How DeepSeek’s AI Model Impacts Nvidia Stock

A few weeks ago, Chinese AI research lab DeepSeek released its open-source AI model, DeepSeek-R1, which has drawn significant attention in the tech world. According to a paper authored by the lab, the DeepSeek-R1 model outperforms cutting-edge models such as OpenAI’s o1 and Meta’s Llama AI models across multiple benchmarks. This is impressive, given that the model is also open-source, cost-effective, and requires significantly less computational power compared to its rivals. Now DeepSeek’s approach appears to have set off alarm bells in Silicon Valley, where most big tech giants have been relying more on brute force – amassing a larger stock of GPU chips and servers and running long model training periods. We believe this development could potentially have implications for Nvidia stock (NASDAQ:NVDA), the dominant player in the AI hardware space. Also, check out our analysis on How Nvidia Stock Could Drop 50%

DeepSeek’s Innovations

DeepSeek has reportedly restructured the foundation of AI models, emphasizing software-driven resource optimization over hardware dependency. Although the company apparently utilizes tens of thousands of Nvidia’s H100 and H200 AI GPUs to train its models, it has faced constraints due to U.S. export controls limiting access to the latest chips. To overcome this, DeepSeek has implemented innovative engineering tweaks, such as custom communication schemes between chips to improve data transfer efficiency, memory-saving techniques, and reinforcement learning methods to minimize computational power requirements. These optimizations result in drastically lower costs compared to traditional large language models. This cost efficiency is reflected in the API pricing for DeepSeek-R1, which costs just $0.55 per million input tokens and $2.19 per million output tokens—significantly undercutting OpenAI’s API rates of $15 and $60, respectively.  However, it remains unclear how quickly DeepSeek can expand its reach. The company’s commercial ambitions could face challenges due to the U.S. chip ban. More importantly,  geopolitical tensions between the U.S. and China could create trust issues for companies considering using Chinese-built large language models. These factors could limit DeepSeek’s penetration in Western markets. Separately, if you want upside with a smoother ride than an individual stock, consider the High Quality portfolio, which has outperformed the S&P, and clocked >91% returns since inception.

Impact On GPU Makers Like Nvidia

That being said, we believe that DeepSeek’s advancements could prompt a moment of reckoning for big tech companies. DeepSeek’s resource-efficient methods could force a reconsideration of brute-force AI strategies that rely on massive investments in computing power. Nvidia has been the largest beneficiary of this approach through the AI boom, with its GPUs regarded as the best performing for training and deploying AI models. Over the past two years, companies have funneled massive resources into building  AI models, driving Nvidia’s revenue up by over 125% in FY24 to $61 billion, with net margins nearing 50%. If the industry begins to take inspiration from the methods DeepSeek uses in its open-source models, we could very well see demand for AI Computing power cool off.  The underlying economics of the broader AI ecosystem have been weak in the first place, and most of Nvidia’s customers likely aren’t generating meaningful returns on their investments. This could accelerate the shift toward more cost-effective, resource-optimized AI models.

Now the increase in NVDA stock over the last 4-year period has been far from consistent, with annual returns being considerably more volatile than the S&P 500. Returns for the stock were 125% in 2021, -50% in 2022, 239% in 2023, and 171% in 2024. The Trefis High Quality (HQ) Portfolio, with a collection of 30 stocks, is considerably less volatile. And it has comfortably outperformed the S&P 500 over the last 4-year period. Why is that? As a group, HQ Portfolio stocks provided better returns with less risk versus the benchmark index; less of a roller-coaster ride as evident in HQ Portfolio performance metrics. Given the current uncertain macroeconomic environment around rate cuts and multiple wars, could NVDA face a similar situation as it did in 2022 and underperform the S&P over the next 12 months – or will it see a strong jump?

We see a possibility that the “fear-of-missing-out” driven AI wave seen over the last two years could ease off due to diminishing incremental performance gains from larger models and also as the availability of high-quality training data becomes a bottleneck. This shift toward more efficient models could compound the impact of a potential slowdown for GPU makers such as Nvidia. Moreover, Nvidia also faces mounting competition from the likes of AMD as well as its own customers such as Amazon, who have been focusing on developing and deploying their own AI chips. While Nvidia does have a comprehensive software ecosystem around its AI processors, including programming languages that should help it better lock customers into its products, the company could face pressure. Nvidia’s premium valuation may not fully reflect these risks. We value Nvidia stock at about $93 per share, roughly 35% below the current market price. See our analysis of Nvidia valuation: Expensive or Cheap.

[1] Returns as of 1/27/2025
[2] Cumulative total returns since the end of 2016

Invest with Trefis Market-Beating Portfolios
See all Trefis Price Estimates