Goldman Sachs: Nvidia's sell-off "overdone" in major technology transition period "calculating short-term cost returns is futile"

AI faith shaken, NVIDIA stock price pulls back from highs, but Goldman Sachs believes recent selling is "overdone".

On September 9th, Goldman Sachs analyst Toshiya Hari told Yahoo Finance that despite the recent poor stock performance, he remains optimistic about NVIDIA:

"Firstly, the demand for accelerated computing remains very strong. We tend to focus on mega-cap enterprises (such as Amazon, Google, Microsoft, etc.), but you will see the demand expanding even in sovereign nations."

Hari still maintains a buy rating on NVIDIA, when asked if the Goldman Sachs team believes NVIDIA's stock has been oversold, Hari said, "Yes, we believe so."

In a recent report, the Goldman Sachs stock research team wrote that since the beginning of 2023, there has been an "almost 180-degree shift" in investors' attitudes towards artificial intelligence. Investors' patience is wearing thin, they want to see - rather than be told - improvements in AI-driven revenue streams and profit margins.

Nevertheless, the Goldman Sachs team wrote that with profound generational technological shifts like artificial intelligence, "making judgments based on short-term cost and return economics is futile."

Goldman Sachs believes the focus should be on the long game, estimating that by the second half of 2025, generative AI will begin to make substantial contributions to industry growth.

Hari stated that NVIDIA's competitive position remains very strong, in the commercial chip field, NVIDIA is the top choice, even compared to custom chips, "they also have an advantage in terms of innovation speed."

At the end of August, NVIDIA released an earnings report that exceeded expectations, but it was still not good enough for Wall Street, leading to a continuous pullback in stock prices. Last week, NVIDIA plummeted nearly 10%, losing about $400 billion in market value. On Monday, NVIDIA regained some lost ground, rising by 3.5%.

A recent research report from Bank of America also stated that it is premature to question AI capital expenditure, at least until 2026:

Corporate capital expenditure is necessary, AI capital expenditure not only drives new business opportunities, but is also crucial for the existing moats and profit pools of search and social-related enterprises. Currently still in the first wave of the big language model craze, with NVIDIA Blackwell starting to ship, the industry will give itself another 1-2 years to focus on building Blackwell chips.

Bank of America believes that with the recent pullback, NVIDIA's stock price has fallen to the lowest quartile range of valuation in the past five years, combining current doubts and unfavorable factors to bring investors a buying opportunity.