Semiconductor stocks are still one of the best ways to play the AI boom.
Artificial intelligence (AI) is driving one of the biggest technology shifts in decades. Meanwhile, the companies helping build out the infrastructure to run AI workloads are seeing some of the strongest growth. That combination is creating some valuable investment opportunities for those with the funds available to buy in right now.
Let’s look at three top AI infrastructure stocks to buy at the moment.
Image source: Getty Images
1. Nvidia
No company has benefited more from the AI infrastructure buildout than Nvidia (NVDA 1.65%). While its graphics processing units (GPU) were initially designed to speed up graphics rendering in video games, hence the name, today its chips are the backbone of AI data centers.
And that is likely to continue well into the future. Why? Because the company formed a huge moat around its business through its CUDA software platform, which it created as a way for its chips to be programmed for tasks outside of their original design.
While the use of GPUs in areas outside of video games was slow to develop, Nvidia smartly gave its software platform away for free, pushing it into research labs and universities where early AI work was being done. As a result, developers learned to program GPUs using CUDA, while also writing a lot of code on top of it.
Nvidia didn’t stop there, though. It also pushed into networking with its proprietary NVLink high-speed interconnect system, which helps connect its GPUs in order for them to act as a single unit. It later acquired Mellanox to help ensure that its chips can scale efficiently across massive AI clusters.
This all gives Nvidia a big advantage, which was on full display when it captured a whopping 92% share of the GPU market Q1. With AI infrastructure spending continuing to ramp up, Nvidia is a top AI stock to own.
2. Broadcom
While Nvidia dominates GPUs, Broadcom (AVGO 1.48%) has carved out its own powerful AI growth story. Its networking components — including Ethernet switches, optical receivers, and DSPs — are essential for moving massive amounts of data across AI clusters and compete with Mellonox’s InfinBand technology. As clusters grow in size, networking demand scales with them. That dynamic has already driven huge growth, with its AI networking revenue surging 70% last quarter.
Broadcom’s bigger opportunity, however, lies in custom AI chips. Nvidia’s GPUs are expensive, and some hyperscalers (owners of massive data centers) have turned to custom AI chips designed for specific tasks, such as inference, to help improve performance and lower costs. Broadcom is helping many of these companies develop their custom chips, given its expertise in application-specific integrated circuits (ASICs).
It helped Alphabet design its highly regarded tensor processing units (TPUs), which led to more custom chip business. It’s now working with several hyperscalers on new designs, with the company saying its three customers furthest along in the process plan to deploy 1 million chip clusters by its fiscal 2027. That alone represents a $60 billion to $90 billion opportunity, before factoring in newer customers like Apple. While custom AI chips are likely not going to be as large a market as GPUs, given the high initial costs to design the chips, they are typically used in large-scale deployments.
Broadcom’s story also doesn’t stop at hardware. Through its VMware unit, the company is also becoming an important player in AI cloud computing infrastructure. VMware’s Cloud Foundation helps enterprises manage AI workloads across hybrid and multi-cloud environments, and adoption has been accelerating.
Between its networking portfolio, custom chip opportunities, and VMware’s role in enterprise AI, Broadcom has multiple growth levers that should drive strong results for years.
3. Taiwan Semiconductor Manufacturing
Taiwan Semiconductor Manufacturing (TSM 2.58%) is the world’s most advanced foundry and the go-to manufacturer for advanced chip designers, including Nvidia and Broadcom. Unlike chip designers, TSMC wins no matter which company comes out on top, since it handles the manufacturing of nearly all of their most advanced chips. That makes it one of the best ways to play the AI infrastructure boom.
TSMC’s strength lies in its unmatched scale and leadership in advanced node chip production. The denser a chip, the more powerful and power-efficient it becomes, which is why the semiconductor industry is continually trying to shrink node sizes. While competitors have struggled with yields on smaller-sized nodes, chips built on 7-nanometer and smaller nodes now account for nearly three-quarters of TSMC’s revenue. 3nm nodes already represent nearly a quarter of its revenue, and it will soon push into 2nm nodes.
As the only foundry that can manufacture advanced chips at scale, TSMC has strong visibility into growth, as it works with its largest customers to increase capacity to meet their growing demand. Management expects AI chip demand to grow at an over 40% compound annual rate (CAGR) through 2028, providing a multiyear tailwind. Its competitors’ struggles also give it some nice pricing power.
As an integral part of the semiconductor value chain, TSMC is poised to continue to ride the AI infrastructure buildout wave.
Great Job newsfeedback@fool.com (Geoffrey Seiler) & the Team @ The Motley Fool Source link for sharing this story.
