Apple is starting to focus on AI chips, Morgan Stanley believes TSMC will benefit the most

Wallstreetcn
2024.03.26 03:23
portai
I'm PortAI, I can summarize articles.

Morgan Stanley raised TSMC's target price to NTD 860, 10% higher than the current stock price, expecting TSMC to outperform the entire semiconductor market in the long term

Morgan Stanley is full of expectations for Apple's Edge Artificial Intelligence (Edge AI) program. In a report released last week, it is predicted that this plan will reignite iPhone sales growth and help Apple catch up in the increasingly fierce AI competition.

Meanwhile, Morgan Stanley's analyst team, led by Charlie Chan, believes that the biggest external beneficiary of the Edge AI program may be TSMC.

In a report released last week, the team stated that due to AI training considerations, Apple's demand for its self-developed M2 Ultra chip will continue to increase. This may further increase Apple's demand for TSMC's InFO-LSI packaging technology and offset the decrease in iPhone processor orders.

Morgan Stanley has further raised TSMC's target price to 860 New Taiwan Dollars and emphasized its importance in the global AI supply chain, expecting TSMC to outperform the entire semiconductor market in the long term.

As of the time of writing, TSMC's stock price is 787 New Taiwan Dollars, with nearly 10% upside potential from Morgan Stanley's target price.

Apple's demand for AI training chips is expected to drive TSMC's revenue growth.

According to Morgan Stanley's recent research on the semiconductor supply chain, the demand for TSMC's InFO-LSI packaging technology is significantly increasing, mainly for Apple's M2 Ultra.

M2 Ultra is used in Apple's high-end Mac computers for AI training. Morgan Stanley points out that as demand for Apple's M2 Ultra chip increases, TSMC may benefit from this.

Starting from the second half of 2023, multiple reports suggest that M2 Ultra may be used for training large language models (LLMs) due to its large memory capacity (192GB unified memory and up to 800GB/s bandwidth), compared to Nvidia's H100 (with 80GB of HBM3 memory).

Recently, Apple made a significant breakthrough in LLM training, detailing its achievements in optimizing multimodal LLM training.

Morgan Stanley expects that the significant increase in demand for M2 Ultra will bring TSMC low single-digit revenue growth, which is expected to offset the potential decrease in 3nm wafer orders due to weak demand for Apple's iPhones

Will TSMC raise wafer prices next year?

Given the increasing demand for high-performance chips in the cloud AI field, Morgan Stanley believes that the possibility of TSMC raising wafer prices in 2025 has increased.

According to a Morgan Stanley report, based on supply chain checks, TSMC is currently in discussions with its major customers and plans to raise wafer prices by 10% in 2025.

Although we believe that the final price increase will be close to 5%, the trend of this price hike has become more apparent.

At the same time, as Apple accounts for 20% to 25% of TSMC's revenue, Morgan Stanley has slightly raised its revenue expectations for TSMC in 2025 and 2026 by 1% to 2%.

From 2024 to 2026, TSMC's earnings per share (EPS) are expected to increase from 32.34 New Taiwan Dollars to 55.84 New Taiwan Dollars.

Based on TSMC's 2025 EPS, Morgan Stanley believes that TSMC's current P/E ratio of 16 times is not high, making the stock price very attractive. Morgan Stanley expects that TSMC's stock price will be revalued to 18 times the expected EPS for 2025.

Therefore, we are raising TSMC's target price to the industry's highest of 860 New Taiwan Dollars and once again emphasize that TSMC is our top pick in the covered sector.

Morgan Stanley believes that TSMC will stand out in the AI semiconductor industry in the long term.

Whether the growth driver comes from GPUs or ASICs, cloud AI chips or Edge AI chips, it is very likely that they will all rely on TSMC's leading manufacturing technology.

Therefore, compared to other stocks in the global semiconductor supply chain, TSMC's growth prospects in the field of artificial intelligence appear more promising