If I Could Only Buy One AI Stock, This Would Be It todayheadline

Share prices of semiconductor giant Nvidia (NVDA 0.90%) are down by about 5% so far in 2025. Despite solid fundamentals, the company’s stock tanked on marketwide concerns about a potential slowdown in artificial intelligence (AI)-related spending in the coming months. Wall Street panicked after Chinese start-up DeepSeek announced training an open source AI model, DeepSeek-R1, for less than $6 million — significantly lower than the $100 million used to train OpenAI’s GPT-4.

However, many industry experts are now casting doubts on DeepSeek’s claims about training the model with only 2,000 H800 graphics processing units (GPUs) in contrast to 25,000 H100 GPUs for GPT-4. Scale AI CEO Alexander Wang has suggested that DeepSeek has used as many as 50,000 H100 chips but has not disclosed this due to U.S. export controls. While Wang has not provided any evidence, it raises questions about the credibility of DeepSeek’s claims.

However, even if DeepSeek has succeeded in realizing cost efficiencies in the generative AI space, it can eventually prove beneficial for Nvidia. With technology becoming more cost effective and efficient, demand will surge, which in turn will drive up the total addressable market and sales of AI-optimized chips.

Hence, Nvidia seems negatively impacted by disproportionate fears. It may make sense for astute investors to buy this stock on the recent dip. Here are a few reasons why I think it’s a compelling pick in 2025.

Robust financials

Nvidia has been successful recently in demonstrating top- and bottom-line performance at scale. In its fiscal 2025’s third quarter (ended Oct. 27, 2024), revenue soared 94% year over year to $35.1 billion, driven mainly by a 112% year-over-year jump in data-center revenue to $30.8 billion. The company also boasts impressive gross margins in the mid-70s percentage range, although these may be modestly affected in the initial ramp-up period of Blackwell systems.

Thanks to the company’s robust profitability, Nvidia is committed to returning significant value to shareholders. In Q3, the company returned $11.2 billion as dividends and share repurchases.

Finally, management expects revenue to reach $37.5 billion plus or minus 2% in the fourth quarter. The high revenue visibility is attributed to continued demand for Hopper architecture chips and the initial ramp-up of Blackwell systems. With demand for AI chips far outpacing supply, the company is expected to enjoy significant pricing power in the future months.

Blackwell catalyst

Nvidia’s next-generation, end-to-end AI infrastructure solution, Blackwell, is expected to play a pivotal role in maintaining Nvidia’s dominance in the accelerated computing space, especially since AI workloads are becoming increasingly more demanding and complex. This infrastructure solution supports seven different chips, multiple networking offerings, and air-cooled and liquid-cooled data centers.

Furthermore, according to recent MLPerf Training results, Blackwell has demonstrated 2.2 times better performance and 4 times lower computing costs compared to previous Hopper generation chips running the GPT-3 benchmark model. The subsequent cost savings and high flexibility are expected to translate into even higher demand for Blackwell systems across industries and deployment environments.

Nvidia is currently in full production mode for Blackwell systems. The company shipped 13,000 Blackwell GPU samples in Q3 and expects Blackwell revenue to exceed previous estimates of “several billion dollars” in the fourth quarter.

Enterprise AI adoption and physical AI

Enterprises are also increasingly opting for the Nvidia AI Enterprise platform, (which includes NVIDIA NeMo and NIM microservices) as an operating platform for developing Co-Pilots and custom AI agents. Prominent companies such as Salesforce, Cloudera, SAP, and ServiceNow are using this platform to accelerate the development of AI applications.

Subsequently, the Enterprise AI platform is expected to become a major revenue driver in the coming months considering that almost 1,000 companies are already using NVIDIA NIM. Management expects Nvidia AI Enterprise revenue to more than double year over year in fiscal 2025. Nvidia has successfully transitioned from a hardware company to a full-stack AI infrastructure provider.

Nvidia also sees significant opportunities in areas such as industrial AI, autonomous systems, and robotics wherein foundation models have to interact and understand the physical world. The company is gearing up to capitalize on these physical AI opportunities with Nvidia’s Omniverse platform and its new Cosmos technology.

Valuation

Nvidia is trading at about 28 times forward earnings, which may seem expensive by traditional standards. However, several factors justify the seemingly rich valuation. Analysts expect the company’s revenue and earnings per share (EPS) to grow year over year by 112% and 127%, respectively, in fiscal 2025.

The chip giant is well poised to benefit from the two fundamental shifts in global computing: a trillion-dollar upgrade cycle at data centers to shift from CPU-based computing to accelerated computing and the emergence of AI factories producing 24/7 digital intelligence. Finally, despite sustainable competitive advantages, such as around 90% share in the global GPU market and a robust software ecosystem built to support its hardware offerings, the company trades at a price/earnings-to-growth (PEG) ratio of just 0.2.

Thus, considering the company’s future growth potential, Nvidia seems like a smart pick for 2025.