Key insights
- Quanta Services projects strong growth in AI server sales, doubling in 2025 and growing triple digits in 2026. Capacity expansion is planned in Thailand and the US. The transition from GB300 to VR200 platforms will occur in late 2024, with VR200 shipments ramping up in 2027. NVIDIA's standardization efforts for Rubin computing trays could impact market share, potentially benefiting the ASIC segment. This signals continued investment and growth in the AI infrastructure space, indirectly benefiting US
Investing.com - Morgan Stanley reported Thursday that Quanta Services (TPE:2382) expects notebook shipments to remain flat to slightly up quarter-over-quarter in the second quarter, with full-year guidance unchanged at a decline in the high single digits to double digits year-over-year.
The company anticipates AI server sales will increase in double digits quarter-over-quarter in the second quarter. For the third quarter, GB300 shipments are expected to plateau without sharp declines, while the fourth quarter will likely mark a platform transition period from GB300 to VR200.
AI server sales doubled year-over-year in 2025 and are projected to grow in triple digits year-over-year in 2026. General server sales guidance remains at double-digit growth year-over-year in 2026, with units also expected to grow.
Notebooks represented less than 15% of the revenue mix in the first quarter and will stay below 20% for the full year.
Quanta's AI server capacity is set to double by the end of 2026 compared to the end of 2025, primarily in Thailand and the United States. The company's capital expenditure guidance stands at NT$30 billion.
VR200 rack shipments are expected to be limited in 2026, with sampling shipments likely to begin in the third quarter. The majority of shipments will occur in 2027.
Some customers are willing to accept a consignment model to ease original design manufacturers' capital expenditure expansion and working capital pressures, aiming to prevent potential supply constraints and roadmap delays from longer investment cycles.
NVIDIA is seeking to standardize computing trays for Rubin, though Quanta remains uncertain whether cloud service provider customers will accept this approach. NVIDIA may risk losing market share to the ASIC segment if it insists on standardizing the trays.