Morgan Stanley Has Once Again Listed Nvidia, The AI Chip Giant, As Its Top Pick In The Semiconductor Sector!

Morgan Stanley has re-listed Nvidia as the top pick for the semiconductor sector, citing the company's strong competitive advantage in a data center resource constraint environment while benefiting from the rapid rise of agent AI applications.

After a recent roadshow meeting with Nvidia's CEO, CFO, and head of investor relations, Morgan Stanley analysts listed several key factors supporting their renewed confidence in the stock. The firm believes that Nvidia has unique advantages in addressing resource constraints such as land, electricity, and data center space, and its component financing capabilities position the company well in the current market environment. Morgan Stanley points out that the stock is not currently valued, with a price-to-earnings ratio of only 15 times for fiscal year 2028. 

TMGM Analysis: Financial Market News, Economic Calendar & Market Insights

After completing investor meetings with company management, the firm emphasized several core advantages. NVIDIA's customer base has surpassed hyperscale cloud providers and frontier labs, becoming more diversified. Management revealed that about half of its business, possibly the faster-growing segment, comes from a broader range of AI model companies, emerging cloud service providers, sovereign entities, and enterprise clients. This diversified customer structure helps alleviate resource constraints such as land, electricity, and data center space faced by hyperscale cloud providers in the US. 

Morgan Stanley also highlighted Nvidia's strategic move to significantly increase revenue per gigawatt. The company showcased the Feynman architecture, expected to launch in 2028, by which time revenue per gigawatt will increase from $40 billion to over $50 billion. Additionally, the Vera Rubin platform is expected to achieve up to 25 times the token processing efficiency improvement compared to Blackwell.

Morgan Stanley also analyzed the impact of agent AI on the demand side. Although continuously running agents will drive up CPU demand, NVIDIA management says GPU spending will still account for the largest share. The company expects that a single agent user can generate up to ten times the number of tokens per human user per day.

Market Insight:

Morgan Stanley pointed out that NVIDIA has leveraged $500 billion in non-proprietary capital to build AI data centers. The company expects CPU business revenue to grow from $20 billion in 2026 to $40 billion in 2027.