A Selloff That Doesn’t Match the Numbers
Broadcom posted strong fiscal third-quarter results, and the market sold the stock anyway. That disconnect – between what the company actually reported and how investors responded – is drawing attention from anyone watching the semiconductor space closely heading into the final stretch of 2025.
The dip landed despite Broadcom maintaining what analysts describe as a clear line of sight to massive AI-driven revenue expansion over the next two years.
For investors willing to look past short-term price action, the question isn’t whether Broadcom is growing – it clearly is – but whether the selloff created an entry point that the underlying business fundamentals actually support.
What the Fiscal Q3 Results Actually Showed
Broadcom’s fiscal Q3 results came in strong by most conventional measures. The company has been building its AI revenue base steadily, and the trajectory it’s set out points toward a roughly 400% surge in AI-related revenue over the next two years. That’s not a vague aspiration – it’s a projection tied to existing customer relationships, product cycles, and the scaling demand for custom silicon and networking infrastructure that underpins modern AI data centers.
The company’s position in AI infrastructure is worth understanding specifically. Broadcom makes the custom AI accelerators – known as XPUs – that hyperscalers like Google and Meta use as alternatives or complements to off-the-shelf GPUs. It also dominates the high-speed networking interconnects that link those chips together inside data centers. Both segments are growing as AI workloads scale, and both give Broadcom exposure to AI spending that doesn’t depend on any single customer or chip design winning the market outright.
That diversification inside the AI buildout is part of why the revenue forecast carries weight. When a company’s AI growth story rests on a single customer or a single product cycle, a delay or a contract loss can unravel the math quickly. Broadcom’s exposure is spread across the infrastructure layer itself – the switches, the custom silicon, the connectivity – which means it collects revenue as the category grows rather than betting on one player inside it.
Why the Stock Dropped and What It Means for Buyers
Markets don’t always punish bad results. Sometimes they punish good results that weren’t good enough relative to what was already priced in. Broadcom’s dip after fiscal Q3 likely falls into that second category. The stock had already priced in substantial optimism, and even a clean beat can trigger profit-taking when expectations are elevated and the broader market is looking for reasons to rotate.
That creates a specific dynamic for investors evaluating the stock now. The business hasn’t changed. The AI revenue projection – a 400% increase over two years – is still intact. The customer relationships with major hyperscalers haven’t shifted. What changed is the share price, which moved lower without a corresponding change in the company’s actual financial outlook or competitive standing. That gap between price and fundamentals is exactly what value-oriented investors in growth stocks look for, though it doesn’t guarantee the stock stops falling in the near term.
Timing a dip-buy in a high-multiple semiconductor stock is never straightforward. Broadcom trades at a premium because the market assigns it premium growth expectations – and premium-valued stocks can stay under pressure longer than the underlying numbers suggest they should. The 400% AI revenue projection gives long-term holders a concrete anchor, but investors with shorter time horizons or lower tolerance for drawdowns face real risk sitting in a name that the market has decided, at least temporarily, to reprice lower.
The Bigger Picture Behind the 400% Figure
A 400% revenue increase over two years is a number that demands scrutiny rather than acceptance. In Broadcom’s case, the math is rooted in where AI infrastructure spending is heading.
Hyperscalers – the large cloud providers running the world’s biggest AI training and inference operations – have publicly committed to spending levels on AI infrastructure that dwarf anything seen in previous technology build cycles. Broadcom sits at an unusual intersection of that spending: it designs custom chips for some of the largest players, and it manufactures the networking equipment that makes large-scale AI clusters function. As those clusters grow in size and complexity, the demand for both product lines scales with them. The 400% projection isn’t built on Broadcom taking market share from competitors so much as it’s built on the market itself expanding at a rate that lifts all serious participants in the infrastructure layer.
There’s also a competitive moat argument embedded in the AI revenue story. Custom silicon design – the kind Broadcom does for Google’s TPUs and other proprietary accelerators – requires years of engineering collaboration with the customer. Those relationships don’t switch easily. Once a hyperscaler has built its AI infrastructure around a specific chip architecture co-developed with Broadcom, migrating to a different vendor is costly and disruptive. That stickiness gives Broadcom’s revenue forecast more durability than a simple market-growth projection would imply on its own.
Putting It Together
Broadcom’s fiscal Q3 results were strong. Its AI revenue is on track for 400% growth over the next two years. Its customer base includes the largest AI spenders in the world, and its product portfolio covers two of the most critical infrastructure layers – custom silicon and high-speed networking – in the AI data center buildout. The stock dipped anyway, creating a situation where the investment case is arguably stronger than the share price currently reflects. Whether that gap closes quickly, slowly, or not at all before the next earnings cycle is the only question left – and it’s one Broadcom’s next quarterly report will begin to answer.
