Intel "falls", Taiwan Semiconductor "eats full"?

Recently, the news about Intel considering splitting its business has attracted widespread attention in the industry. Morgan Stanley pointed out that if Intel considers options such as closing some factories, selling assets, and potentially splitting or divesting its foundry and product design teams, it would be slightly positive for Taiwan Semiconductor.

On September 2nd, Morgan Stanley analyst Gokul Hariharan released a report stating that compared to the opportunities brought by AI semiconductors, Intel's outsourcing TAM has shrunk, and it may need to heavily rely on Taiwan Semiconductor's low-power platform to remain competitive in the market.

Capital Reduction or Market Tilt for Taiwan Semiconductor

Previously, Intel has stated that it is cutting capital expenditures, reducing its fiscal year 2024 capital spending from $31-33 billion to $25-27 billion, with an expected year-over-year decrease to $20-23 billion in fiscal year 2025, and adjusting expansion plans for some new factories.

Morgan Stanley stated that if Intel decides to further reduce capital expenditures or halt the production of new fabs, it would directly impact the capacity of its external foundries. Considering the current global shortage of cutting-edge technology, such measures could further tilt the market towards Taiwan Semiconductor, especially as Taiwan Semiconductor expands its business footprint.

Uncertain Future for Splitting Foundry Entity

Morgan Stanley believes that if Intel splits into independent foundry and product design entities, Taiwan Semiconductor may benefit from it.

He mentioned that in order to stay competitive, Intel's product division may have to outsource more core computing products to Taiwan Semiconductor. As for the foundry business, due to its dependence on funding for expensive advanced research and capacity building, Intel's prospects in this business are less clear.

After all, there have been precedents, such as GlobalFoundries, which, despite obtaining significant external funding, eventually had to abandon process development and become a specialized foundry.

Morgan Stanley pointed out:

"We believe that even as an independent entity, Intel's foundry division is unlikely to become a strong competitor to Taiwan Semiconductor in the advanced process field in the next 3-5 years."

Limited Returns for Intel's Merger Choices

The market has raised questions about whether Intel's fabs could merge with another supplier (such as Samsung foundry or GlobalFoundries) to become a strong competitor to Taiwan Semiconductor?

Morgan Stanley believes that such a merger is unlikely to succeed:

"Although Intel's advanced process development has accelerated in the past 3-4 years, the cost, execution speed, and adaptation to market changes (especially the rapid growth of AI accelerators and advanced packaging) are still far below expectations. In addition, this could be very expensive in the coming years, with limited returns."

Data shows that in the second quarter, Intel's foundry operating profit margin was -65%, much lower than Taiwan Semiconductor's 42.5%. And due to weak overall demand for Intel products, revenue in 2025 is projected to be only around $5 billion, significantly lower than Morgan Stanley's forecast of $6-8 billion by the end of 2023In addition, Morgan Stanley believes that artificial intelligence may become a bigger driver for Taiwan Semiconductor than Intel. Its revenue has reached the median in 2024 and may approach 20% in 2025, while the incremental outsourcing opportunities brought by Intel's outsourcing are not as attractive as they were a few years ago.

Bullish on Taiwan Semiconductor

Currently, the market expects Intel to reduce outsourcing after 2026 and instead use its own 18A process for production.

However, Morgan Stanley points out that this shift may impact Taiwan Semiconductor's outsourcing revenue. But considering the strong demand for Taiwan Semiconductor among other strategic clients, and the typically unstable pace of Intel's product launches, Taiwan Semiconductor may not be greatly affected by this.

Furthermore, analysts believe that Intel's key AI PC processor (LunarLake) in 2025 is likely to be entirely dependent on Taiwan Semiconductor's process. With the expected intensification of competition in ARM PC CPU in 2025, Intel may need to heavily rely on Taiwan Semiconductor's low-power platform to maintain competitiveness in the market.

Morgan Stanley states that Taiwan Semiconductor, with its strong process roadmap (N3 and N2) and leading position in packaging technology, as well as its monopoly in AI accelerators and edge AI, is expected to continue to maintain its structural growth momentum. Additionally, with inventory levels bottoming out and cyclical growth drivers in the next 12 months, Taiwan Semiconductor's market outlook remains optimistic.

Currently, Morgan Stanley rates Taiwan Semiconductor as "Overweight" with a target price of NTD 1200