Bank of America: The "AI bubble theory" is a "real benefit" for chip stocks, avoiding excessive crowding, NVIDIA is undervalued

Author: Dong Jing

Source: Hard AI

Regarding the recent "AI bubble theory," Bank of America has put forward a distinct contrary viewpoint: it is a "reverse benefit" for chip stocks, helping to prevent excessive crowding. NVIDIA is significantly undervalued, as its stock price does not reflect its true growth potential.

On November 11, according to Hard AI, Bank of America stated in its latest research report that the prevailing skepticism in the market regarding AI capital expenditures is not a risk but rather a "reverse benefit" that ensures lasting returns in this sector. This skepticism helps prevent excessive crowding in the sector.

The bank believes that the recent volatility in AI chip stocks primarily stems from correctable macro factors rather than a deterioration in the fundamentals of the AI spending cycle. In fact, based on the strong performance of the AI supporting industry and NVIDIA's optimistic outlook, underlying demand remains solid.

Bank of America stated that the doubts about OpenAI's grand plans are overly simplistic and overlook the fact that the real drivers of AI investment are the defensive strategies of large tech giants. Notably, as the industry leader, NVIDIA's current stock price does not reflect its true growth potential and is significantly undervalued.

AI Skepticism: A Healthy "Reverse Positive" Signal

Despite the average decline of 7-8% in large AI semiconductor stocks last week, Bank of America believes this is not a signal of problems in the AI spending cycle. On the contrary, stock price fluctuations are more driven by macro factors, such as concerns about a U.S. government shutdown, weak employment data, tariff turmoil, and misinterpretations of comments about OpenAI. These are all correctable short-term noises rather than fundamentally bearish signals.

The real fundamental signals are quite the opposite. The report emphasizes that AI-related supporting sectors performed strongly, with storage and small to medium optical module stocks rising 14% last week, indicating that the construction of AI infrastructure is fully underway.

Additionally, the data center business outlook for fiscal year 2025/26 disclosed by NVIDIA at the recent GTC conference, with a scale exceeding $500 billion, further confirms the strength and sustainability of AI demand.

Therefore, Bank of America believes that the widespread skepticism actually helps the market remain calm, reiterating a "buy" rating on core beneficiary stocks of AI construction, including NVIDIA and Broadcom in the data center sector, as well as Lam Research (LRCX), KLA (KLAC), and Applied Materials (AMAT) in the semiconductor equipment sector.

Refuting "AI Spending is Unsustainable": Defensive Investments by Giants are Key

A common bearish argument in the market is: "Since OpenAI cannot justify its $1.4 trillion long-term commitment, AI stocks must be overvalued." Bank of America dismisses this viewpoint as "lazy and taken out of context."

First, OpenAI's grand plans have yet to materialize and will be constrained by real-world conditions such as power and data center space. More importantly, the vast majority of current AI spending does not come from startups but is driven by highly profitable publicly traded "hyperscalers." Bank of America believes that for these tech giants, upgrading to accelerated computing (from traditional CPU computing) is not only "mission-critical" but also a form of "defensive" investment.

The report uses Google as an example: Its capital expenditure of up to $92 billion is aimed at "defending" its leadership position in the search business, which exceeds $200 billion, and preventing users from shifting to emerging competitors like ChatGPT and Perplexity.

Bank of America states that, at the same time, private AI companies represented by OpenAI (with over 1 million commercial customers) and Anthropic (with over 300,000 commercial customers) are rapidly attracting enterprise users, which in turn will continue to exert pressure on publicly listed software and Infrastructure as a Service (IaaS) providers, forcing them to increase AI investments. Therefore, the core driving force behind AI spending is solid and strong.

NVIDIA: Not a bubble, but undervalued

The report highlights that NVIDIA's stock is highly attractive, as its pricing only reflects "very cautious AI deployment."

According to NVIDIA's recent disclosure of approximately $500 billion in data center orders for the fiscal year 2025/26, Bank of America estimates that the company's earnings per share (EPS) in 2026 could reach around $8.

This means that a company with year-on-year sales/EPS growth rates of up to 50%/70% has a price-to-earnings ratio of only 24 times, which is comparable to the market average and can be considered "not a high valuation."

To further illustrate its potential, the report makes a scenario assumption:

Even if by 2030, AI capital expenditures only reach 50% of NVIDIA's forecast range of $3 trillion to $4 trillion, the company's EPS could exceed $40 per share at that time. This means that based on the current stock price, its forward price-to-earnings ratio would be below 5 times.

Bank of America believes this figure is "irrational" and uses it to support its view: Despite media headlines filled with bubble theories, NVIDIA's stock price is actually undervalued.