ON Semiconductor, the Phoenix-based power and sensing chip supplier, could emerge as a sleeper winner in AI infrastructure, according to a prediction highlighted by Yahoo Finance.

The thesis rests on a straightforward observation: AI data centers consume enormous amounts of power, and every watt delivered to a rack must pass through power management and conversion silicon. That is the territory ON Semiconductor (onsemi) has built its business on for decades, alongside its larger automotive and industrial franchise.

The company's portfolio spans silicon carbide (SiC) power devices, MOSFETs, and power management integrated circuits — product families that server makers and data center operators increasingly need as rack power densities climb. AI accelerators from NVIDIA and its rivals have pushed individual rack designs toward power levels that make efficient power conversion a first-order engineering constraint rather than a rounding error.

ON Semiconductor's traditional strength lies elsewhere. The company is a major supplier to the automotive market, where its SiC devices serve electric vehicle powertrains and its image sensors support advanced driver assistance systems. That automotive exposure has been a double-edged sword in recent quarters, with EV demand softness weighing on results. An AI-driven power business could diversify that revenue mix.

The Yahoo Finance prediction frames onsemi as underappreciated in the AI conversation. Most investor attention on AI infrastructure flows to GPU designers, hyperscalers, and advanced packaging suppliers. Power semiconductor vendors sit further down the value chain, and the market has largely priced them as automotive and industrial plays rather than data center beneficiaries.

The scale of the opportunity is tied to physics. AI servers draw multiples of the power consumed by conventional servers, and data center operators face both electricity costs and grid interconnection limits. Higher efficiency at each conversion stage — from grid to rack to board to chip — translates directly into lower operating cost and more usable compute per megawatt. Wide-bandgap semiconductors such as SiC, and the power ICs around them, are central to that efficiency push.

ON Semiconductor has invested heavily in SiC manufacturing capacity in recent years, building vertically integrated production from boule growth to device fabrication, primarily on 150mm and migrating toward 200mm wafers. That capacity was aimed largely at EV customers, but the same technology serves high-voltage power delivery in data centers.

The prediction carries the usual caveats of a speculative pick. It is an analyst-style thesis rather than a confirmed revenue disclosure: Yahoo Finance's prediction piece does not attach specific AI revenue figures or design-win announcements from onsemi itself. The company has not, in this material, quantified its AI-related bookings or named data center customers.

What is concrete is the direction of the market. AI infrastructure buildouts continue at pace, and power delivery has become a recognized bottleneck. Suppliers that can address high-voltage, high-efficiency conversion are seeing demand from a customer base that historically bought far less from them.

For ON Semiconductor, the question is whether AI power demand can grow large enough to move a company whose revenue base remains anchored in automotive and industrial markets. The Yahoo Finance prediction suggests it can. Investors will look for confirmation in onsemi's coming earnings reports — specifically, any disclosure of data center design wins or a revenue segment breakout that shows power products shipping into AI infrastructure at material volumes.