【US Stock Decoding】Intel's Valuation Hurdle

Since the beginning of this year, riding on the wave of AI, GPU suppliers Nvidia (NVDA.US) and Advanced Micro Devices (AMD.US) have gained a lot of popularity.

Despite recent rumors about Nvidia facing antitrust investigations leading to a significant drop in its stock price, its stock price has risen by 114.50% since the beginning of the year. In contrast, AMD's stock performance has been less than satisfactory, with a cumulative decline of 4.44%.

Compared to AMD's mediocre performance, Intel's performance can be described as "extremely poor". Intel (INTC.US) has seen its market value halved since the beginning of the year, with a cumulative decline in stock price of 60.82%.

For the 12 months ending June 2024, Intel's revenue reached $55.1 billion, higher than AMD's $23.3 billion, and more than half of Nvidia's total revenue of $96.3 billion for the 12 months ending July 2024. However, Intel's market value is only $83.1 billion, equivalent to 36.45% of AMD's total market value of $228 billion, and 3.19% of Nvidia's market value of $2.61 trillion.

Why does Intel have such a significant discount in valuation compared to its peers?

Currently, the general market view is that the lag in technological advancement, limitations in development prospects, and the failure to timely adapt its business ecosystem to industry changes are the main factors leading to Intel's poor performance.

Finet believes that the underperformance of semiconductor wafer foundries may be another key factor leading to the significant decline in Intel's stock price this year.

Intel's Wafer Foundry Development Plan

Under the guidance of returning CEO Gelsinger, Intel released its IDM 2.0 strategy in March 2021, aiming to revitalize its manufacturing capabilities and expand its market business into the semiconductor industry. This includes investments in production facilities to enable the company to produce its own designed chips and supply products to external customers.

Intel aims to become the world's second-largest wafer foundry after TSMC (TSM.US) and hopes to become a major chip supplier for European and American chip technology giants, aiming to reduce dependence on the Asian supply chain.

Previously, the main reason for the shortage of Nvidia's AI chip H100 was its dependence on the supply chain. If Nvidia had its own foundry, it could be self-sufficient, or if Nvidia's customers such as Tesla (TSLA.US), OpenAI, etc., had their own chip manufacturing capabilities, they would naturally not need to pay high costs for scarce AI chips.

Intel's IDM 2.0 strategy allows it to achieve chip self-sufficiency and have the capacity to serve other customers - such as creating chips for Microsoft (MSFT.US), making it a very suitable strategic layout Ideals are beautiful, but reality is harsh.

In the first half of 2024, Intel's wafer fabrication business operated at a loss of $5.304 billion, a 25.42% increase from the same period last year.

Recent reports indicate that Intel will test Broadcom (AVGO.US) silicon wafers produced through its more efficient 18A production process—used as components for semiconductor basics. Broadcom found that the process is not yet ready for mass production, although Broadcom later clarified that no conclusion has been reached. The capital market is less hopeful about Intel's silicon manufacturing capabilities, further dragging down Intel's stock price. It is important to note that 18A is a crucial step for Intel to return to the ranks of leading chip manufacturers.

How difficult is wafer foundry?

In "Chip Wars," it is mentioned that in the early 21st century, Intel possessed the world's most advanced semiconductor process technology, staying ahead of competitors by introducing smaller transistors while maintaining the pace set during the Moore era. However, the gap between Intel and competitors like TSMC and Samsung began to narrow.

Intel missed the smartphone era led by Apple (AAPL.US), surviving on the monopoly of x86 achievements. The main reason was the high cost required to build a new platform, which made Intel, already enjoying the profits of monopolizing PC chips, reluctant to take risks.

Missing out on Apple and the rise of the internet economy, Intel's management at the time was indignant that tech giants like Apple managed cost-effectiveness by outsourcing production, rather than striving to adapt to the changes of the mobile era and compete with Asian suppliers.

It is worth noting that the investment cost of chip production is high, with thin profits. It cannot compare to selling ads on applications to earn over 60% in profits effortlessly. This is why major U.S. tech giants are venturing abroad to seek lower-cost manufacturing.

Today, building an advanced logic wafer fab costs at least $20 billion. The quantity and yield of chips produced by a company are correlated—the larger the production scale, the higher the yield. This has led to many chip manufacturers being eliminated by costs, while TSMC, with excellent resource utilization efficiency, has grown stronger, becoming the only supplier capable of providing Apple with cutting-edge chips.

Wafer foundry is not just about technological and process leadership; resource coordination and integration capabilities must also be excellent to profit through scale in a low-margin industry.

