For a brief while, Broadcom’s $22.2 billion quarter report in late spring seemed like the kind of figure that should make the stock unstoppable. Revenue has increased by 48% annually. 67% is a record operating margin. $10.3 billion in free cash flow in just one quarter. This was a dominant performance by all reasonable standards. After that, the stock fell.
That is the current Broadcom paradox. The price of AVGO’s stock is currently at about $382, down from a 52-week high of $495 that was only a few weeks ago in early June. It seems that markets were looking for something different, or more accurately, assurance that the growth narrative was continuing. The stock lost some of its gains almost immediately after management made a statement that investors didn’t particularly like. The issue wasn’t the number per se. It was the tone.
Over the past eighteen months, there has been a noticeable shift in the narrative surrounding Broadcom. Unlike Nvidia, this company never dominated headlines. The CEO of Broadcom, Hock Tan, is not the type of executive who seeks attention. He manages a precision-driven business with long-term chip contracts, solid client agreements, and consistent software integration profits. Instead of press conferences, the drama usually takes place in design labs and multi-year partnership agreements. Despite this, Broadcom has a $1.82 trillion market capitalization and generated $10.8 billion in revenue from AI semiconductors in Q2 alone—a 143% increase over the same period last year.
A product mix that most investors still don’t fully comprehend is what’s driving those numbers. In addition to chips, Broadcom also produces XPUs, which are specialized AI accelerators made for the training and inference tasks of the biggest AI labs in the world. Anthropic, OpenAI, Meta, and Google. These aren’t just regular consumers looking to buy parts. They have multi-generation contracts with Broadcom that cover chips that are not yet available. In an industry where demand can reverse in as little as six months, Tan’s statement that visibility now extends through 2028 is uncommon.
The networking aspect is just as significant but receives less attention than it merits. Broadcom also produces the massive-scale chips required to connect any cluster of AI accelerators. With large tech companies investing hundreds of billions in AI infrastructure, data centers are growing quickly, and Broadcom gains value at two different stages of this expansion. Even at a P/E ratio above 63, the company’s current trajectory is difficult to discount due in part to this dual position.
According to some analysts, the price of AVGO stock is expected to rise by about 70% from its current level to $650 by the end of 2028. According to that forecast, revenue will continue to grow at a rate of about 52% per year, and margins will remain close to their current levels before progressively contracting as the cycle of AI spending returns to normal. That estimate might be overly optimistic. It might also be conservative, since the company projects that full-year AI semiconductor revenue will reach $56 billion in 2026 and surpass $100 billion by fiscal 2027.
In general, investors appear to think that the AI infrastructure spending cycle is still years away. Whether Broadcom’s portion of that cycle can still be justified is the question. Nothing in semiconductors is permanently locked, but custom chip relationships are sticky. There are new rivals attempting to get into the XPU market. Sometimes, clients with sufficient scale construct their own silicon. Although it’s still unclear if any of those threats will become significant in the near future, they are still worth keeping an eye on.
For the time being, Broadcom is in a unique position: it is underappreciated for a business of its size, well past its growth inflection, and inexplicably still accelerating. In retrospect, the post-earnings decline seems more like fatigue than uncertainty. Investors haven’t had much reason to leave based on the numbers, at least.
