➀ AMD announces a multi-year agreement to supply its Instinct™ MI450 AI accelerators for OpenAI's next-gen AI infrastructure;

➁ OpenAI secures an option to acquire a 10% stake in AMD under specific terms;

➂ The deal strengthens AMD's position in the AI computing market, marking a strategic expansion in its AI capabilities.

➀ AMD and OpenAI announce a multibillion-dollar partnership for AI data centers, with AMD supplying 6 gigawatts of MI450 chips starting 2026;

➁ OpenAI secures warrants for 160M AMD shares (10% stake) at $0.01 per share upon meeting deployment milestones, potentially valuing the deal at $60B+;

➂ The deal contrasts with NVIDIA's Intel partnership, as AMD and OpenAI commit to near-term regulatory filings amid an AI infrastructure arms race.

➀ Business leaders like Goldman Sachs CEO David Solomon and Amazon founder Jeff Bezos debated whether the AI investment boom is a bubble, with Solomon cautioning potential unrealized returns and Bezos acknowledging short-term risks but emphasizing long-term societal benefits.

➁ Hyperscaler AI capital expenditures are projected to reach $490 billion by 2026 and $684 billion by 2029, while AI companies have accumulated $268 billion in debt over 2024-2025, raising concerns about financial over-engineering akin to historical bubbles like the 2007 subprime crisis.

➂ Nvidia’s $4.6 trillion market cap and circular financing practices (e.g., investing in firms that buy its chips) highlight structural risks, though Bezos argued that despite potential turbulence, AI-driven innovations will ultimately deliver transformative societal gains.

➀ Key ASIC Berhad signs RM1.11 million contract with a Middle Eastern navigation firm to co-develop an AI-driven RF-integrated navigation chip;

➁ The chip targets ultra-low power consumption and advanced navigation accuracy for applications in automotive and IoT sectors;

➂ Collaboration aims to strengthen Key ASIC's presence in AI-optimized semiconductor solutions for global markets.

➀ The US-China semiconductor competition has reportedly led to the resurgence of Alibaba founder Jack Ma, as Chinese authorities showcased Alibaba's in-house AI chip as a domestic alternative to NVIDIA's H20 GPU during Premier Li Qiang's visit to a Qinghai data center;

➁ The Chinese government urged domestic customers to prioritize local chips over NVIDIA's H20, citing security concerns, which halted H20 production;

➂ Jack Ma re-emerged publicly after five years of low visibility following the halted Ant Financial IPO, with his viral bar meeting and the slogan 'Make Alibaba Great Again' signaling his renewed influence.

➀ China is constructing a $37 billion centralized AI compute network, converting 760-acre farmland in Wuhu into data hubs for Huawei and telecom giants to bridge the gap with US AI dominance;

➁ The initiative combines UB-Mesh networking, redistribution of idle GPU resources, and subsidies for domestic AI chips amid US export restrictions;

➂ Strategic data clusters near major cities aim to boost inference speeds, while remote centers focus on LLM training, reflecting China's dual urban-rural AI scaling strategy.

➀ TDK introduces its InvenSense SmartMotion solution for smart glasses, addressing motion tracking and image stabilization challenges while optimizing power efficiency;

➁ The ICM-45685 sensor enables high-accuracy head orientation tracking, posture recognition, and supports AI/AR applications;

➂ Integrated with TDK’s sensor ecosystem, it enhances augmented reality, fitness eyewear, and AI-driven smart glasses with ultra-low power consumption.

➀ Ambient Scientific unveils an AI-native processor delivering 100x efficiency gains over traditional MCUs through in-memory analog compute architecture;

➁ Features 10 AI cores, <100 µW always-on mode, 2MB SRAM, and sensor fusion optimization for voice/vision tasks;

➂ Targets battery-powered edge devices with 2026 production plans, enabling cloud-free AI inference in wearables and IoT.

➀ China's Cyberspace Administration instructs domestic tech companies to avoid purchasing Nvidia's H20 GPUs and RTX Pro 6000D AI chips, restricted under U.S. export rules;

➁ Despite this, China still seeks Nvidia's high-end chips, but its officials are angered by U.S. Commerce Secretary Howard Lutnick's remarks about selling only lower-tier chips to China;

➂ Nvidia CEO Jensen Huang expressed disappointment in U.S.-China tensions and emphasized the strategic importance of maintaining U.S. chip technology as the global standard.

➀ China's Cyberspace Administration (CAC) banned major tech firms like ByteDance and Alibaba from purchasing Nvidia’s AI chips, including the newly released RTX Pro 6000D and previously restricted H20.

➁ Beijing asserts that domestic AI chips from Huawei and Cambricon now rival Nvidia’s China-specific products, prompting companies like Tencent to accelerate local infrastructure development.

