➀ Elon Musk's xAI is acquiring an overseas power plant to ship to the U.S. for a data center housing 1 million AI GPUs, requiring up to 2 GW of power (equivalent to 1.9 million homes).

➁ Current xAI Colossus supercomputer uses 200,000 NVIDIA H100 GPUs and 300 MW of power, relying on gas turbines and Tesla Megapacks for energy stability.

➂ The AI industry's energy demands are escalating rapidly, with companies prioritizing dedicated power infrastructure over grid reliance, despite environmental and logistical challenges.

➀ Researchers developed a self-powered temperature sensor using a flexible ionic elastomer combining thermoplastic polyurethane (TPU) and ionic liquids, eliminating the need for external power;

➁ The sensor leverages triboelectric effects from temperature-induced microstructural changes, generating voltage signals that correlate with heat levels, achieving rapid and reversible responses;

➂ Potential applications include wearable electronics for real-time thermal monitoring, with future plans to optimize performance and integrate with AI systems for advanced signal analysis.

➀ Arteris won the 'AI Engineering Innovation Award' at the 2025 AI Breakthrough Awards for its system IP technology;

➁ The company was recognized for accelerating AI chip design and SoC creation to meet growing compute demands;

➂ This award highlights Arteris' leadership in enabling AI hardware advancements through interconnect solutions.

➀ The 75th ECTC highlighted three key packaging themes: Co-Packaged Optics (CPO) integration, Hybrid Bonding (HB) advancements, and thermal management for Backside Power Delivery (BPD).;

➁ CPO discussions emphasized materials innovation (e.g., glass substrates) to address thermal and optical challenges in AI chiplets, while HB sessions focused on scaling density, reducing defects, and integrating metrology for memory/3D IC applications;

➂ BPD thermal solutions explored interlayer cooling, multi-level modeling, and novel structures like microchannels to tackle heat dissipation in 3D architectures.

➀ The BMFTR-funded SPINNING project demonstrated a diamond spin-photon quantum computer with 12 qubits, showcasing superior coherence times (over 10 ms) and lower error rates (<0.5%) compared to superconducting IBM models (Eagle, Heron);

➀ Photonic coupling over 20 meters enabled scalable, distributed quantum registers with high fidelity (over 0.9), offering flexible integration with classical computing systems;

➂ The project involved collaboration among 28 entities, advancing diamond material processing, qubit fabrication, and AI-driven quantum applications, funded with €16.1 million by the German government.

➀ AWS's custom Trainium and Graviton chips gain momentum in AI infrastructure, challenging Nvidia's dominance;

➁ Increased adoption driven by cost efficiency and performance optimization tailored for cloud workloads;

➂ AWS reinforces its silicon strategy with enhanced software stacks and developer ecosystems to attract enterprise clients

➀ PCI-SIG released PCIe 7.0 specifications, delivering 128 GT/s raw bit rate and up to 512GB/s bidirectional bandwidth in x16 slots using PAM4 signaling and flit-based encoding;

➁ A new optical PCIe interconnect solution (PCIe 6.4/7.0) was introduced, enabling extended reach and multiplexing for applications like NVMe storage pools and CXL memory shelves in hyperscale AI data centers;

➂ PCIe Gen8 development is already underway, with adoption expected to accelerate in high-performance computing and AI-driven infrastructure.

➀ SiMa.ai and Enclustra co-developed a System-on-Module (SoM) using MLSoC Modalix 50 TOPS chip for edge AI applications;

➁ The solution reduces development costs and time, supports Gen AI and multi-modal AI in robotics/drones with Python/PyTorch compatibility;

➂ Features 50 TOPS performance, multiple I/O options, compact form factor, and industrial-grade thermal management.

➀ Tesla developed a groundbreaking 'Stress' tool to detect defective cores in its wafer-scale Dojo processors, critical for preventing silent data corruption that could invalidate weeks of AI training;

➁ Each Dojo 'Training Tile' contains 8,850 RISC-V cores and leverages TSMC's InFO_SoW packaging, with Stress enabling real-time monitoring across millions of cores without downtime;

➂ The method not only identifies faulty cores but also exposed rare design flaws, positioning Tesla alongside Google and Meta in hardware reliability while shaping future wafer-scale chip adoption.

➀ AMD acquires Untether AI's entire engineering team in a non-traditional talent acquisition, leading to the discontinuation of Untether's AI inference products;

➁ The move enhances AMD's capabilities in AI compilers, SoC design, and energy-efficient inference solutions to challenge Nvidia;

➂ Industry analysts suggest this signals a strategic shift toward AI inference optimization as GPU-driven training demand plateaus.

