엔비디아·AMD 넘어 AI 반도체 경쟁서 승리할 저평가 칩주 주목

Most investors already know that there's a massive build-out of AI infrastructure underway and that it's likely to continue at least until 2030. However, the key question for long-term-thinking investors is where to invest as the build-out phase begins to slow and physical AI spending ramps up aggressively, with technologies like autonomous driving and robotics expanding. One answer is ON Semiconductor (ON +0.44%), and the good news is that the stock trades on excellent multiples and is more attractive than better-known AI plays like Nvidia (NVDA +0.14%) and Advanced Micro Devices (AMD +2.80%). ON Semiconductor is exposed to the AI infrastructure The power and sensing business is best known for its play on power density in its traditional core end markets, namely automotive (mainly electric vehicles and autonomous driving) and industrial. Those two end markets still dominate its sales now. NASDAQ: ON Key Data Points However, there are three reasons why it's set to be a major player in AI-related spending. First, it has a fast-growing power chip business selling into AI data centers. In fact, management expects its sales to the AI data center market to grow at a compound annual growth rate of 50% from 2026 to 2030 and hit $2.5 billion. Based on Wall Street estimates, sales of $2.5 billion would represent 19% of its overall sales in 2030. Moreover, the company is a technology partner with Nvidia in building out the architecture for the new 800-volt direct current (VDC) data centers, likely to drive the next phase of investment in AI data centers. The growing data center exposure is great news for the company and also signals strong medium-term prospects for Nvidia and AMD. But here's the thing: After the initial build-out of AI infrastructure, with massive spending on cloud compute and LLM training, the market is likely to shift toward implementation in the form of physical AI. ON Semiconductor is a physical AI stock now Second, management is positioned to be a major player in physical AI spending, driven by its upcoming acquisition of AI compute company Synaptics (SYNA -0.28%). ON Semiconductor already has the power, sensing, and control technology to support physical AI solutions, and adding the "brain" of Synaptics's AI edge intelligence solutions will enhance the new company's overall content in physical AI. It's undoubtedly going to be a fast-growing market, and a good way of understanding it is to compare the AI infrastructure build-out as building training clusters, inference, and language models to make AI function, but given the need for, say, a robot or an autonomous vehicle to make immediate decisions, the necessity of edge compute is essential. To be clear, Nvidia and AMD are also set to be major players in physical AI. For example, Nvidia builds out models (such as the open-source AI model Alpamayo for autonomous driving) to encourage automakers to buy Nvidia's chips, and AMD's acquisition of Xilinx in 2022 gave it a powerful position in edge computing for physical AI. However, the point is that ON Semiconductor has relatively more exposure to physical AI than Nvidia or AMD. Third, the growth of physical AI will organically expand ON Semiconductor's market, as it implies increased EV and autonomous vehicle sales (as well as semiconductor content per vehicle) and a host of industrial applications such as robotics and automation. ON Semiconductor is attractively valued Management recently held an investor day and outlined its expectations for 12%-14% annual revenue growth through 2030, with free cash flow (FCF) increasing from $1.6 billion to $3.5 billion in 2030. To put that figure into context, the company's market cap is only $32.8 billion at present, and if the company can achieve its aims, driven by growth in AI infrastructure and physical AI, the current price will prove significantly undervalued in my view.