Broadcom (AVGO), the Backbone of AI: A Case for the New Magnificent Seven

How AVGO Passed Meta and Tesla

Amid the current AI revolution and arms race, Magnificent Seven tech companies like Nvidia (NVDA), Apple (AAPL), Microsoft (MSFT), Alphabet (GOOG), and Meta (META) have been getting the bulk of the attention, both in the media and in the market. This attention is deserved - these companies are at the forefront of AI development, after all - but there is one company that has silently benefitted from the AI revolution, perhaps more than any other: Broadcom (AVGO). 

 

Over the past year, Broadcom has vaulted into the world’s top-10 most valuable companies, currently sitting firmly in the no. 6 spot globally. This puts their valuation ahead of Meta and Tesla, with a P/E ratio half that of Tesla, and a market cap growing ever closer to $2 trillion. 

 

This climb obviously isn’t just hype-driven; if I were to take a random poll of 100 full-time traders right now asking what stock they are most hyped about, I would wager that less than 10 of them would say Broadcom. Rather, this climb has been fueled by real revenue and deal execution. 

 

Broadcom closed their acquisition of VMware in late 2023, and then posted revenue of $51.6 billion for the full 2024 fiscal year, as VMware boosted their infrastructure software segment. This firmly asserted infrastructure software as a second revenue engine alongside semiconductors, and that blend of cash-rich chips plus sticky software has helped support durable margins and cash flow. 

 

AI’s Plumbing: Where AVGO Wins

When people think about AI development and implementation, their minds go straight to semiconductors and GPUs. While these hardware elements are critical to the AI ecosystem, they are only half of what’s required; robust software is also a necessity for AI, and that’s where Broadcom comes in. 

 

Ultra-fast networks to move data between thousands of chips are Broadcom’s wheelhouse. In 2025, the company began shipping Tomahawk 6, the first-ever 102.4-TBPS Ethernet switch silicon designed for 800G era fabrics. Since the release of Tomahawk 6, the company has also found success rolling out end-to-end AI networking pieces around it. 

 

Furthermore, Broadcom’s Jericho3-AI fabric can wire together enormous GPU clusters (up to 32,000 accelerators), positioning Ethernet as a credible alternative to Nvidia’s InifiniBand for training-class workloads. Broadcom has continued to extend that family with the development of Jericho4 to stitch clusters across sites. 

 

As if the switches and fabrics weren’t enough, Broadcom is also pushing AI NICs, like their 800G Thor Ultra, which aligns with emerging Ultra Ethernet standards. It has deep custom-silicon relationships that matter in AI; most notably long-running work on Google’s TPU program, which industry experts believe will continue expanding through 2026. 

 

Essentially, when hyperscalers build the next wave of AI capacity, it falls to Broadcom to provide the “plumbing,” like the chips and systems that keep those pricey GPUs fed with data. Broadcom’s logo never shows up on the front of AI servers, but their tech is used on the backend of the majority of AI buildouts. 

 

Why AVGO Belongs in the New Mag-7

Despite the excellent run Broadcom has seen to this point, it shows no signs of slowing down anytime soon. There are three major factors that make a compelling case for Broadcom headed into 2026 and beyond; a visible AI roadmap, their two distinct revenue engines, and custom silicon momentum. 

 

Potentially the most compelling case for Broadcom moving forward is just how clear their map to continued growth is. The company’s cadence, with Tomahawk 6 at 102.4 TBPS, Jericho3 and Jericho4 fabrics, and 800G NICs, aims directly at the bottlenecks AI operators talk about most; bandwidth, congestion, and scale across and between datacenters. When you hear in media that AI companies are planning to increase their capital expenditure in 2026, Broadcom is who they will be spending that money on. 

 

The second major bullish variable for Broadcom is their two distinct revenue engines; chips and software. VMware gives Broadcom a second profit stream with recurring software tied to enterprise infrastructure. This compliments their pre-existing revenue stream from the chip side, which is cyclical in nature, and can serve to support sturdy free cash flow through those cycles. 

 

The last bit of momentum for Broadcom is the sector-wide momentum behind custom silicon. Industry reporting continues to point to Broadcom as the go-to partner for hyperscaler accelerators and AI ASIC programs, which is a wedge that can grow alongside standard Ethernet wins. 

 

When you stand back and look at the full picture, Broadcom looks less and less like an under-the-radar chip name, and more and more like core infrastructure in the AI era that deserves a spot alongside Nvidia when discussing companies integral to the AI revolution.

Gus Downing is host of the tastylive Network show Risk and Reward. @GainsByGus
For live daily programming, market news and commentary, visit tastylive or the tastylive YouTube channel.
Trade with a better broker, open a tastytrade account today. tastylive Inc. and tastytrade Inc. are separate but affiliated companies.