Nvidia (NASDAQ:NVDA) stock is down 15% year to date, and even more from its highs. The sell-off has reflected several things including, I feel, the currenlty-much-talked-about end to US exceptionalism, concerns about valuations in the artificial intelligence (AI) segment, and developments in China, notably DeepSeek’s efficient AI models. Let’s take a closer look.
Demand fears persist
Investors are increasingly wary that advancements like DeepSeek’s compute-efficient models could slow demand for Nvidia GPUs. DeepSeek is a Chinese AI lab and its ability to train high-performance AI models on cheaper hardware has raised concerns about the sustainability of Nvidia’s growth trajectory, particularly as the initial wave of AI infrastructure build-out potentially slows.
The release of DeepSeek R1 in January 2025 triggered a significant market reaction, with Nvidia’s market cap plummeting by nearly $600bn, marking the stock’s worst performance on record. Despite these fears, demand from hyperscalers remains robust, with major tech firms like Microsoft, Google, and Meta projected to significantly increase their capital expenditures, which could continue to drive Nvidia’s revenue growth into 2025.
Losing the Midas touch
At Nvidia’s GTC keynote on 18 March, CEO Jensen Huang focused on expanding AI’s real-world applications, particularly in robotics and physical AI systems. He introduced the Rubin AI chips and emphasised the potential for AI to transform industries through automation and intelligent systems. Huang’s vision highlighted a future where AI moves beyond data centres into everyday life, with robotics playing a central role.
Despite the ambitious announcements and focus on new opportunities, Nvidia’s shares remained relatively flat during the keynote. This suggests investors are cautiously evaluating the company’s long-term prospects amid broader market uncertainties. Huang‘s speeches have previously provided the stock with additional momentum.
Are investors looking at an opportunity?
Nvidia stock is cheaper today than it was two months ago. It’s trading at 26 times forward earnings, representing a 23% premium to the information technology sector. However, it’s a lot cheaper than Nvidia’s long-term number. Over the past five years, Nvidia’s price-to-earnings (P/E) ratio has stood at 46 times on average.
Nonetheless, growth expectations for Nvidia remain incredibly strong. Earnings growth is expected to average around 30% annually over the next three to five years. In turn, this gives us a price-to-earnings-to-growth (PEG) ratio of 0.81. According to famous fund boss Peter Lynch, this is a clear sign of undervaluation.
But it’s not that straightforward. The forecasts are bullish and the price targets are bullish too. However, it all comes back to that idea of more efficient AI models. There’s certainly some concern that these models will impact long-term demand for Nvidia’s GPUs. And while many analysts have suggested that this will democratise AI, potentially leading to more demand outside the hyperscaler market, the concerns are weighing on the stock.
Personally, having built a sizeable position in Nvidia, I’m simply holding and watching. However, with robotics potentially leading Nvidia higher in the long run, I’m constantly evaluating my position.
