Late last year, Morgan Stanley issued a forecast suggesting Amazon, Microsoft, Alphabet, and Meta Platforms could spend a combined $300 billion on data center infrastructure and chips to power their artificial intelligence (AI) ambitions during 2025.

A close look at some of the forecasts issued by those companies more recently suggests they might even exceed Morgan Stanley's estimate. A lot of that spending will flow to hardware giants like Nvidia (NVDA -8.69%), Broadcom (AVGO -6.05%), and Advanced Micro Devices (AMD -1.63%), which supply the most advanced data center chips and networking equipment for AI development.

The iShares Semiconductor ETF (SOXX -3.53%) is an exchange-traded fund (ETF) that holds a portfolio of 30 top hardware stocks including the three mentioned above. It's currently down 21% from its record high, which offers investors a rare opportunity to buy it at a discount, and with so much AI hardware spending in the pipeline this year, here's why it might be a good idea to do so.

Image source: Getty Images.

Every top AI hardware stock packed into one ETF

The iShares Semiconductor ETF has a very narrow focus, which is reflected by its small number of holdings. It invests exclusively in U.S. companies that design, manufacture, and distribute chips, but especially those that will benefit from powerful trends like AI.

Highly concentrated ETFs such as this one can be a double-edged sword for investors. On the one hand, they can deliver spectacular returns when they are in favor, but on the other hand, they can underperform the rest of the market if investors pull back from the underlying theme (in this case, the semiconductor industry). As a result, investors should own the iShares ETF as part of a diversified portfolio of other funds and individual stocks.

With that said, the ETF offers an incredible amount of exposure to the AI hardware spending boom. Despite holding 30 stocks, its top five positions account for 36.4% of the total value of its portfolio, and they are some of the biggest names in the industry:

Data source: iShares. Portfolio weightings are accurate as of Feb. 27, 2025, and are subject to change.

Nvidia designs the most advanced graphics processing units (GPUs) for data centers, which are the key pieces of hardware when it comes to developing AI models. The company recently announced its fiscal year 2025 financial results, and it delivered a record $130.5 billion in total revenue, which was a 114% increase from the prior year. Demand is unlikely to slow anytime soon, because CEO Jensen Huang says new AI models that are capable of reasoning (or "thinking") can require 100 times more compute than previous generations.

Each of the four tech giants I mentioned at the top (Amazon, Microsoft, Meta, and Alphabet) are big buyers of Nvidia's chips. However, some of them are also making their own chips in partnership with Broadcom, which assists in the design and manufacturing processes. Broadcom is also a leading supplier of AI data center networking equipment, like its industry-leading switches, which regulate how fast data travels between chips and devices.

Advanced Micro Devices has emerged as a competitor to Nvidia in the data center space with its own lineup of AI GPUs. The company is now sending samples of its new MI350 to customers, which is expected to rival Nvidia's latest Blackwell-based GB200.

However, AMD is also the industry's top supplier of AI chips for personal computers, which could be a huge growth area in the future because AI models are quickly becoming efficient enough to run on devices without the help of external data centers. Qualcomm is another emerging player in this space with its Snapdragon chips, which are used in Microsoft's Surface laptops and Samsung's Galaxy smartphones.

Outside of its top five positions, the iShares ETF holds other popular AI semiconductor stocks like Micron Technology, which makes memory and storage solutions, and Taiwan Semiconductor Manufacturing, which is the world's largest chip fabricator.

The iShares ETF has a track record of generating strong returns

The iShares ETF is down almost 4% in 2025 so far, and it's down 21% from its record high, which was set in the middle of last year. However, it has delivered a compound annual return of 11.1% since its inception in 2001, beating the average annual gain of 8.4% in the S&P 500 over the same period.

In other words, the ETF has successfully navigated several tech booms driven by the internet, enterprise software, the smartphone, and cloud computing over the last two decades, so the recent dip might be a buying opportunity.

Based on its top holdings, the ETF is already positioned for the AI boom, which could last for years. Morgan Stanley's estimate for 2025 might only be the start because, in early 2024, Huang said he expects data center operators to invest $1 trillion over five years to upgrade their infrastructure to meet demand from AI developers.

A lot of that spending will be absorbed by the 30 companies that make up the iShares ETF, so it might be the ultimate investment for those looking to capture the subsequent value creation.

However, as I mentioned earlier, it's important for investors to buy the ETF as part of a diversified portfolio. If companies struggle to monetize AI software over the next couple of years, they might pull back on their hardware spending, which would lead to a period of underperformance for stocks like Broadcom, Nvidia, AMD, and more.