Why Broadcom’s (AVGO) Selloff is Once-in-a-Cycle Opportunity

As the adage goes, an opportunity knocks only once—and it feels like this may be Broadcom’s (AVGO) moment. After a massive rally of nearly 180% since April, following the so-called Liberation Day correction, the stock reached new all-time highs ahead of its Q4 earnings. It then hit a wall, as more skeptical investors saw the setup as an opportunity for a downward re-rating.

At first glance, that reaction looks contradictory. Broadcom delivered a beat across the board and reported a record backlog, further reinforcing revenue visibility at a time when AI infrastructure continues to require massive investment. Still, a mix of growing skepticism around the AI CapEx cycle and concerns over stretched valuations ended up overshadowing the fundamentals in the short term.

Looking beyond the near-term noise—which, in my view, does not point to structural risks in demand for Broadcom’s silicon and AI networking, at least not yet—I see no reason to take my foot off the accelerator on this stock. Investors willing to buy shares at this point of valuation reset are likely getting a good deal, which is why I maintain a Buy rating on AVGO.

Why AI Networking Sits at the Core of the Broadcom Thesis

For those just getting familiar with Broadcom, the company designs, tests, and builds communication silicon—along with switches and routers—that are essential for enabling AI to operate at scale, connecting chip to chip and server to server. Broadcom’s key competitive advantage is its customer base: the world’s largest technology companies, which arguably cannot afford to slow investment in the midst of the AI boom.

In addition, Broadcom operates a software infrastructure business, led by VMware, which accounts for roughly 42% of total revenue and provides a meaningful source of recurring cash flow.

That said, the core of the investment thesis today lies in AI networking, which sits within the semiconductor segment. The company reported a record $73 billion backlog in its Q4 earnings—clear evidence that customer demand remains extremely robust. Importantly, this backlog does not convert into revenue all at once. The real question is how quickly it will be monetized.

More skeptical investors appear to assume that less than $40 billion of this backlog will be converted in 2026. This is implied by total revenue estimates for FY26 of roughly $96.1 billion, which—when broken down by the semiconductor segment’s share—remain well below the reported backlog. Such assumptions reflect a conservative view on delivery pace, rooted in the traditional belief that semiconductors are cyclical and, possibly, an underestimation of the structural growth in AI networking.

CEO Hock Tan, who rarely provides long-term guidance, stated that the $73 billion backlog is expected to be delivered over approximately the next 18 months, while also suggesting that the backlog could continue to grow during that period. Under this framework, it does not seem unrealistic to project that around $48.6 billion of the backlog could be delivered as early as 2026.

Post-Earnings Selloff Misses the Bigger Picture

Since AVGO reached its all-time high on December 10, just ahead of its earnings release, the stock has already corrected by roughly 22% of its market value.

Amid skepticism that the company will actually deliver close to $50 billion of backlog by 2026, one recurring concern has been the risk of “customer tooling.” In short, the fear is that hyperscalers such as Amazon (AMZN) and Google (GOOGL) will increasingly design their own silicon and eventually “cut out” Broadcom.

Amazon’s AWS is often cited as the most advanced example, with Graviton and Trainium. While TTrainium is often viewed as lagging Google’s TPU in performance for certain workloads, the more important point is that AWS remains heavily dependent on Broadcom for networking, including both switches and interconnects. This suggests that even a more vertically integrated hyperscaler does not truly replace Broadcom—it simply concentrates the relationship.

The same logic applies to Google’s TPU, designed by Google and produced at TSMC (TSM), with MediaTek involved as a design partner. Here, the difference in scale is critical. Google’s TPU volumes are generally understood to be in the hundreds-of-thousands range, serving very specific workloads within a closed ecosystem. Broadcom, by contrast, operates at a multi-million-unit scale when combining custom silicon and networking, serving multiple customers across multiple generations. That scale also translates into a much steeper learning curve.

It’s also worth emphasizing that, in semiconductors, scale is the outcome of cost, yield, and performance working together. Companies that lack scale tend to pay more per wafer, achieve lower yields, and iterate more slowly—structural disadvantages that are difficult to overcome.

A Reset in Valuation, Not in Demand

Bringing the market’s concerns together, I believe Broadcom’s sharp post-earnings correction can be explained by the confluence of two main factors.

The first is a market that has grown more skeptical of the AI hype cycle, with rising fears that CapEx could peak and decline sooner than expected.

That interpretation, however, appears to be driven more by a shift in sentiment than by hard data. Signals from the production chain—particularly from TSMC —include firm orders, increased wafer signaling (advance volume commitments), and ongoing capacity planning, all pointing to real, contracted demand rather than a slowdown.

The second factor is skepticism around “stretched” valuations, especially relative to other AI leaders such as Nvidia. At its peak ahead of earnings, AVGO was trading at 57.2x forward earnings, well above Nvidia’s 36.5x forward multiple at the time. Following the recent correction, however, AVGO’s forward multiple has fallen to 32.3x, effectively resetting to roughly the same level seen at the beginning of the year.

Looking at the current valuation alongside the likelihood of rising earnings, driven by backlog upside and very robust growth in AI networking, I would argue that any remaining valuation premium largely reflects perceived demand risk. I see little evidence of a demand problem at Broadcom, especially given the company’s record backlog and improved revenue visibility. In fact, at this point, it may be more reasonable to assign greater demand risk to Nvidia than to Broadcom, particularly given Nvidia’s growing exposure to enterprise and neocloud customers, where CapEx tends to be less predictable.

Is AVGO a Buy?

The consensus on AVGO remains decidedly bullish. Of the 29 analyst ratings issued in recent months, 27 are Buys, and only two are Holds. Following the post–Q4 earnings pullback, several analysts raised their price targets, bringing the average target price to $461.93, which implies roughly 41.7% upside from current levels.

See more AVGO analyst ratings

Looking for the Signal in the Noise

In short, I view the recent correction in Broadcom shares as a case of multiple compression and a modest shift in sentiment, rather than any deterioration in the company’s fundamentals. Arguably, this is precisely the kind of setup value-oriented investors look for when buying the dip. A record backlog, strong revenue visibility, and robust growth in AI networking continue to support the underlying story, in my view. Over time, I expect the market to better price in backlog conversion and revise earnings higher, which should ultimately allow the stock to trade at richer multiples once again. All told, AVGO remains a Buy.