➀ ROHM launches first standalone AI MCUs with on-device learning and inference capabilities, eliminating cloud dependency;

➀ Utilizes proprietary Solist-AI for real-time anomaly detection in industrial devices, reducing downtime and maintenance costs;

➂ Features 1,000x faster AI processing and low power consumption, supported by developer tools for seamless integration.

➀ NVIDIA is developing a Blackwell-based B30 AI chip for China with multi-GPU scaling, potentially using NVLink or PCIe 6.0 ConnectX-8 SuperNICs;

➁ The U.S. export ban on H20 chips impacted NVIDIA and AMD, with AMD facing $800 million losses from MI308 ban;

➂ Jensen Huang warned that U.S. export controls risk driving Chinese rivals like Huawei to innovate and set future global standards.

➀ AMD launches affordable Strix Halo variants including 8-core Ryzen AI Max 385, targeting mainstream laptops;

➁ Geekbench scores show lower performance than flagship Max+ 395, though variables like power settings may affect results;

➂ HP ZBook Ultra G1a and Framework Desktop adopt the new APU, priced from $2,599 with shipments starting June 2025.

➀ NVIDIA develops 800V HVDC architecture to support 1 MW server racks by 2027, addressing existing 54V DC system limitations;

➀ Current 54V DC systems struggle beyond 200kW, requiring impractical copper quantities for gigawatt data centers;

➂ New 800V HVDC integrates with 13.8kV AC sources, reduces copper use by 45%, and eliminates inefficient AC-DC conversions.

➀ U.S. semiconductor sanctions aimed at curbing China's growth may inadvertently boost its domestic industry, fostering a resilient ecosystem;

➀ Taiwanese IC substrate firms face divergent outcomes: Unimicron struggles with sanctions, while Zhen Ding thrives via a 'China for China' strategy;

➂ Nvidia CEO Jensen Huang criticizes U.S. export bans as ineffective, noting Chinese competitors gain ground due to restricted access to advanced chips.

➀ Ed proposes a £20 million UK government prize for developing reconfigurable AI chip prototypes to address AI progress bottlenecks;

➁ The UK leverages Arm’s low-power design expertise, university photonics, packaging technologies, and phase-change materials research;

➂ HMG plans funding for a collaborative research hub, imec access, and manufacturing partnerships, with Ed positioning himself in profit-generating oversight roles.

➀ Cadence launched Cerebrus AI Studio, the first agentic AI-driven multi-block SoC design platform, accelerating chip development by up to 5X;

➁ The platform uses autonomous AI agents for hierarchical design optimization, multi-user collaboration, and advanced analytics to enhance power-performance-area (PPA) targets;

➂ Early adopters like Samsung and STMicroelectronics reported 8-11% PPA improvements and 4X productivity gains, addressing semiconductor complexity and engineering shortages.

➀ Intel unveiled its Arc Pro B50 and B60 GPUs at Computex 2025, targeting the workstation and AI market with 16GB (B50) and 24GB (B60) memory capacities, emphasizing cost efficiency and AI inference capabilities;

➁ The B50 features a 70W bus-powered design, while the B60 supports up to 200W, with a focus on Linux-based virtualization (SR-IOV) and scalable configurations like dual-GPU systems for AI workloads;

➂ Intel also showcased its Panther Lake 18A processors, aiming to challenge TSMC's process leadership, with GPUs expected to launch in Q3 2025 to address the underserved low-end AI inference and server market.

➀ The U.S. Commerce Department prohibits global use of Huawei's Ascend AI chips, citing violations of export controls due to alleged use of U.S. technology;

➁ Nvidia views Huawei as a formidable competitor but highlights challenges in replacing its dominance, as Huawei faces sanctions limiting access to advanced chipmaking tools;

➂ The crackdown underscores U.S. efforts to maintain AI leadership, even as Chinese firms like Biren and Moore Threads emerge as potential long-term challengers.

➀ The U.S. Department of Commerce's BIS banned the use of Huawei’s Ascend 910B/C/D AI chips, claiming they utilize U.S. technology, with violators facing severe penalties;

➁ Huawei’s Ascend 910C-based servers reportedly outperform NVIDIA’s most powerful GPUs, but production shortages limit their availability even for Huawei itself;

➂ BIS revoked the AI Diffusion Rule and plans new regulations, potentially easing restrictions for U.S. companies exporting AI chips to China.