The artificial intelligence (AI) boom has been a big growth driver for the semiconductor industry. However, it may just be getting started. According to OpenAI CFO Sarah Friar, the AI infrastructure buildout may make the internet buildout look capex-light and we’re just at the beginning of the ramp up.
Let’s look at three breakout growth stocks that should benefit from this trend.
Advanced Micro Devices (NASDAQ: AMD) has long lived in Nvidia‘s (NASDAQ: NVDA) shadow, but a potential shift in the AI market could create a big opportunity. The first phase of AI has been about training massive models, where Nvidia’s CUDA software ecosystem gave it a big edge. However, as the industry pivots toward inference, AMD is carving out an important niche.
Inference is eventually expected to grow to become much larger than training, and cost-per-inference becomes a much more important factor. One of the world’s largest AI companies is already using AMD’s graphics processing units (GPUs) for a big chunk of its inference traffic, while 7 of the top 10 AI companies now use its GPUs.
AMD is also part of the UALink Consortium, which is looking to break the grip of Nvidia’s proprietary NVLink standard. NVLink is a high-speed interconnect system that links Nvidia’s GPUs to allow them to act as a single unit. It’s another one of the chipmaker’s big advantages. However, if the consortium can create an open interconnect standard, it would allow customers to more easily mix and match vendors in AI clusters.
Beyond GPUs, AMD’s central processing units (CPUs) have been steadily gaining share in the data center space, while its gaming and PC chips have also been seeing strong growth. At the end of the day, AMD doesn’t need to topple Nvidia to thrive. With inference demand accelerating and the company looking like it could take some market share, AMD is well-positioned for long-term growth.
Broadcom (NASDAQ: AVGO) is another semiconductor company with big AI opportunities in front of it. Its core business is networking components, where its Ethernet switches and optical interconnects are used to move huge amounts of data across AI clusters. As these AI clusters continue to get bigger and bigger, networking needs scale with them. Last quarter, Broadcom’s networking components business saw its revenue surge 70%, demonstrating the company’s strong position in this area.