AMD's 2nm CPU, Nvidia's $500B AI Push, and Google's $10B Tax Break - Silicon Brief #14

April 21, 2025 - Silicon Brief
Hi there, In this edition of Silicon Brief, we're diving into the latest advancements in AI chips and data centers. AMD has unveiled its new 2nm Epyc CPU, codenamed 'Venice', which is set to be produced at TSMC's Arizona facility. This development marks a significant step in the evolution of semiconductor technology, promising enhanced performance and efficiency for data-intensive applications. Meanwhile, Micron is restructuring to meet the growing demand for AI data centers, highlighting the increasing importance of AI-driven infrastructure in today's tech landscape. On the data center front, Nvidia is making headlines with its ambitious plan to build a $500 billion AI infrastructure in the US. This move underscores the company's commitment to expanding its footprint in the AI sector. Additionally, Google's Project Mica has secured $10 billion in tax breaks in Kansas City, further emphasizing the strategic investments being made in AI and data center technologies. Stay tuned as we explore these developments and their implications for the future of technology.

Energy & Sustainability

Mainspring Energy Inc, a Menlo Park-based startup, has raised $258 million to advance its linear generators for powering data centers. The funding was led by General Catalyst, with contributions from Khosla Ventures and Amazon's Climate Pledge Fund. Equinor has created a new business unit, PWR, to consolidate its power generation assets in response to rising demand from AI and data centers. The International Energy Agency has released a special report examining how artificial intelligence affects global energy systems, focusing on energy demand from data centers and AI's role in optimizing energy processes. A report by Seagate Technology stresses the need for sustainable practices in data centers as AI demands grow, highlighting concerns over energy use and infrastructure costs. Fujitsu, Supermicro, and Nidec have joined forces to enhance energy efficiency in data centers. The collaboration will integrate Fujitsu's liquid-cooling software, Supermicro's GPU servers, and Nidec's cooling systems, aiming for improved power usage effectiveness.

Technology

City of Peachtree Corners has integrated NVIDIA L4 Tensor Core GPUs to enhance its smart city infrastructure, focusing on traffic and crowd management improvements.
Fixstars Corporation has announced its role as an AI software technical advisor in a strategic partnership with Optage Inc. and GMI Cloud to develop an advanced AI cloud platform in Japan.
Tachyum has become a member of the UALink consortium to ensure its Prodigy Universal Processor is compatible with new AI accelerator ecosystems, aiming to establish open standards for high-performance computing connections.

Regulation & Policy

AlphaSense has identified significant disruptions in the semiconductor industry due to proposed tariffs on imports from China, Taiwan, and South Korea, affecting tech companies and investor confidence. The U.S. government has mandated that Nvidia obtain a license to export its H20 AI chips to China, citing concerns over their potential use in supercomputers. This change is expected to cost Nvidia $5.5 billion.

Data Centers

Townsend Group has invested in CleanArc Data Centers to support the development of renewable-focused hyperscale data centers, including a major project in Virginia. Canadian telecommunications company Telus plans to deploy 500 Nvidia Hopper GPUs at its Rimouski data center, which will utilize renewable energy and natural cooling to enhance efficiency. BDx Data Centers has achieved certification in the NVIDIA DGX-Ready Data Center program, enabling its Jakarta AI Campus to support high-performance AI infrastructure.

Chips & Hardware

Xpeng will implement its Turing AI chip in semi-autonomous vehicles this quarter, surpassing the current Nvidia chip. AMD unveiled its 6th generation Epyc processor 'Venice', the first CPU using TSMC's 2nm technology, with production set for 2026. Texas Instruments launched new automotive chips, including a high-speed lidar laser driver and mmWave radar sensor, to enhance vehicle safety and autonomy.

Subscribe to Silicon Brief

Weekly coverage of AI hardware developments including chips, GPUs, cloud platforms, and data center technology.