➀ JPR reports 73.6 million GPUs shipped in Q3 2024; ➁ GPU shipments are expected to decline at a -1.9% CAGR from 2025 to 2028; ➂ AMD and Intel released new CPUs in Q3.

➀ NVIDIA's Rubin AI GPU architecture, originally set for 2026, is now expected six months earlier; ➁ Utilizing TSMC's 3nm process and next-gen HBM4 memory; ➂ Follows the Blackwell architecture, with NVIDIA collaborating with Taiwanese partners on R100-powered AI servers.

➀ Indiana Jones and the Great Circle requires a GeForce RTX 4090 for 4K 60FPS ray tracing; ➁ The game's PC requirements have been announced, with high-end hardware needed for full ray tracing mode; ➂ The game is set to launch on December 9.

➀ NVIDIA is developing the Thor chip for future EVs, partnering with Chinese EV companies like BYD and SAIC Motors. ➁ The Thor platform, with five versions, aims for a full transition by late 2025 or early 2026. ➂ NVIDIA is moving into the automotive market with its new Drive Thor chip.

➀ Elon Musk's xAI startup has ordered NVIDIA GB200 AI servers for $1.08 billion, seeking priority access. ➁ The servers are to be delivered by January 2025 and are expected to enhance xAI's Colossus supercomputer capabilities. ➂ The order is a significant upgrade for xAI's AI computing power.

➀ AMD's Radeon RX 8800 XT, the top RDNA 4 GPU, is set for mass production in December with 25% less power consumption and 45% better ray tracing performance than the 7900 XTX; ➁ NVIDIA's upcoming GeForce RTX 5070 is expected to surpass the RDNA 4 stack at CES 2025; ➂ The RX 8800 XT is the last RDNA 4 card before UDNA, with performance and power consumption rumors suggesting significant improvements.

➀ The rise of RISC-V cores and the challenges of certification; ➁ The role of Breker Verification Systems in the certification process; ➂ The complexity of certifying RISC-V ISA implementations and the efforts of RISC-V International.

➀ NVIDIA's GeForce NOW Black Friday Deal offers 50% off the first three months for new Ultimate and Performance memberships; ➁ Ultimate users get access to RTX 4080 GPUs, streaming up to 4K at 120 FPS; ➂ Both plans have a 100-hour monthly limit, with options to continue on a free tier or pay extra.

➀ NVIDIA's GeForce RTX 5090 and RTX 5090D share the same GPU configuration and clock speeds; ➁ The RTX 5090D was developed to bypass export restrictions, similar to the RTX 4090D; ➂ Both models are set to launch in January 2025 with similar pricing.

➀ NVIDIA is set to use Samsung's new GDDR7 memory chips for its upcoming GeForce RTX 50 series GPUs; ➁ The RTX 50 series will feature both 28Gbps and 32Gbps GDDR7 memory; ➂ Samsung's GDDR7 memory chips will be used by NVIDIA's GeForce RTX 50 series GPUs in early 2025.

➀ The GeForce RTX 5070 features fewer CUDA Cores than the RTX 4070 SUPER and 40% fewer than the RTX 5070 Ti; ➁ NVIDIA's upcoming mid-range RTX 50 Series card might not offer a significant improvement over the GeForce RTX 4070; ➂ The performance of the RTX 5070 is speculative and will depend on clock speeds and memory bandwidth.

➀ NVIDIA is set to unveil the GeForce RTX 5090 and RTX 5080 at CES 2025, with a launch expected shortly after; ➁ The RTX 5090 may be priced at $1999 with 32GB of GDDR7 memory and over 20,000 CUDA cores; ➂ The RTX 5090 is expected to offer 4K 240FPS+ gaming performance.

➀ AI servers are expected to consume over 1000 kW of power due to NVIDIA's upcoming Rubin Ultra AI GPU and HBM4 memory; ➁ Current AI servers use 130-250 kW, but future models like Blackwell Ultra and Rubin AI servers will require 250-900 kW, with up to 576 GPUs per rack by 2026-2027; ➂ NVIDIA's future AI servers with Rubin Ultra AI GPUs will use 1000 kW or more by 2028-2029+.

➀ NVIDIA is set to launch four GeForce RTX 50 series GPUs in Q1 2025; ➁ The GPUs will feature the new Blackwell architecture and next-generation GDDR7 memory; ➂ The release schedule aligns with previous rumors, with the RTX 5090 and RTX 5080 expected in January.

➀ NVIDIA is reportedly developing the GeForce RTX 5070 Ti with 8960 CUDA cores and 70 Streaming Multiprocessors; ➁ The GPU is expected to be the GB203 'Blackwell'; ➂ The card is rumored to have a 300W TDP.

➀ The NVIDIA GeForce RTX 5090 will use the GB202 'Blackwell' GPU, 22% larger than the AD102 'Ada Lovelace' GPU in the RTX 4090; ➁ The GB202 GPU has a die size of 744mm², making it 22% larger than the AD102 GPU in the RTX 4090; ➂ The RTX 5090 is expected to have 20,000+ CUDA cores, 32GB of next-gen GDDR7 memory, and up to 2.0TB/sec memory bandwidth.

➀ Comino has revealed liquid-cooled workstations and servers with Nvidia RTX 5090 graphics cards, indicating that the Blackwell flagship GPU might be close. ➁ Comino is offering priority access to potential buyers, with pre-orders available. ➃ The servers are designed for high performance and can operate in harsh environments up to 40°C.

➀ Amazon has started early Black Friday deals with discounts on a range of PC hardware like GPUs, CPUs, PSUs, and motherboards; ➁ The article lists specific deals available on Amazon; ➂ The deals are expected to change as Black Friday approaches.

➀ NVIDIA reported a Q3 revenue of $35.1 billion with a gross margin of 74.6%, much of which comes from its Data Center business; ➁ NVIDIA's GeForce RTX Gaming and AI PC sector saw a 15% revenue growth year-over-year, but a shortage is expected in the current quarter; ➂ NVIDIA CFO Colette Kress indicated a shift in focus to the next-gen Blackwell-powered GeForce RTX 50 Series, with the launch expected in early 2025.

➀ WSTS报告2024年第三季度半导体市场增长1660亿美元，同比增长10.7%；➁ 英伟达因AI GPU的强劲表现成为最大半导体公司；➂ 2024年第四季度展望显示数据中心市场将推动Nvidia、美光和AMD的收入增长；➃ 2025年展望显示AI将推动服务器增长，但增速将放缓。