Major Investment Banks Comment on Nvidia's Financial Report: The Cheapest AI Core Asset, or an Overestimated Myth in Computing Power?

Nvidia is proving to skeptics that the AI infrastructure investment boom is far from over with one earnings report after another. The world's largest AI chipmaker reported first-quarter revenue of $81.6 billion on Wednesday, up 85% year-over-year, once again significantly exceeding expectations; the guidance for the next quarter is even more impressive, with a new record of $91 billion on the table, indicating that not only is the growth rate not slowing down, but it is actually accelerating.

Morgan Stanley, JPMorgan, Bernstein, and Jefferies, four investment banks, collectively raised their target prices after the earnings report, with a highly consistent core judgment: as long as the capital expenditure of hyperscale cloud vendors continues to expand, Nvidia's growth ceiling has not yet been reached. However, analysts also admit that the mass production pace of the next-generation Vera Rubin chip, the competitive pressure of ASIC in-house chips, and the completely blocked Chinese market, are three swords hanging over this optimistic narrative.

Investment Banks Fully Increase Holdings, Raise Target Prices Across the Board

All four major investment banks maintained a bullish rating and collectively raised their target prices.

• Morgan Stanley slightly raised the 12-month target price from $285 to $288, reaffirmed "accumulate" and the preferred stock status in the semiconductor sector, and raised the revenue forecast for fiscal year 2027 to $393 billion

• JPMorgan raised the target price from $265 to $280, citing stronger network business performance and the continuously revised outlook for capital expenditure by hyperscale cloud vendors, maintaining an "accumulate" rating

• Bernstein was more optimistic, raising the target price from $300 to $315, using a 25x FY2028 earnings per share valuation method, raising the FY2027 revenue forecast to $398.8 billion, and reiterating an "outperform" rating.

• Jefferies raised the target price from $275 to $300, expecting fiscal year 2026 revenue to reach $394.4 billion, maintaining a "buy" rating.

The core consensus among various institutions is: demand for Blackwell architecture still far exceeds supply, the AI computing power investment cycle has not yet peaked, and Nvidia's moat in the data center ecosystem remains solid.

Bernstein analyst Stacy Rasgon pointed out that the company currently holds supply commitments of up to $119 billion, an increase of about $30 billion from the previous quarter, which to some extent alleviates concerns about supply constraints. Morgan Stanley emphasized that, calculated at about 17 times earnings per share for next year, Nvidia's valuation discount relative to similar AI infrastructure targets is "too significant to ignore."

Discrepancies: The Significance of Vera CPU, ASIC Threat, and Chinese Market

Analysts' disagreements mainly focus on three dimensions.

The first is the strategic significance and statistical approach of Vera CPU. Management disclosed that the company is expected to achieve $20 billion in Vera CPU revenue this year and claims to achieve a leading position in server CPU market revenue. However, there are reservations about this approach among various parties. Morgan Stanley believes that this $20 billion target may include "head nodes" CPUs delivered as part of GPU server racks, rather than CPUs sold independently; Jefferies also pointed out that the $20 billion figure may cover the entire rack-level supporting memory, and the actual content value of a single CPU may be magnified 5 to 10 times.

The second is the long-term threat of ASIC competition. Jefferies clearly stated that it is cautious about whether investors will return to Nvidia, citing ongoing market concerns about hyperscale cloud vendors' in-house chips (XPU) and the diversification of computing power. Morgan Stanley countered that market concerns about share loss "may have reached a low point," and competitors often compare their new generation of ASIC products with Nvidia's current products, rather than with the soon-to-be-released Vera Rubin, which is a fundamentally flawed comparison logic.

The third is the uncertainty of the Chinese market outlook. Nvidia's first-quarter data center revenue in China was zero, compared to $4.6 billion in the same period last year; the proportion of revenue from China has dropped from about 10% to only 6%. The company stated that it has obtained an H200 export license, but the actual revenue contribution is still zero due to uncertainty at the customer level. Currently, no data center computing business revenue from China is included in any quarterly guidance, and investment banks therefore view it as pure potential upside.

The Most Important Forward-Looking Indicators

Analy