At first, people said Intel was the stock of America's national fortune, and I scoffed—how could stock trading have become a MEME?

Later, it skyrocketed, from 20 to 80, and people were saying it was undervalued.

There's no way around it, missing out isn't scary, what's scary is not knowing why you missed out.

As I missed out, I researched all the way:

Intel is actually two companies:

One is Intel Products — a mature cash flow machine selling x86 CPUs (servers + PCs + AI PCs). The total revenue in 2025 is expected to be about 46 billion dollars, with a gross margin of 40-45%, which is a typical "long slope, not too thin snow" business in the sense of Duan Yongping.

The other is Intel Foundry — an advanced foundry burning money gambling on American semiconductor sovereignty. The external revenue for 2025 is only a few hundred million dollars, with an annual operating loss of 8.5 to 13.4 billion dollars, but it secured a 10% equity stake from the U.S. government, 2 billion dollars from SoftBank, 5 billion dollars from NVIDIA, and an initial order of 14A from Elon Musk's Terafab—this is a typical 0→1 monopoly candidate.

The valuation logic, growth logic, and risk structure of these two companies are completely different. If they are scored together using a single P/E ratio, the result will inevitably be a mess.

To see Intel clearly, you must look at them separately.

1. Look at the trends: Three structural trends, where does Intel stand

Trend One: AI infrastructure is a structural theme for the decade

This one is largely undisputed. Gartner, McKinsey, and Goldman Sachs are saying similar things—by 2028, global inference computing power consumption will exceed that of training by more than three times, and in China, it could be 4:1. The AI inference market is expected to be about 106 billion dollars in 2025, and 255 billion dollars in 2030, with a compound annual growth rate of 19%.

More critically, the economic structure of inference is different from that of training. Training involves one-time investment, where GPUs are the main actors; inference involves ongoing consumption of hundreds of millions of requests a day, with CPUs playing an irreplaceable role as the scheduling center. When Agentic AI transforms from concept to enterprise deployment, the main battleground for serial logic processing, context switching, tool invocation, and API orchestration is the CPU.

This trend is favorable for Intel, but Intel is not the biggest beneficiary. The biggest beneficiaries are NVIDIA (for training), TSMC (for manufacturing), and SK Hynix (for HBM). Intel is a participant, not the main force. What it can capture is the "control plane" part of CPUs in AI systems—this is an important position, but not a top-tier one.

Trend Two: U.S. semiconductor sovereignty is an irreversible geopolitical strategy for a decade

This trend is truly unique to Intel.

In August 2025, the Trump administration exchanged unallocated funds from the CHIPS Act for a 10% stake in Intel; this is not a one-time event, but a paradigm shift in national industrial policy. The previous CHIPS model was "government gives money, companies do it themselves"; the subsequent model is "government invests, government endorses, government backs." The possibility of this model extending to other strategic industries is being discussed—nuclear power, critical materials, defense chips, etc.

Intel's unique position in this trend is: the only option for advanced process foundries within the U.S. mainland. Even if TSMC's plant in Arizona starts producing 2nm by 2028, the U.S. Department of Defense's Secure Enclave project will always be given to Intel. The Pentagon will not hand over its most secret chips to a company headquartered in Taipei, no matter where its production capacity is.

This is a "secret" in the Thiel sense—a truth that few agree with but may be extremely important: if the situation worsens, America's AI hegemony without Intel is a house of cards. This secret itself may not necessarily need to be fulfilled, but it defines Intel's "lower limit"—the U.S. government will not allow Intel to disappear.

This trend manifests as a unique monopoly characteristic for Intel.

Trend Three: Supply chain diversification pressure from hyper-scale customers

AWS, Microsoft, Google, Meta, OpenAI—none of these customers are willing to place all AI infrastructure solely on TSMC. It's not that they don't trust TSMC; it's that any board cannot accept a "single point of failure" in the supply chain.

This creates a structural window—only Intel can provide truly advanced process foundry services within the U.S. mainland. Microsoft Maia2, AWS custom AI chips, and Google's IPU collaborations all establish themselves under this logic.

However, the benefits of this trend for Intel are conditional. The condition is that Intel can secure the orders—yield must be good, and production capacity must be large. The harsh reality revealed by institutional-level analysis is: Intel's monthly production capacity at 18A is about 10,000 to 15,000 wafers, while TSMC's N2 is around 1.3 to 1.4 million wafers, a 10-fold gap. Even if all hyper-scale customers want to allocate 20% of their orders to Intel, it cannot handle them.

The upper limit of benefits from this trend is locked by Intel's own capacity ramp-up speed.

Conclusions drawn from the overlap of three trends

The track Intel is on is real—AI infrastructure is on an upward trajectory for the next decade, U.S. semiconductor sovereignty is an irreversible trend for a decade, and supply chain diversification will persist for the next decade. But Intel's position within these three trends is:

AI infrastructure: participant (main force is NVIDIA)

U.S. semiconductor sovereignty: unique main force (no substitute)

Supply chain diversification: candidate beneficiary (locked by capacity)

This is a "favorable trend but not the biggest beneficiary" target. This position has its advantages—the lower limit is supported by national strategy, making it hard to disappear. But it also has its limitations—the upper limit is suppressed by the respective dominance of NVIDIA and TSMC, making it difficult to truly achieve explosive growth.

2. Look at the business model: two companies, two perspectives

(Please visit the link for better formatting)

Intel INTC Research Report: America's National Fortune Stock?

https://mp.weixin.qq.com/s/5T_AGkb8D8mU-psyL2dxYA

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。