Lite-On expects AI-related revenue to reach 40%–50% of total revenue by 2027, with AI server orders visible through 2028. The company points to the industry's shift toward 800V DC power delivery as a major driver reshaping the data center power supply market.
Overview
Lite-On expects AI-related revenue to account for 40%-50% of its total revenue by 2027, with order visibility for AI server orders extending through 2028. The company's outlook, reported by DIGITIMES on November 15, 2026, links that growth to high-end AI power systems and to a broader industry transition toward 800V DC power delivery. At the same time, Lite-On says 800VDC specifications are undergoing major revisions, a factor that will delay volume production until late 2027. The combination of strong order visibility and a moving specification target makes Lite-On a useful lens for understanding how AI data-center power architectures are evolving.
The 800V DC Technical Shift
The move to 800V DC is more than a voltage change. It touches the entire power path inside and around the data center, from facility-level rectification and distribution to rack-level conversion, protection, and control. Higher-voltage DC distribution changes the engineering trade-offs around current handling, busbar design, connector and contactor requirements, isolation, and thermal management. It also affects how power shelves, server nodes, and cooling systems are co-designed. For power-supply vendors, the transition shifts value toward high-end AI power systems that can meet tighter efficiency, density, and reliability requirements. That is the context in which Lite-On expects AI power systems to become a much larger part of its business.
Order Visibility Through 2028
Lite-On's order visibility extends through 2028 as US cloud customers continue expanding AI infrastructure. That visibility is significant because AI server power is not a spot-market business; it requires qualification, design-in, and production ramps tied to specific platform generations. The company expects AI-related revenue to reach 40%-50% of total revenue by 2027, driven by high-end AI power systems. The revenue mix target implies that AI power is moving from an emerging category to a central pillar of Lite-On's business. It also suggests that the company is aligning its product roadmap and capacity planning with the multi-year buildout plans of US cloud customers.
Specification Revisions and Production Timing
The 800VDC transition is not proceeding on a fixed schedule. Lite-On says 800VDC specifications are undergoing major revisions, which will delay volume production until late 2027. That timing matters. Order visibility through 2028 gives the company a long demand runway, but the specification revisions create a gap between customer planning and volume manufacturing. Power-supply vendors must design to evolving requirements while preserving safety margins, efficiency targets, and compatibility with adjacent infrastructure. The delay also means that the 800V DC ramp will overlap with, rather than immediately replace, existing power architectures. For Lite-On, the challenge is to support current AI server programs while investing in the next-generation 800V DC platform.
Market Context: AI Infrastructure Demand
AI infrastructure demand from US cloud customers is the demand-side backdrop for Lite-On's outlook. AI servers place different demands on power systems than general-purpose servers, and the industry's shift toward 800V DC is reshaping the data-center power supply market. The transition is driven by the need to deliver more power through dense server racks while maintaining efficiency and reliability. As specifications evolve, the power supply market is separating into vendors that can support high-end AI power systems and those that remain focused on conventional architectures. Lite-On's guidance places it in the former category, with AI revenue expected to reach 40%-50% of total revenue by 2027.
Industry Impact on the Power Supply Chain
The 800V DC transition changes the competitive landscape for power-supply vendors. It raises the bar for high-voltage design, safety certification, and system-level integration. It also increases the importance of early engagement with cloud customers, because specification revisions can force repeated design cycles. Lite-On's order visibility through 2028 suggests that customers are willing to commit to long-lead power programs even while 800VDC specifications are in flux. That is a signal that power delivery is becoming a strategic constraint in AI data-center buildouts. Vendors that can manage the specification risk and deliver high-end AI power systems on schedule will be positioned to capture a larger share of the market.
Implications for AI Data Centers
For AI data-center operators, the 800V DC transition has implications for facility design, rack architecture, and deployment timelines. Higher-voltage DC distribution can change the layout of power conversion stages and the way power is delivered to dense AI racks. But the specification revisions mean that operators cannot assume a fixed 800V DC design will be available on their preferred schedule. Volume production is now expected in late 2027, so early deployments will likely rely on a mix of existing power architectures and transitional designs. Operators will need to work closely with power-supply vendors to qualify products against evolving specifications while keeping AI infrastructure buildouts on track.
Future Outlook
Lite-On's outlook points to a multi-year AI power upcycle with a moving technical target. AI-related revenue is expected to reach 40%-50% of total revenue by 2027, and AI server orders are visible through 2028. The 800V DC transition is a major driver of that growth, but specification revisions will delay volume production until late 2027. The key milestones to watch are the finalization of 800VDC specifications, customer qualification cycles, and the ramp of high-end AI power systems. If Lite-On can bridge the gap between current AI server programs and the 800V DC generation, it stands to benefit from a data-center power market that is being reshaped by AI infrastructure demand.