with a 29% growth in revenue compared to Q1-23, excluding Nvidia, the growth falls to under 10%. 2. Nvidia's strong performance is driving overall growth, with significant contributions from data centre and AI sales. 3. Inventory levels and revenue predictions suggest a cautious outlook for the industry, with concerns over the sustainability of current growth trends.

1. Black Myth: Wukong will be released next month with a 'Full Ray Tracing' mode that requires significant GPU power. 2. The game features realistic lighting, shadows, and reflections using ray-traced global illumination. 3. NVIDIA's GeForce RTX 40 Series GPUs are recommended for optimal performance, with the RTX 4070 being the minimum requirement for 1440p and the RTX 4080 for 4K.

1. NVIDIA announces a Black Myth: Wukong GeForce RTX 40 Series Bundle, excluding RTX 4060 GPUs. 2. The bundle includes a copy of the game with eligible RTX 4070 and higher GPUs. 3. The game's advanced ray-tracing visuals require at least an RTX 4070 GPU for optimal performance.

1. NVIDIA's next-gen GeForce RTX 5090 rumored to have a base GPU clock of near 2.9GHz, a significant improvement over the current RTX 4090. 2. The RTX 5090 is expected to feature next-gen GDDR7 memory, providing a 30% performance boost in gaming. 3. The new GPU is anticipated to offer substantial improvements in 4K gaming and ray tracing performance.

1. Northwestrepair discovered a scam involving a relabeled GeForce RTX 3080 Ti GPU inside a supposed GeForce RTX 4090. 2. The fake RTX 4090 had a GA102 GPU, which is pin-compatible with the AD102 GPU used in the RTX 3090 Ti. 3. The scammers polished the die and laser-etched new labels, making it difficult to detect.

1. Northern Data Group will launch Europe's first cloud service using Nvidia's H200 GPUs. 2. The service will be provided by Taiga Cloud in datacenters, powered by renewable energy. 3. The deployment aims to enhance high-performance computing solutions and support AI and ML innovation.

1. Elon Musk's xAI startup is training its next-generation Grok 3 LLM model using 100,000 NVIDIA H100 AI GPUs. 2. Musk agreed with Cohere CEO Aidan Gomez's analogy of models training on each other's data as a 'human centipede effect'. 3. Grok 2 is set to release in August, with significant improvements in purging internet training data.