Overview: onsemi’s Strategic Pivot to Power-Density Architectures

onsemi (NASDAQ:ON) has outlined a comprehensive strategy centered on power-density technologies for artificial intelligence data centers, electric vehicles, industrial systems, and "Physical AI." During a recent analyst event, President and CEO Hassane El-Khoury emphasized that future AI infrastructure is increasingly constrained not just by compute and memory, but by the ability to efficiently deliver and manage power under sustained peak loads. "More power is really not enough," El-Khoury stated. "We need better power." To address this, the company has established ambitious long-term revenue, margin, and free-cash-flow targets through 2030, aiming to generate $2.5 billion in revenue specifically from AI data centers.

Technical Details: Embedded Power Platform and vGaN Innovations

At the core of onsemi's technological roadmap is the newly introduced Embedded Power Platform (EPP). Described as a platform rather than a single product, EPP utilizes a silicon wafer as an active package, embedding heterogeneous semiconductor products—including silicon carbide (SiC) and gallium nitride (GaN) components—into a wafer-level architecture. This approach is designed to simultaneously improve electrical, thermal, and mechanical performance while reducing system-development time. The company also highlighted its vertical gallium nitride (vGaN) technology, which is supported by its existing Syracuse facility without requiring new greenfield fab investments.

onsemi cited several significant product-performance benchmarks:

  • An 800-volt-to-6-volt or 12-volt conversion system using vGaN that delivers 20% better efficiency and a 20% reduction in size compared to competing solutions.
  • An EPP-based SiC JFET solid-state circuit-breaker design that is 50% smaller and operates 20% cooler than the nearest competing product.

Market Context: The Shift to High-Voltage DC in AI Infrastructure

Achyut Shah, president of onsemi’s Power Solutions Group, explained that AI processors are transitioning from consuming hundreds of watts to kilowatt-scale requirements, while rack-level power needs are escalating from tens of kilowatts toward megawatts. This transition necessitates a fundamental shift from traditional AC distribution architectures to high-voltage DC systems extending from the grid directly to computing equipment. Shah noted that the initial transition toward higher-voltage rack designs and sidecar architectures will begin ramping in the next 12 to 18 months, with broader adoption of solid-state transformers expected around late 2028 or early 2029. Additionally, onsemi announced it won a "major" Vcore socket, with revenue expected by the end of 2026.

Implications for AI Data Centers: Scaling Rack-Level Power

The transition to higher-voltage architectures presents a massive semiconductor content opportunity for onsemi. Shah estimates that content per rack could increase from approximately $15,000 to more than $115,000. Consequently, the company expects its AI data-center revenue to exceed $500 million in 2026—more than doubling from the prior year—and double again in 2027. Utilizing a hypothetical 40% market compound annual growth rate, onsemi projects its AI data-center revenue will grow at least 50% annually, surpassing $2.5 billion by 2030. This growth trajectory is modeled to outpace the underlying market by at least 10 percentage points.

Industry Impact: Automotive and Physical AI Expansion

Beyond data centers, onsemi is targeting robust growth in automotive and industrial sectors. Sudhir Gopalswamy, group president for Intelligent Sensing and Analog, stated that automotive revenue is expected to grow about 9% annually through 2030, even assuming flat global light-vehicle production. The strategy leverages EPP-based traction inverters that can double power while using half the area, achieving four times the power density. Furthermore, onsemi is advancing SiC device-voltage capability toward 1,400 volts to support 1,000-volt batteries. Subaru is currently evaluating onsemi’s embedded power module technology for future vehicle platforms to achieve higher power density and dual-sided cooling. In industrial markets, the company targets 10% growth, noting that "Physical AI" applications represent a potential $6 billion addressable market by 2030, with humanoid robots offering up to $900 in potential content per unit.

Future Outlook: Financial Targets and Capital Allocation

CFO Thad Trent detailed onsemi's financial trajectory, noting the company has invested $8.7 billion over the past five years. Looking ahead, the company targets 12% to 14% annual revenue growth from 2026 through 2030, reaching just under $11 billion in revenue. The financial model aims for a 53% gross margin, a 38% operating margin, and a capital intensity of about 5% of revenue. Free-cash-flow margin is targeted at 30% to 35%, exceeding $3.5 billion in 2030. These baseline targets exclude the proposed Synaptics transaction, which is expected to add over 15% long-term growth and more than $200 million in synergies. onsemi plans to maintain balance-sheet flexibility and return 100% of free cash flow to shareholders, with $5.3 billion remaining under its existing share-repurchase authorization.