Tag with "TSMC"

Google is reportedly planning to deploy up to 15 million of its own TPU v9 AI chips by 2028, potentially surpassing Nvidia in server AI accelerator shipments. This strategic shift from renting GPUs to in-house chip development aims to cut costs and reduce supply chain dependence. The move could reshape the AI hardware landscape, with Intel and Samsung foundries likely to benefit from Google's multi-supplier strategy.

Meta is teaming up with Broadcom and TSMC to produce its own AI data center chip, Iris, which will start mass production in September 2026. The chip aims to optimize content ranking and recommendations on Facebook and Instagram, cutting costs and reducing reliance on third-party GPUs. Meta also plans to double its computing capacity by 2027.

Taiwan Semiconductor Manufacturing Company (TSMC) reported better-than-expected Q3 2025 results with 30% revenue growth to NT$989.92 billion ($32.47B), driven by strong AI chip demand from clients like Nvidia and Apple. The performance offset weaker consumer electronics sales, showcasing TSMC's dominance in high-performance computing.

Tesla's Dojo 2 chip, set for mass production, boasts a 10x performance boost over its predecessor and rivals NVIDIA's Blackwell B200. Elon Musk teased an even more powerful Dojo 3 chip for next year, signaling Tesla's push for AI and autonomous driving independence.

SoftBank founder Masayoshi Son is partnering with TSMC to establish a $1 trillion industrial park in Arizona focused on AI and robotics. The project aims to revitalize U.S. manufacturing and create a smart manufacturing ecosystem, potentially attracting global tech giants and investors.

Taiwan Semiconductor Manufacturing Company (TSMC) has reported a record quarterly revenue of NT$868.5 billion, driven by a surge in demand for AI-related hardware. The positive financial results indicate a robust outlook for 2025, despite some market challenges and uncertainties surrounding the tech industry and geopolitical factors.

**Summary** 1. OpenAI and Broadcom are working on a custom inference chip to reduce operating costs. 2. The collaboration aims to enhance AI efficiency through hardware-software optimization. 3. Broadcom’s stock got a slight bump after the news broke, with AI silicon sales expected to hit $12 billion. 4. OpenAI envisions massive AI-powered data centers using these custom chips. 5. Although OpenAI won’t likely sell hardware, they may deploy these chips at the network edge.