Like much of the stock market, Nvidia (NVDA -0.64%) has experienced significant volatility in 2025. Even after recovering some of its losses, the stock remains down 15% year to date. Investors are concerned that President Donald Trump’s tariffs could reduce demand for the company’s data center chips, which are the best in the industry for developing artificial intelligence (AI) applications.

Although semiconductors are exempt from the most aggressive tariff policies, many of Nvidia’s customers still face increased costs and potential sales declines, which could force them to reduce their capital expenditures (capex).

Meta Platforms (META 0.34%), Microsoft (MSFT 0.15%), Alphabet (GOOG 0.14%) (GOOGL 0.12%), and Amazon (AMZN -1.95%) are four of the biggest buyers of Nvidia’s AI chips, and they just gave Nvidia investors a positive update on their planned AI spending for this year.

Image source: Nvidia.

Nvidia’s chips are the gold standard in AI

Nvidia’s H100 graphics processing unit (GPU) was the industry’s dominant AI data center chip throughout 2023 and for most of 2024. It was based on the company’s Hopper architecture, which has been superseded by the higher-performing Blackwell and Blackwell Ultra architectures. In fact, the new Blackwell Ultra GB300 GPU can perform AI inference up to 50 times faster than the H100 in specific configurations, which is important for developers of next-generation “reasoning” models.

Traditional large language models (LLMs) deliver one-shot responses, providing users with fast and convenient access to information, but they often make mistakes. Reasoning models spend more time “thinking” in the background to clear up as many errors as possible before generating a response. This means they make better use of the data they already have, which reduces pre-training workloads (endlessly feeding more data into models to make them “smarter”).

But since reasoning models use up more tokens (words, symbols, and punctuation) during the thinking process and are slower to generate responses, Nvidia CEO Jensen Huang says they need up to 100 times more computing power than traditional models to maintain a convenient user experience. Blackwell Ultra chips (which will ship to customers in the second half of 2025) are a step in the right direction, but developers are already eyeing Nvidia’s next-generation Rubin GPUs, which are expected to deliver a further 3.3 times more compute performance.

Rubin GPUs will pave the way for the most powerful reasoning models to date, and they are slated for release in 2026. Simply put, Nvidia is investing heavily in innovation to continue driving the AI industry industry forward for the long term, which is why investors are so anxious for proof that demand is holding up.

Meta, Microsoft, Alphabet, and Amazon delivered good news for Nvidia investors

Nvidia is an American company, but the majority of its chips are fabricated offshore by Taiwan Semiconductor Manufacturing, so they are technically imported products. However, Trump exempted semiconductors from his “Liberation Day” tariffs on imports, as he understands the importance of keeping the U.S. at the forefront of AI technology.

The bigger concern is the effect of tariffs on Nvidia’s customers. They are spending tens of billions of dollars per year on AI chips and data center infrastructure, so any slowdown in their core operations could force them to pull back. Tariffs apply to physical goods, so a company like Amazon will take a hit because it draws part of its revenue from its e-commerce platform, which imports retail products from other countries.

However, Amazon’s cloud services platform, digital advertising business, and streaming segment aren’t directly subjected to tariffs because they offer digital products and services. Other Nvidia customers like Meta, Microsoft, and Alphabet also primarily sell digital products and services, so they could weather the global trade tensions better than most companies.

As a result, they each delivered reassuring news for Nvidia investors when they reported their financial results for the first quarter (ended March 31):

• Meta raised its 2025 capex forecast range to $64 billion to $72 billion (from $60 billion to $65 billion previously).
• Microsoft held its capex forecast steady, so it’s on track to spend around $80 billion during its fiscal 2025 (ending June 30).
• Alphabet didn’t change its capex forecast with $75 billion in spending planned for 2025.
• Amazon also left its capex forecast unchanged, so it’s still likely to spend around $105 billion this year.

Nvidia stock looks like a bargain right now

The dip in Nvidia stock has created a fantastic buying opportunity for investors too. The stock currently trades at a price-to-earnings (P/E) ratio of 39, which is a significant discount to its 10-year average and median, both of which are above 50.

Data by YCharts.

Nvidia generated $115.2 billion in data center revenue during its fiscal 2025 (ended Jan. 26), which was an eye-popping 142% increase from the prior year. But Huang predicts that data center spending will top $1 trillion annually by 2028 as reasoning models demand even more computing power, so the company still has a long potential growth runway.

Data center operators like Meta, Microsoft, Alphabet, and Amazon typically plan their infrastructure spending years in advance, even if they only offer guidance for the coming 12 months. Therefore, they might be willing to look past any short-term slowdown in the economy from the tariffs and global trade tensions. That’s likely why they haven’t reduced their capex forecasts.

Plus, demand for Nvidia’s chips is outstripping supply, so companies can’t afford to cancel orders because they would risk falling behind in the AI race. As a result, Nvidia stock looks like a great buy at the current price, especially for investors who are willing to stay the course until 2028 and beyond.