3 AI Chip Stocks to Buy as the Sell-Off Continues

While temperatures have been scorching across much of the U.S., AI chip stocks have cooled off, with the sector facing increased selling pressure this summer. The dip appears largely centered on concerns that AI infrastructure spending could slow. However, hyperscalers have largely indicated that their capital expenditures will only increase next year, and Bank of America recently projected that worldwide, cloud and AI data center capex will jump by 40% to 50% year over year to around $1.5 trillion in 2027.

With data center infrastructure spending still booming, this pullback could be a great chance to scoop up these three AI semiconductor stocks.

1. Nvidia: Still leading the way

Nvidia (NASDAQ: NVDA) remains the king of AI infrastructure, and following the pullback, it trades at just 15 times analysts' estimated earnings for its fiscal 2028 (which ends in January 2028). That's a bargain for a company that is still delivering rapid revenue growth, including 85% growth last quarter.

What I really like about Nvidia, though, is how the company has quietly transformed itself into a complete AI infrastructure package. The company's graphics processing units (GPUs) remain its biggest revenue driver, and its ubiquitous CUDA software platform provides a wide moat for its chips in AI model training.

However, the company also has a top-notch networking portfolio; its "acquisition" of Groq gave it a chip designed specifically for inference; and it's diving headfirst into the data center central processing unit (CPU) market, which is set to boom as the use of agentic AI takes off. This lets it offer complete end-to-end systems for specific AI tasks and should help drive continued strong growth.

At its current valuation, Nvidia is a stock to own.

2. AMD: Riding inference and agentic AI trends

Advanced Micro Devices (NASDAQ: AMD) stock has been hot this year, but it, too, has pulled back from its highs. The exciting thing about AMD is that it is riding two of the most powerful trends in AI right now.

The first is inference, where its GPUs compete well against Nvidia's offerings, given their chiplet design, which allows them to be packaged with more memory. AMD has formed partnerships with OpenAI and Meta Platforms, and big orders for its newest GPUs should begin shipping soon. It's also been reported that it may have a deal with Anthropic.

On top of that, AMD is a leader in data center CPUs.

Because CPUs are the right hardware for managing AI agents, the number of CPUs used in AI data centers is expected to skyrocket. Where previously, the ratio of GPUs to CPUs in AI data center servers built for training stood at 8:1, experts foresee that ratio evolving to 1:1 with infrastructure designed to support agentic AI. AMD sees the total addressable market for data center CPUs growing at a 35% annualized rate to $120 billion by 2030.

Between its GPU and CPU opportunities, AMD looks poised for strong growth.

Image source: Getty Images.

3. Broadcom: The custom chip leader

Broadcom (NASDAQ: AVGO) is another company on the verge of explosive revenue growth that has been caught in the AI infrastructure sell-off. The pullback in the stock has taken its valuation down to just 19 times analysts' earnings estimates for its fiscal 2027 (which ends in November 2027).

However, the company should see its custom chip revenue surge to well over $100 billion next year. That's more than the nearly $64 billion in total revenue it generated last year and five times the AI revenue it produced. The company's custom chip business is taking off with the success of Alphabet's Tensor Processing Units (TPUs), which it helped the search leader develop. That has led to other hyperscale customers turning toward its ASIC (application-specific integrated circuit) services to help them develop custom AI chips.

Broadcom has also been a leader in data center networking. This is a fast-growing business that can also tie directly into its custom chip business. Given the company's growth prospects, the stock is just too cheap at these levels.

Bank of America is an advertising partner of Motley Fool Money. Geoffrey Seiler has positions in Advanced Micro Devices, Alphabet, Broadcom, and Meta Platforms. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Broadcom, Meta Platforms, and Nvidia. The Motley Fool has a disclosure policy.