NVIDIA is a leading supplier of AI chips with the H100. However, designing the H100 and more advanced AI chips is already quite challenging, requiring high technical and human resource costs. If production is outsourced, it will undoubtedly affect its design capabilities and quality, potentially leaving it without enough resources to build its software ecosystem CUDA, which best showcases its advantages today. Therefore, outsourcing production to a third-party independent wafer fab like TSMC is the most cost-effective choice.

In today's highly specialized and internationalized environment, this long and winding industry chain consists of many supply links: for example, Applied Materials (AMAT.US) supplies machines for depositing chemical films in the silicon wafer processing process, Lam Research (LRCX.US) provides professional technology for silicon etching, KLA Corporation (KLAC.US) can measure nanometer-level errors on wafers and photomasks, Cadence Design Systems (CDNS.US), Synopsys (SNPS.US) HIMX (now part of Siemens Electronics) supplies electronic hardware and software design solutions, while ASML.US provides cutting-edge lithography machines...

After decades of accumulation and investment, TSMC has formed a very mature supply alliance, with dozens of chip design companies (whose customers design products around TSMC), selling IP, materials, and manufacturing machinery to supply services and production materials. The key to TSMC's success lies not in how high its R&D investment is, but in its ability to coordinate the supply alliance that has been accumulated for many years, ensuring that its suppliers all prioritize it and ensure the most efficient resource allocation and coordination within the timeframe.

Compared to twenty years ago, TSMC's foundry capabilities have far surpassed Intel.

TrendForce's latest report (September 2024) shows that in the second quarter of 2024, TSMC is still undoubtedly the top wafer fab, occupying 62.3% of the market share, with a gap as wide as 11.5% from the second largest wafer fab, Samsung, which holds only 11.5% market share. It can be seen that it is not easy for Intel to stand out in the wafer foundry field.

Furthermore, TSMC has another advantage - it is a neutral participant with no conflicts of interest with customers, while Intel itself is a downstream manufacturer in the wafer industry chain, conflicting with customers who commission third parties to produce wafers, and may also raise concerns about the leakage of customer intellectual property.

In summary, factors such as cost, technology, resources, and identity may constrain Intel's ambitious foundry plans. However, in the eyes of the capital market, cost-effectiveness is the most important consideration.

Financial View of Intel's IDM 2.0

As mentioned earlier, building a wafer fab requires an investment of at least $20 billion, not to mention the annual operating expenses and capacity expansion expenses.

Intel is expanding its capacity, investing around $20 billion in building wafer fabs in Arizona and New Mexico, investing over $28 billion in building wafer fabs in Ohio, acquiring Tower Semiconductor, and planning to invest up to €80 billion in the EU along the entire semiconductor value chain over the next decade. Thus, the investment will require at least $128 billion, which is half more than Intel's current market value of $83.1 billion.

However, Intel has launched the Semiconductor Collaborative Investment Program (SCIP) to mitigate cost pressures through capital cooperation with third parties.

Currently, Intel's semiconductor collaborative investment program involves two main entities, one being Fab 34 in Ireland, which introduced asset management company Apollo Global in the second quarter of 2024, which acquired a 49% stake in the factory for $11 billion, which will be used for wafer production in Intel 4 and Intel 3 processes. The other is the semiconductor collaborative investment program for two chip factories in Arizona, which will be similar in form to Fab 34.

In 2023, Intel sold a 20% stake in its IMS nanomanufacturing business to Bain Capital and a 10% stake to TSMC, generating a net profit of $1.4 billion from the transaction, and spun off the intelligent driving solutions company Mobileye (MBLY.US) for a public listing, unlocking over $30 billion in value In order to invest resources in IDM 2.0.

Moreover, after announcing a dividend of $0.125 per share in the third quarter, Intel stated that it will suspend dividends starting from the fourth quarter of 2024 to allocate funds to support the investments needed to execute its strategy, which naturally is not well-received by investors.

As of June 29, 2024, Intel held $11.287 billion in cash and cash equivalents, $17.986 billion in short-term investments, and generated a net cash inflow of only $1.069 billion from operating activities in the first half of the year, which seems insufficient to support its extensive capacity expansion plans.

From the perspective of sustaining business growth, Intel's operating profit is considerable if it focuses solely on its core business. In the first half of 2024, Intel's product revenue reached $23.732 billion, with an operating profit of $6.223 billion. However, all of this was used to offset the losses of $5.304 billion from its wafer fabrication plants.

Summary

The poor outlook of the foundry business has dragged down overall performance. Intel's attempts to free up resources for this underperforming business through layoffs and dividend cuts have failed to appease investors.

Rumors have recently circulated that Intel may spin off its wafer business, leading to a sudden increase in its stock price by over 9%, indicating investor sentiment.

Since Intel's ailment is known, its future stock performance is likely to depend on the decision regarding its wafer business