Morgan Stanley Bullish on Long-Term Growth of AI Semiconductor Infrastructure

Overview

This Morgan Stanley report provides a deep analysis of the future development of AI computing infrastructure, covering multiple areas including CPU, GPU, ASIC, optical components, and Chinese chips. The report forecasts that by 2030, the global semiconductor market will reach $1.5 trillion, with AI semiconductors accounting for half of it. The report particularly focuses on TSMC's leading position in advanced packaging technology, NVIDIA's dominant advantage in the AI accelerator market, and the rising opportunities for domestic Chinese chip manufacturers.

Core views

The core viewpoint of the report is that AI computing demand will continue to drive growth in the semiconductor industry. Capital expenditure from cloud service providers (CSPs) remains strong, with the top 10 CSPs globally expected to spend $685 billion in 2026. As a key supplier, TSMC's CoWoS advanced packaging capacity is expected to expand to 200kwpm by 2027. In specialized fields, the report favors the AI training and inference chip markets, especially NVIDIA's GPU product lines. It also points out that despite NVIDIA's powerful GPUs, cloud service providers still need custom ASIC chips to meet specific needs. China's AI chip market is expected to grow from $19 billion in 2024 to $67 billion by 2030, benefiting domestic manufacturers such as Huawei and Cambricon due to the trend of local substitution.

Analysis framework

The report uses a comprehensive supply chain analysis method, studying upstream wafer manufacturing (such as TSMC), midstream packaging and testing (such as CoWoS capacity), and downstream applications (such as cloud service provider demands). By comparing the technical approaches and market positioning of major global AI chip manufacturers, it evaluates the investment value of each segment. For the Chinese market, the report analyzes multiple dimensions such as policy support, technological gaps, and domestic substitution space. It specifically focuses on China's progress in expanding advanced process capacities, HBM memory compatibility, and its potential impact on the global supply chain structure.

Methodology notes

• Industry/sector analysis frameworkSupply-demand framework

  Semiconductor industry supply-demand analysis

  The report analyzes the alignment between foundry capacity expansion and AI chip demand growth to assess industry supply-demand balance, such as comparing TSMC's CoWoS capacity plans with customer demand.

• Valuation methodPE/PEG valuation

  Semiconductor company valuation

  Uses PE and PEG valuation methods for covered semiconductor companies, assigning different multiples based on industry conditions and company growth prospects.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• TSMC

  AI chip manufacturing leader, benefiting from growth in CoWoS advanced packaging demand

  Strengths

  Technological leadership, covers global major AI chip manufacturers

• NVIDIA

  Dominant player in the AI training chip market

  Strengths

  CUDA ecosystem advantage, leading product performance

• Huawei/Cambricon

  Main suppliers of Chinese AI chips

  Strengths

  Local market advantage, policy support

  Weaknesses

  Still a gap in technology compared to international leaders

Key data

• Global semiconductor market size (2030e)$1.5 trillionAI semiconductor accounts for 50%
• Cloud service provider capital expenditure (2026e)$685 billionTotal of Top 10 CSPs
• TSMC CoWoS capacity (2027e)200kwpmSignificant expansion compared to current levels
• China AI chip market size (2030e)$67 billionCAGR approximately 20% from 2024-2030

Impact & implications

The report believes that the continuous growth of AI computing demand will reshape the semiconductor industry landscape. On one hand, leading foundries and packaging companies like TSMC will gain more market share; on the other hand, the rise of specialized AI chips (such as ASICs) will change the traditional GPU-dominated market structure. For the Chinese market, the promotion of technical autonomy will accelerate the development of the domestic supply chain, but it also faces challenges in technology catching-up and capacity construction.

Risks

• AI demand falling short of expectations leading to overcapacity
• Escalation of US-China tech disputes affecting supply chains
• Risk of changes in technical pathways

What to watch

• Changes in cloud service provider capital expenditures
• Progress of advanced packaging capacity ramp-up
• Improvement in China's chip self-sufficiency rate