As the AI race intensifies, the competition among major cloud service providers (CSPs) to establish robust AI computing capabilities, applications, and ecosystems. Morgan Stanley recently released a research report titled AI Supply Chain: ASIC 2.0 – face-off on 3nm projects, stating that Taiwan Semiconductor Manufacturing Company (TSMC) emerges as a pivotal AI enabler, set to benefit significantly regardless of which technology—GPUs or ASICs—gains dominance.
TSMC – the AI enabler that benefits most amid the AI race
With the competition among major CSPs to build up their own AI computing, applications, and ecosystems, we believe TSMC will be the AI enabler that benefits most, as we indicate in our Global Insight on ASIC. In this report, we indicate that those major CSPs are still trying to develop their own AI ASIC chips, competing with Nvidia's GPU. These efforts not only can help them to build up a second source for their AI chips but also can bring an alternative which is potentially more power-efficient and cost-saving. That said, no matter which type is winning more market share in the future, ASIC or GPU, we believe TSMC can still benefit from production from its leading edge node and advanced packaging.
During its 3Q24 analyst meetings, TSMC suggested that it may double, or more than double, its CoWoS capacity Y/Y again in 2025, after more than doubling it Y/Y in 2024. It is difficult to pinpoint when AI capex might reach an inflection point, but supply chain data points are still positive, in our view. Based on the current supply chain visibility, we raised our bull case forecast to US$235bn in 2025e, which includes around US$200bn of AI GPU production value and US$30bn of merchant ASIC, other GP GPU solutions, and AI CPU. In our bull case, we now expect a 43% CAGR for the cloud AI semi market during 2023-2030e.
TSMC will benefit no matter who wins in the cloud AI market, ASIC or GPU
Many investors believe most of TSMC's AI semi foundry service caters to Nvidia, but we estimate that Nvidia's AI GP GPU will actually contribute just ~70% of TSMC's AI semi revenue in 2025. We believe AI ASICs will become a more significant revenue contributor at nearly 25% of TSMC's AI revenue by 2027e, vs. ~65% from Nvidia GPUs. Our analysis is as follows:
1. AI GPU revenue contribution to TSMC
Last year, we only forecast around 25% of Nvidia revenue contribution to TSMC from AI revenue. However, this year the increasing percentage is mostly due to the strong demand for both Hopper and Blackwell, and the higher-than-our-previous expectation of the price for CoWoS and testing (including the probe pad cost from TSMC). We therefore believe TSMC can generate US$6.8bn revenue from Nvidia's AI GPU in 2024e, accounting 8% of TSMC's overall revenue this year.
2. AI ASIC revenue contribution
For AI ASIC, we expect it to account for 4% of TSMC 2024e total revenue this year (including CoWoS and testing), equating to ~US$3.5bn, which is mostly Google TPU v5 and AWS Trainuium/Inferentia 2. Into 2025, we expect this revenue to grow 70% to around US$6.7bn revenue contribution to TSMC.
3. CoWoS revenue contribution
Advanced packaging (CoWoS and SoIC, excluding InFO), which is heavily used by AI semis, may generate more than US$6.5bn revenue for TSMC in 2024e based on 32kwpm of CoWoS capacity and more than double in 2025.
4. AI sever CPU
For AI server CPU, Nvidia's Grace CPU accounts for the majority of TSMC AI CPU production, which will have a larger contribution in 2025. We expect it to contribute US$8bn/114bn of front-end wafer revenue in 2024 and 2025, respectively. Then, we expect the revenue contribution from AI CPU to double in 2026, accounting for ~1% of total TSMC AI related-revenue in 2026.
In our bull case, we expect the global cloud AI market to reach US$405bn in 2027e and US$498bn in 2028e, which is in line with AMD's expectations during its recent Advancing AI event. We see room for our bull case to play out if increasing numbers of companies acquire a competitive edge and continue investing heavily in AI computing to enhance their productivity; we also consider an early pull-in schedule for AGI applications, such as humanoid robots.
Thus, a 43% CAGR over 2023-30e is possible, in our view, implying a >US$500bn market for cloud AI semis in 2030e. In our bull case scenario, we also see the global semi market size expanding to US$1.2 trillion, within which cloud AI semi would account for ~45% of total spending and the incremental increase is also driven by AI investments.
TSMC: Earnings estimates revision summary
We nudge our EPS estimates up 1% for 2024, 2% for 2025, and 5% for 2026: This mainly reflects increased CoWoS revenue; 2026e has more incremental change. We also fine-tune 2025 wafer price assumptions to match the pricing change for TSMC going into 2025: hikes for 5nm and below, cuts for 7nm and above.
We raise our price target from NT$1,330 to NT$1,388: This is because of our earnings estimate revisions.
We continue to use a residual income model to derive our base case value, which is also our price target. All our key RI model assumptions are unchanged – cost of equity of 9.2% (beta of 1.2, risk-free rate of 2.0%, equity risk premium of 6.0%), intermediate growth rate of 10.0%, and terminal growth rate of 4.0%.
We also raise our bull case value to NT$1,680 (from NT$1,610) and our bear case value to NT$770 (from NT$740).
TSMC stock is trading at 17x our 2025e EPS, at its 10-year average multiple of 17x, which we view as attractive. Our price target implies a multiple of 23.5x our 2025e EPS, higher than the plus-one standard deviation average multiple of 19x. We think our implied target multiple is justified because we continue to believe that AI will outgrow non-AI industry in the coming years, and TSMC is the key enabler of this AI trend including both front-end wafer production and back-end advanced packaging.
Comments