Nvidia's (NVDA) stock performance has plateaued with shares up less than 3% year-to-date, signaling the end of its previous rocket ship rally. Investors are increasingly cautious about the stock for several reasons, including the unprofitability of generative AI software and the emergence of potentially cheaper open-source rivals from China, which could challenge Nvidia's major customers like OpenAI and Meta Platforms.
Nvidia is set to release its fourth-quarter and full-year earnings report after the market closes on February 26. This report is crucial for investors as it will provide insights into Nvidia’s growth prospects and the state of the AI industry, which is currently facing challenges such as high spending and low profitability (with OpenAI reporting a $5 billion loss in 2024).
The company remains optimistic, forecasting around $37.5 billion in revenue, reflecting a 70% year-over-year growth. Although this indicates a deceleration compared to last year's 265% revenue surge, it remains impressive for a company valued at $3.4 trillion. For context, Amazon recently reported a top-line growth of just 10% in its latest quarter.
Nvidia's dominance is attributed to its ability to release new and improved graphics processing units (GPUs) for training and running AI models. Despite the high cost of these products (with the latest Blackwell chips priced between $30,000 and $40,000 each), they offer substantial performance enhancements, potentially leading to operational cost and energy savings compared to older chip designs.
The introduction of DeepSeek R1, an open-source large language model (LLM) from China, poses a potential threat to Nvidia's growth thesis. This Chinese application reportedly outperforms industry leader ChatGPT on key tasks and was developed for a mere $5.6 million using Nvidia's less advanced H800 chips, designed to meet U.S. export restrictions to China.
If companies can produce top-tier LLMs with less advanced chips, it might undermine the market for Nvidia's most sophisticated hardware, potentially affecting its growth and margins. Despite this, there are positive signs for Nvidia and its shareholders.
Nvidia's fiscal fourth-quarter report includes data up to January 2025. Since DeepSeek R1 was released late in this period, it is unlikely to have a significant impact on guidance. Moreover, some experts have challenged DeepSeek's claims, with SemiAnalysis estimating its hardware costs at over $500 million, contrary to the $5.6 million claimed by its developers. Additionally, DeepSeek faces regulatory scrutiny over data safety, having been banned from South Korean app stores due to privacy issues.
Major Nvidia customers, such as Meta Platforms, continue to invest in AI hardware. Despite the potential threat from DeepSeek, Meta is not reducing its AI spending. CEO Mark Zuckerberg anticipates investing hundreds of billions of dollars in the industry over the long term.
DeepSeek is unlikely to disrupt Nvidia's momentum in the fiscal fourth quarter or the calendar year 2025. The upcoming release of new Blackwell AI chips is expected to sustain the company’s elevated growth rate. With a forward price-to-earnings ratio (P/E) of 32, the stock is reasonably priced based on its fundamentals.
However, while Nvidia's stock is attractive due to its low valuation, it is best considered a hold rather than a buy. Although DeepSeek won't immediately deflate the generative AI bubble, it highlights vulnerabilities—especially since leading LLM developers like ChatGPT remain unprofitable. Given the speculative and risky nature of the industry, this valuation is justified.
