The hype around generative artificial intelligence is showing signs of fatigue. For the AI hardware giant Nvidia (NVDA), it achieved a stellar market performance in 2025, with its stock price soaring by 32%, significantly outperforming the S&P 500 index. However, the slowdown in the hardware boom has led the market to question whether this company, with a market capitalization of $4.3 trillion and ranking first globally, can sustain its growth momentum into 2026.
Nvidia’s success is built on a solid foundation, with its key strength lying in constructing formidable competitive barriers. This advantage largely relies on its proprietary CUDA software platform. After decades of accumulation within the developer ecosystem, this platform has formed an extensive system of tools and user knowledge, creating a moat that is almost unshakable, even if competitors can offer comparable hardware specifications. Thanks to the vigorous development of generative AI, Nvidia has fully unleashed its accumulated advantages. Its third-quarter revenue increased by 62% year-over-year to $57 billion, with the sustained strong sales of new data center AI chips being the primary driver.
Although Nvidia’s business model is favorable to shareholders, it is costly for its customers. The company’s gross profit margin in the third quarter was as high as 73.4%, a figure typically seen in pure software companies. Its customers, such as industry leaders who continue to report massive losses, have a fiduciary duty to control costs. For example, the developer of ChatGPT is believed to have incurred losses exceeding $11.5 billion in the most recent quarter, and some analysts suggest that related losses could expand sharply in the future. The financial pressure on this customer base reveals the instability of the industry’s foundation, which could ultimately affect investment sentiment.
Furthermore, a growing number of top-tier customers, such as Alphabet and Amazon, are increasing their investments in Application-Specific Integrated Circuits. These custom chips are tailored for specific tasks and are far less expensive than Nvidia’s general-purpose GPUs, representing a potential long-term threat.
Despite challenges such as escalating customer losses and the shift towards custom chips, the likelihood of Nvidia facing a sudden collapse appears low. Its stock price remains reasonable relative to its growth prospects, with a forward price-to-earnings ratio of approximately 23x, lower than the average of the Nasdaq 100 index. This indicates that current market valuations have already factored in expectations for a significant slowdown in its growth.
The latest research report from Bernstein points out that Nvidia’s current valuation relative to the Philadelphia Semiconductor Index is at a historically very attractive level, providing a favorable foundation for future stock price returns. The report analyzes that Nvidia is currently trading at a discount of approximately 13%, placing it at an extremely low percentile within its historical valuation distribution. As a core beneficiary of AI, its stock price is around 25 times its expected earnings for the next year, a forward P/E ratio that sits at the lower end of its range over the past decade.
The report further emphasizes that this valuation level not only appears “quite cheap” in absolute terms, but historical data also shows that investors who bought and held at such valuation lows have achieved very substantial average returns over a one-year period. Bernstein maintains an “Outperform” rating.
