While some investors tend to index on names like Nvidia and Advanced Micro Devices, Wall Street is increasingly building a bullish narrative around another semiconductor leader: Broadcom (AVGO 1.12%).
As cloud hyperscalers continue to increase their capital expenditure (capex) budgets, Broadcom is quietly becoming a key enabler of the AI infrastructure revolution. Let's dive into how Broadcom is swiftly emerging as an important player powering the ongoing data center buildout boom.
Image source: Getty Images.
Broadcom is the nervous system for AI data centers
Broadcom's primary role within AI data centers revolves around its high-performance networking gear. As AI applications evolve into more complex utilities beyond chatbots, compute capacity is no longer the single biggest bottleneck straining workloads.
Rather, the means by which data flows between graphics processing units (GPUs), servers, and storage systems is becoming a mission-critical issue. Broadcom's position along the AI chip value chain sits squarely between switching and networking silicon. Broadcom's Ethernet and switching equipment help move massive data sets with low latency.
As AI workloads scale, Broadcom is uniquely positioned to complement existing GPU clusters within broader data center architectures. In essence, Broadcom collects a royalty as AI infrastructure buildouts accelerate, regardless of which compute architecture is preferred.
NASDAQ: AVGO
Key Data Points
Hyperscale workloads are transitioning to custom silicon
One of the lesser-talked-about topics with hyperscalers is that cloud providers, such as Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform, are trying to identify ways to lower their overall cost of compute while also gaining greater control over their AI stack.
As part of this performance optimization strategy, many developers are beginning to design their own custom application-specific integrated circuits (ASICs). Broadcom currently works with Meta Platforms, Apple, ByteDance, and Alphabet on custom silicon solutions.
Broadcom's ability to design sophisticated, high-volume processing custom chips allows the company to acquire incremental market share from general-purpose GPU suppliers. Importantly, this transition to complement custom silicon with existing GPU clusters is not a fleeting decision.
Rather, Broadcom is quietly becoming more deeply embedded within multiyear hyperscale infrastructure playbooks. In other words, Broadcom's custom silicon relationships should scale alongside big tech's rapidly expanding compute needs.
From Wall Street's point of view, Broadcom's business is becoming evermore tied to secular tailwinds fueling AI infrastructure, as opposed to less predictable, cyclical hardware upgrades.
The capex supercycle is a multiyear tailwind
According to industry data compiled by FactSet Research, big tech is forecast to spend at least $500 billion on AI capex this year. Taking this a step further, McKinsey & Company suggests that a grand sum of $6.7 trillion will be spent on AI infrastructure through 2030 -- with the largest component allocated to serving AI workloads.
This drives home the point that, as AI model training and inference deployment accelerate, GPUs and AI accelerators are not the only pieces of hardware positioned for explosive growth. Broadcom has a unique ability to monetize the current capex supercycle, given its exposure to networking, interconnects, storage, and custom silicon.
This optionality gives Broadcom a structural advantage across various pockets of the AI chip ecosystem in relation to its peers. While it may not fetch as much attention as its counterparts, Broadcom is positioned to generate robust, compounding growth throughout the AI infrastructure era.
To me, Broadcom is a compelling buy-and-hold opportunity for investors seeking durable growth beyond the obvious AI stocks.
