3 Stocks to Buy on the AI Infrastructure Sell-Off

While the market has been driven higher by artificial intelligence (AI) stocks over the past few years, these stocks have recently pulled back, creating a potential buying opportunity. The fear is that spending could eventually start to slow.

But right now, hyperscalers (owners of large data centers) have indicated that they are getting strong returns on their investments and that they continue to plan to spend big money building out AI data centers.

Let's look at three AI stocks to buy on this dip.

Image source: Getty Images.

Nvidia: The original AI play

No company has benefited more from the AI infrastructure build-out than Nvidia (NASDAQ: NVDA), which has grown to become the world's largest company by market cap. Its graphics processing units (GPUs) are the primary chips used to train AI models, and its CUDA software, where most foundational AI code has been written, gives it a wide moat in this area.

However, the Nvidia of today is more than just GPUs; it has become a complete AI infrastructure player. It has a strong networking portfolio, which has been one of its fastest-growing areas, while its "acquisition" of Groq brought chips specifically for inference. It's also dived into the data center central processing unit (CPU) market. Together, this now allows it to offer complete systems for specific AI tasks, which should position it for continued strong growth well into the future.

AMD: The AI upstart

Advanced Micro Devices (NASDAQ: AMD) may have lost to Nvidia when it came to AI model training, but the company looks well positioned for inference and agentic AI.

Inference is eventually expected to become larger than training, and AMD's GPUs are much better suited for this task. Inference is much more memory-bound than compute-bound, and AMD's chiplet design allows it to be packaged with more memory. Its recent acquisition of MEXT and its memory optimization technology, and ZT Systems, meanwhile, will allow it to offer complete systems designed for inference. The company already has two large inference partnerships, which should be a major area of growth in the coming years.

On top of that, the company is positioned to be a major beneficiary of agentic AI, which will require a significant increase in the use of CPUs in AI data centers. This will boost AMD, which is a leader in the space and has been taking market share away from rival Intel. AMD is already developing CPUs specifically for agentic AI and sees this growing to be a $120 billion addressable market over the next few years.

Micron: A memory winner

The memory market continues to explode higher, and as one of the big three DRAM (dynamic random access memory) makers, Micron Technology (NASDAQ: MU) continues to be a huge beneficiary. The company just posted incredible numbers for its fiscal third quarter, with revenue surging from $9.3 billion a year ago to $41.5 billion and gross margin expanding to 84.6% from 37.7%.

The demand for high-bandwidth memory (HBM), which is a special form of DRAM packaged with GPUs to optimize performance, remains insatiable, with Micron's supply sold out for 2027 and into 2028. It expects HBM to be a $100 billion market next year. Meanwhile, it has now signed long-term strategic customer agreements that include both HBM and NAND (flash memory) that are non-cancellable, take-or-pay arrangements with annual volume commitments, with some extending until 2030. This should help reduce some of the typical cyclicality of the memory business.

Micron is working to continue increasing capacity, although the market is expected to remain supply constrained for the foreseeable future. Between its SCAs and supply-demand dynamics, this makes the stock a good one to buy after the recent pullback.

Geoffrey Seiler has positions in Advanced Micro Devices. The Motley Fool has positions in and recommends Advanced Micro Devices, Intel, Micron Technology, and Nvidia. The Motley Fool has a disclosure policy.