Computex Highlights: Strong AI Semiconductor Demand, Sufficient TSMC Capacity
AI summary card
Computex Highlights: Strong AI Semiconductor Demand, Sufficient TSMC Capacity
Morgan Stanley summarizes the Computex conference, emphasizing that Agentic AI drives CPU server demand, TSMC capacity supports Nvidia's growth in 2027, and MediaTek's AI PC chip shipments are expected to reach 5-8 million units.
- Agentic AI becomes a focus, Arm CPU gains attention
- Sufficient TSMC capacity to support growth in 2027
- Nvidia RTX Spark shipment forecast at 5-8 million units
- MediaTek EPS contribution estimated at 5-10%
Report interpretation
Overview
The report summarized the key points of the first day of the Computex 2026 conference, focusing on how Agentic AI drives demand for CPU servers, TSMC's sufficient capacity to support Nvidia's robust growth in 2027, and MediaTek's AI PC chip shipment estimates and impact.
Core views
The Computex conference focused on Agentic AI and CPU servers, with Nvidia, Arm, and Qualcomm all emphasizing that Arm-based server CPUs will meet future Agentic AI needs. The Nvidia Vera CPU significantly outperforms x86, improving by 1.7x and 1.9x in compilation and Python tasks respectively. Supply chain checks show preparation quantities of 2.5-4 million units, with Rubin GPUs entering production without delays. TSMC has sufficient capacity, and Nvidia confirmed that supply in 2027 can support strong growth. Even with supply expansion, demand may still exceed supply, as the AI semiconductor market is not a zero-sum competition. The Nvidia RTX Spark AI PC chip, co-designed by Nvidia and MediaTek, is expected to ship 5-8 million units in 2026, with an average royalty fee of $40 per chip, contributing 5-10% to MediaTek's 2026 EPS.
Analysis framework
The institution analyzed AI hardware trends through Computex keynote speeches and investor Q&A sessions; combined supply chain checks to verify product readiness; used performance comparison data to evaluate technical advantages; and estimated shipment volumes and financial contributions based on brand pricing models.
Methodology notes
Shipment Volume and Price Breakdown Model
The report quantifies the impact of the new AI PC chip by estimating its shipment volume and average royalty fee, breaking down its contribution to MediaTek's EPS.
Product Life Cycle Analysis
The report evaluates the short-term impact of product release schedules (e.g., RTX Spark delayed to the second half of 2026) using life cycle stage management principles.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- MediaTek (2454.TW)Benefits from AI PC chip shipments
- Strengths
- Co-designs RTX Spark with Nvidia, entering the high-end market
- TSMC (TSM.US)Benefits from AI-driven capacity utilization
- Strengths
- Sufficient capacity to support Nvidia's growth
Key data
- Nvidia RTX Spark Shipment Forecast5-8 million units2026
- MediaTek EPS Contribution5-10%2026, based on average $40/chip royalty fee
- Nvidia Vera CPU Preparation Quantity2.5-4 million unitsSupply chain check
Impact & implications
The report believes that the AI semiconductor market will continue to grow, benefiting supply chain companies in a non-zero-sum competitive landscape. MediaTek benefits from AI PC chip shipments but must monitor the impact of high prices on demand. TSMC's capacity ensures stable industry supply.