tsmc

The Verge 报道 AI 数据中心热潮正在加剧亚利桑那州的水资源压力：TSMC 每天消耗近 500 万加仑水后，更大的超大规模数据中心项目引发居民自发的两党联合抵制，争论焦点是芯片与数据中心带来的经济利益是否值得以沙漠中最珍贵的水资源为代价。

Microsoft 于 2026 年 10 月 12 日发布下一代 AI 推理芯片 Maia 300，采用 TSMC N2P 计算芯片与 8 组 HBM4，每兆瓦可多产出 40% tokens，2027 年投产并扩展至多吉瓦级 Azure 部署。

Elon Musk comments on the ambition of the Terafab project, which aims to produce around 1 TW of AI compute annually — roughly 50× current global output — arguing that scale alone becomes a qualitative shift.

TSMC presents A14 NanoFlex Pro platform at IEDM with record SRAM density and PPA improvements over N2 technology, on track for volume production in 2028.

Microsoft reportedly plans to unveil its Maia 300 AI chip in September, aiming to close the gap with Nvidia and other custom silicon rivals, backed by a large TSMC order and potential adoption by Anthropic.

Microsoft reportedly plans to unveil its Maia 300 custom AI chip this fall, claiming over 30% more tokens per dollar than existing silicon, and is in talks with TSMC for 300,000 chips by 2027 to reduce reliance on NVIDIA.

Etched announces breakthroughs in low voltage inference and cluster scale memory for their AI inference hardware, with first racks shipping this summer and $1B in customer contracts.

TSMC CFO says the company is accelerating its Arizona factory build-out with an additional $100 billion investment to meet surging AI-driven demand, raising total U.S. investment to $265 billion.

The article examines the growing compute shortage in AI, highlighting simultaneous bottlenecks across GPUs, memory, TSMC capacity, and power infrastructure, and how this scarcity is reshaping corporate strategy and market leadership.

Meta's custom AI chips (MTIA) will begin production in September, aiming to reduce GPU costs. The chips are designed with Broadcom and manufactured by TSMC, part of Meta's strategy to secure compute capacity while still purchasing from Nvidia and AMD.

Google is in talks with Samsung to manufacture a component of its next-generation AI chip, codenamed Icefish, using 2-nanometer technology, while the main part will be made by TSMC. The chip aims to offer an alternative to Nvidia's GPUs and is expected to enter mass production as soon as 2028.

TSMC CEO C.C. Wei warned that the company is struggling to meet surging AI-driven chip demand, stating it could take a 'very long time' to fulfill customer needs even with its US factory expansion, including a $165 billion investment in new American plants.

Taiwan's manufacturing leaders, including TSMC and Foxconn, are leveraging NVIDIA's accelerated computing and AI technologies to build and optimize AI infrastructure for the upcoming Vera Rubin platform, enhancing efficiency and productivity across the supply chain.

TSMC's senior VP says energy efficiency is now the primary constraint in AI chip design, surpassing raw computing power. The shift is driving changes in transistor density, advanced packaging, and chip stacking to reduce power consumption.

Jensen Huang responds to Huawei's Tao's Law, saying it is a breakthrough for Huawei but TSMC has been deeply involved in the related technology for 10 years, and explicitly stated that Huawei is a competitor.

Taiwan's stock market overtook India to become the world's fifth largest, fueled by TSMC's 46% surge and the AI chip boom, adding nearly $2.5 trillion in value.

Tech fund giant Gavin Baker analyzes the AI bubble from a unique perspective, arguing that current AI infrastructure is backed by cash flow and GPUs are operating at full capacity, unlike the 2000 internet bubble, and that TSMC's capacity constraints are key to preventing the bubble from collapsing.