Executives at the chip manufacturer are still optimistic about the revenue potential of AI, as Nvidia and its partners say new GPUs have a lead time of up to 52 weeks.
Despite a 7.5% decline in November revenue, Taiwan Semiconductor (TSMC) is confident chip demand will soon pick up as demand for AI increases and work on the next-generation iPhone gets underway.
TSMC reported monthly revenue of $6.6 billion (NT$206 billion), down 7.5% from the month prior.
Revenue for the first 11 months of 2023 decreased by 4.1% compared to the previous year and came in at approximately $66.33 billion (NT$1.99 trillion).
TSMC faces potential margin pressures due to its current reliance on Apple for 3nm clients, local media in Taiwan report, but expects growth in the second half of the year with new clients such as MediaTek and Intel. The company said it is on track to mass-produce advanced 2nm technology by 2025.
The chipmaker projects December quarter’s sales between $18.8 billion and $19.6 billion, with CEO CC Wei expecting a low point in the coming quarter.
Economic and trade challenges in China are likely to weigh heavily on TSMC’s growth.
The company however received positive news in the US on Thursday wherein TSMC and Arizona labor unions reached a deal to resolve labor disputes at the company’s upcoming chip factory in Phoenix.
TSMC is investing $40 billion in the facility and has agreed to partner with unions for workforce training and safety transparency, as it aims to start production in the first half of 2025, Bloomberg reported.
AI as a revenue driver
At a recent Supply Chain Management Forum in Taiwan, Wei discussed the challenges of high inflation and the growth opportunities in AI, emphasizing the increasing demand for semiconductors and their role in healthcare, remote work, and reducing energy consumption, local media reported.
Virtually all of TSMC’s semiconductor clients, including Nvidia and AMD are banking on AI to drive demand for their chips.
While Nvidia has been in the spotlight for its AI-ready GPUs, AMD has also made AI a pillar of its growth for 2024 as it launches its MI300 series to compete with Nvidia’s flagship AI processors.
AMD estimates a $45 billion market for its AI data center processors this year and anticipates over $2 billion in revenue from the sales of its new AI chips in 2024.
Meta, OpenAI, and Microsoft said they plan to adopt AMD’s new AI chip, intensifying the competition in the space.
Demand for Nvidia’s AI chips is so high that market research firms forecast a 52-week lead time for its flagship H100 AI GPU. Dell recently told investors that AI servers from the company have a 39-week lead time.
Meanwhile, analyst firm IDC is projecting 2023 shipments of PCs to decline 13.7%, with consumer growth down 16%, year over year to 252 million units. This would be the sharpest decline in nearly 40 years.
