Broadcom is set to see huge growth from networking and custom AI chips.
AMD can ride two powerful trends with inference and agentic AI.
With Warner Bros. Discovery in the process of being acquired, I sold my shares in the company and instead have put money into these two top artificial intelligence (AI) names: Broadcom (NASDAQ: AVGO) and Advanced Micro Devices (NASDAQ: AMD). Nvidia has ridden the first wave of the AI boom to great heights, but the next wave promises a new set of big winners. Broadcom and AMD look very well positioned for the next AI supercycle just getting started.
Broadcom is set to ride two powerful trends emerging in AI infrastructure. The first is the increasing importance of networking. Although large language model (LLM) training and inference are all about speed, it is no longer about how fast a single chip can operate, but how quickly a large cluster of chips can work together.
Will AI create the world's first trillionaire? Our team just released a report on the one little-known company, called an "Indispensable Monopoly" providing the critical technology Nvidia and Intel both need. Continue »
And the more AI accelerators that are strung together, the more important the networking components that tie them together become. Without high-speed networking to reduce latency, companies would not get much added benefit from having huge chip clusters.
Broadcom is the leader in data center networking, offering a variety of components that help connect servers and networks to quickly transfer data. The centerpiece of its networking portfolio is its Tomahawk Ethernet solution, which is one of the gold standards for switching and the backbone of many hyperscalers' data centers. With clusters set to reach 1 million chips, this is a fast-growing business.
The other big trend that Broadcom is riding is custom AI accelerators. The company is the leader in application-specific integrated circuit (ASIC) technology, where it provides some of the intellectual property and know-how to turn its customers' chip designs into physical chips that can be manufactured at scale.
It helped Alphabet develop its popular tensor processing units (TPUs). These chips are a big growth driver for the company, given the strong demand not only from Alphabet but also from outside customers like Anthropic, which has started to deploy the chips.
Broadcom also has deals with OpenAI and Meta Platforms to help them design their own custom chips. It has projected it will see $100 billion in custom chip sales alone in fiscal 2027, and this looks like just the start.
Image source: Getty Images.
Like Broadcom, AMD is also at the convergence of two important trends: inference and agentic AI. AMD's upcoming MI450 graphics processing unit (GPU) is expected to be a big leap for the company, especially for inference.
The company is using a chiplet design, which allows for significantly more memory. This can not only reduce costs but also improve inference performance. AMD's ROCm software platform has also improved significantly, and the company is benefiting from most new AI code being written with open-source frameworks.
Management has already partnered with OpenAI and Meta, with each having committed to 6 gigawatts' worth of its new GPUs. Both companies will also take a stake in AMD through warrants and incorporate its ROCm software into their workflows. The issuance of warrants came at a price, but it was an important step for the company to gain access to two of the largest AI infrastructure spenders.
AMD also has a huge opportunity in data center central processing units (CPUs). Server racks dedicated to agentic AI will need a much higher CPU-to-GPU ratio, as CPUs will be needed to provide the sequential logic and handle the real-world interactions required to manage AI agents. The company is the leader in this space, and it is developing high-performance CPUs specifically to handle the demands of agentic AI.
Through its acquisition of ZT Systems, AMD will soon be able to provide end-to-end full rack systems designed specifically for inference and agentic AI, making it a much more formidable AI infrastructure player.
The AI infrastructure market is shifting, and Broadcom and AMD look poised to be two of the biggest winners. As such, the stocks look like two great options right now.
Before you buy stock in Broadcom, consider this:
The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Broadcom wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.
Consider when Netflix made this list on December 17, 2004... if you invested $1,000 at the time of our recommendation, you’d have $581,304!* Or when Nvidia made this list on April 15, 2005... if you invested $1,000 at the time of our recommendation, you’d have $1,215,992!*
Now, it’s worth noting Stock Advisor’s total average return is 1,016% — a market-crushing outperformance compared to 197% for the S&P 500. Don't miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.
See the 10 stocks »
*Stock Advisor returns as of April 17, 2026.
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, Nvidia, and Warner Bros. Discovery. The Motley Fool has a disclosure policy.