Morgan Stanley reinstates this AI chipmaker as its top pick in semiconductors
Morgan Stanley has reclassified Nvidia as its top semiconductor pick, touting the company's robust position in the market. The firm cites Nvidia's resilience against data center constraints and the growing importance of agentic artificial intelligence applications. At a recent meeting, Morgan Stanley analysts, including CFO Colette Kress and Head of Investor Relations Toshiya Hari, discussed several factors bolstering their renewed confidence in Nvidia's stock.
The analysts highlighted Nvidia's strategic advantages in managing land, power, and data center shell constraints, along with its capability to finance components. They noted that Nvidia's stock is trading at a reasonable valuation of 15 times fiscal year 2028 earnings. The firm also emphasized Nvidia's diverse customer base, which surpasses hyperscalers and frontier labs, with a significant portion of revenue coming from AI model companies, neoclouds, sovereign entities, and enterprise customers, thereby reducing exposure to land, power, and shell constraints limiting U.S.-based hyperscalers.
Additionally, the analysts pointed to Nvidia's growing revenue per gigawatt, with the Feynman architecture anticipated in 2028 expected to drive revenue per gigawatt above $50 billion. The firm cited Nvidia's Vera Rubin platform, which promises up to 25 times token improvement compared to Blackwell. The company's strategy to increase revenue per gigawatt substantially was also acknowledged.
Furthermore, Nvidia's agentic AI demand was addressed, with the firm noting that while persistent agents will escalate CPU demand, GPU spending will remain the dominant portion. Nvidia projects that an agentic user could generate ten times the daily tokens of a human user. Morgan Stanley also highlighted the company's mobilization of half a trillion dollars of non-Nvidia capital to invest in AI data centers, projecting CPU revenue to expand from $20 billion in 2026 to $40 billion in 2027.
Written by urgent.news from Investing.com's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.