Author: XinGPT

After reading many bloggers' summaries of Huang Xiaoming's GTC speech today, I feel that none of them have hit the nail on the head.

The key point is actually just one thing: the Nvidia RTX Spark / N1X chip.

The biggest positive is: $ARM

Why is this good for ARM? There are three core layers of logic.

The first layer is the PC architecture narrative. In the past, the biggest problem with Windows on Arm was that Qualcomm was the only one pushing it, making it difficult for the ecosystem to expand. Now, with Nvidia, Microsoft, MediaTek, and multiple OEMs working together, the narrative of "Arm PC" has been upgraded from a low-power office laptop to a "high-performance AI PC / creator PC / local AI workstation" narrative.

The Verge suggests that Nvidia, Microsoft, and Arm are simultaneously hyping up "a new era of PC," implying that Qualcomm's de facto monopoly on Windows on Arm has been broken.

The second layer is royalty per chip. ARM's business model is license fee + royalty per chip. In its FY26 Q3 investor materials, Arm stated that royalty revenue grew by 27% year-over-year, driven by factors including the adoption of technologies with higher royalty rates per chip, such as Armv9 and CSS, as well as increased use of Arm chips in data centers. In other words, for ARM, high-ASP, high-core-count, high-end AI PC SoCs like Nvidia's are more valuable than ordinary low-priced IoT/mobile chips.

The third layer is ecosystem endorsement. Nvidia's entry into the Windows PC main processor market is essentially helping ARM validate a long-standing bull case: Arm architecture is not only suitable for mobile phones and low-power devices, but can also be used in high-performance PCs, local AI, AI workstations, and even further into data center CPUs. The Reuters report also mentioned that Nvidia's Vera CPU has been adopted by OpenAI, Anthropic, SpaceX, and others, and Jensen Huang emphasized that CPU/PC products are an important growth direction for the future.

In terms of the competitive landscape, Qualcomm is the most directly pressured, as Nvidia has taken away its differentiation in the Windows on Arm market. Intel/AMD will also face challenges, but in the short term, they can still maintain their core market share by relying on the x86 software ecosystem and mainstream PC shipments. For ARM, this is equivalent to gaining another heavyweight customer to help it fight in the PC market.

Of course, this is a short-term bullish catalyst. Whether it can translate into long-term profits for ARM depends on the actual shipment volume of AI PCs.

The market has already given its answer: ARM rose 14% in pre-market trading, while Qualcomm fell 7%.

ARM's only problem is its high price: Yahoo Finance analysts predict that ARM's average revenue for FY2027 will be approximately $5.97 billion, and for FY2028 approximately $7.96 billion. Even based on the projected FY2028 revenue, a share price of $400 still represents a P/S ratio of over 50.

This means the market no longer views ARM as a typical IP licensing company, but rather values it as an AI CPU platform company in advance. In other words, a $400 share price requires ARM to evolve from a mobile IP royalty company into a comprehensive company offering data center AI CPUs, Arm PCs, CSS, Armv9, and high-end custom computing platforms.

Of course, there's an even cheaper option: buy SoftBank. ARM's current market capitalization is approximately $370 billion, and SoftBank holds about 86.72% of ARM, corresponding to a holding value of approximately $321 billion. SoftBank Group's market capitalization is approximately 48.8 trillion yen, or about $306 billion. In other words, SoftBank's current overall market capitalization is slightly lower than the market value of its ARM stake. Furthermore, buying SoftBank also grants additional assets such as OpenAI, Vision Fund, PayPay, and SoftBank Corp.

However, holding SoftBank shares long-term presents a problem: SoftBank has been used by Masayoshi Son for financing, resulting in net debt and asset-backed financing, and the underlying assets are not pure. SoftBank also faces significant volatility risk from its non-ARM assets.

Bitwise bought 336,474 HYPE tokens in the past 12 hours.

Popular Articles