➂ Nvidia CEO Jensen Huang expressed disappointment but acknowledged geopolitical complexities, while China’s move aligns with efforts to reduce reliance on U.S. tech amid ongoing trade tensions.

➀ NVIDIA reportedly becomes TSMC's first A16 node customer for its Feynman AI GPUs, breaking Apple's decade-long tradition of adopting TSMC's leading-edge processes with iPhones;

➁ A16, TSMC's 1.6nm-class process, integrates GAA transistors and backside power delivery, offering 8-10% speed gain and improved power efficiency, tailored for AI/HPC applications;

➂ NVIDIA's strategic move to skip N2 for A16 could secure a full-node advantage over AMD and Intel, positioning Feynman GPUs as pioneering AI chips with cutting-edge architecture by 2028.

➀ Rubin CPX die shot reveals dedicated ROP units and display engines, which are redundant for AI workloads;

➁ Speculation arises that it could serve as the foundation for RTX 6090 with 28,672+ CUDA cores and GDDR7 memory;

➂ Planned 2026 release aligns with potential CES 2027 consumer GPU launch, suggesting dual AI/gaming architecture strategy.

➀ Tech leaders like Jensen Huang (NVIDIA) and Larry Ellison (Oracle) saw their wealth skyrocket due to massive AI investments, with Huang's net worth peaking at $243B in 2025.

➁ NVIDIA dominates the AI GPU market, powering Meta, OpenAI, and xAI, while Oracle's $300B OpenAI deal propelled Ellison to become the world's richest man.

➂ Meta's shift to AI superintelligence and Musk's xAI ambitions, including plans for 50M GPUs, highlight the intensifying race for AI dominance despite financial and technical hurdles.

➀ NVIDIA's Blackwell Ultra GB300 NVL72 achieves 45% higher inference throughput than GB200 in MLPerf's DeepSeek R1 test, leveraging upgraded tensor cores and software optimizations;

➀ The system uses NVFP4 quantization to shrink model sizes and NVLink 1.8TBps interconnect for sharding Llama 3.1 405B across 72 GPUs;

➂ NVIDIA positions GB300 as a cost-efficient solution for 'AI factories,' with shipments beginning this month to drive tokenized data center workflows.

➀ Edge AI applications face limitations in performance, memory, and security for deployments in industrial and medical environments;

➁ Axelera AI's Metis M.2 Max accelerator delivers PCIe-level AI performance in a compact M.2 form factor, doubling LLM processing speeds with 16GB memory and 6.5W power efficiency;

➂ Features include Root of Trust security, thermal optimization for constrained spaces, and Voyager SDK for simplified AI model deployment across vision and language applications.

➀ ASML invests €1.3B in French AI startup Mistral AI, becoming its largest shareholder through a €1.7B funding round;

➁ Mistral AI focuses on compact, task-optimized AI models with rapid iteration and permissive licensing, positioning it as a European rival to OpenAI and Meta;

➂ The partnership aims to enhance ASML's AI-driven semiconductor manufacturing processes, including computational lithography and yield optimization.

➀ Ayar Labs与Alchip合作，结合共封装光学（CPO）技术与先进封装工艺，为超大规模数据中心提供高性能、高能效的AI加速器和平台；

➁ 合作中整合了Ayar Labs的光学I/O技术、Alchip的先进封装能力及台积电的COUPE™和SoIC®等封装技术，旨在解决数据传输瓶颈；

➂ 双方计划通过可扩展的解决方案推动下一代AI基础设施应用，进一步提升数据吞吐量和能源效率。

➀ Researchers developed a silicon photonic chip using light for AI computations, reducing energy consumption by up to 100 times;

➁ The chip integrates ultrathin Fresnel lenses for optical convolution, achieving 98% accuracy in tasks like handwritten digit recognition;

➂ Supports wavelength multiplexing for parallel processing, offering potential for sustainable AI scaling and industry adoption.

➀ Fujitsu developed a generative AI reconstruction technology using 1-bit quantization and AI distillation, reducing memory usage by 94% and achieving 3x faster inference while retaining 89% accuracy.

➁ The method enables large AI models like Takane LLM to run on low-end GPUs and edge devices (e.g., smartphones, industrial machines), improving data security and energy efficiency compared to conventional approaches like GPTQ.

➂ The brain-inspired technology allows task-specific specialization, with trials planned for late 2025 and quantized models already released via Hugging Face. Applications show 11x speed gains in sales predictions and 10% accuracy improvements in image recognition.

➀ Intel received $5.7 billion from the U.S. government in exchange for a 10% stake to prevent the sale of its chip fab business;

➁ Struggling against TSMC and Samsung, Intel's foundry division faced a $13 billion loss in 2024, with potential plans to abandon 14A process development;

➂ The deal includes provisions to discourage majority stake sales, while SoftBank invests $2 billion, signaling confidence in government-backed Intel's stability.