MIC Forecast Sees AI Infrastructure Component Constraints Persisting Through 2028
Редакция ЦОД:Главное
Taiwan’s Market Intelligence & Consulting Institute (MIC) expects supply-demand tightness for several components used in AI and ICT infrastructure to continue through 2028, while NAND constraints could ease by the end of 2027. The institute forecasts that servers and the data-center supply chain will be the principal drivers of ICT shipment growth in 2027.
- MIC expects tight supply-demand conditions for HBM/DRAM, ABF substrates, ABF build-up materials and high-end CCL to continue through 2028.
- NAND shortages may ease by the end of 2027, according to MIC’s forecast.
- Global server shipments are forecast at 18.6 million units in 2027, including 5.4 million AI servers.
- MIC forecasts 12.3% year-on-year growth in switch-market value in 2027, linked to continued data-center deployment.
- Server and networking production is being shifted and diversified across Southeast Asia, Mexico and the United States.
MIC expects accelerated AI-infrastructure build-out by technology companies and public-sector entities, alongside US-China technology competition, to keep supply-demand conditions tight for HBM/DRAM, ABF substrates, ABF build-up materials and high-end copper-clad laminates through 2028. NAND is the exception: MIC sees a possibility that shortages in that segment could ease by the end of 2027.
For the data-center market, the forecast combines continued equipment-demand growth with an extended risk of component constraints. MIC projects global server shipments of 18.6 million units in 2027, up 12.4% year on year. AI-server shipments are forecast to reach 5.4 million units, an increase of 17.8% year on year.
MIC also expects the value of the switch market to rise 12.3% in 2027, attributing the growth to continued data-center deployment. It identifies the server and data-center supply chain as the leading contributor to ICT shipment growth in 2027.
The institute notes a shift and diversification of server and network-equipment manufacturing toward Southeast Asia, Mexico and the United States. This could indicate that supply-chain planning will remain a central operational consideration for AI-infrastructure deployments, particularly where access to constrained memory, substrate and board materials affects equipment availability.
The outlook points to a continued procurement risk for critical AI data-center components even as demand for servers, AI servers and networking equipment rises. For operators, integrators and suppliers, this could require earlier purchasing plans and more diversified supply chains, particularly for AI infrastructure.
- TechNews 科技新報 / КитайскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.