Broadcom delivered numbers that once would have ignited a rally. Revenue hit $22.187 billion in the second fiscal quarter of 2026, a 48 percent jump from the prior year. AI semiconductor sales alone reached $10.8 billion, up 143 percent. Adjusted earnings per share came in at $2.44, beating estimates. Free cash flow topped $10 billion with a 67.3 percent margin.
Yet shares plunged as much as 13 percent in extended trading. The reason sat in the guidance. Hock Tan, Broadcom’s chief executive, pointed to $16 billion in AI chip revenue for the current quarter. That figure trailed some analyst targets. He held the full-year AI semiconductor outlook at $56 billion. Investors wanted more. They received steady confidence instead.
Tan offered something else that stood out. Asked about fears that agentic AI systems would erode traditional software licensing models, he delivered a blunt response. “We’re not seeing it.” The remark, reported by Yahoo Finance, cut against the narrative that advanced AI agents might render enterprise software obsolete. Tan saw the opposite. AI appeared to accelerate demand for Broadcom’s software assets, particularly VMware.
This stance matters. Broadcom’s 2019 purchase of VMware once drew skepticism. Integration risks loomed. The company’s hypervisor layer seemed vulnerable to disruption. Tan pushed back. Stronger demand for VMware infrastructure has emerged alongside the AI buildout. Enterprises need reliable foundations to run these massive GPU clusters. The software side benefits.
And the hardware side dominates the story. Custom AI accelerators now drive the bulk of growth. Hyperscalers have turned to Broadcom to design chips tailored to their exact workloads. Google relies on Broadcom for portions of its TPU ecosystem. Meta uses the company for its MTIA chips. Partnerships with Anthropic and others have expanded. These deals lock in multi-year revenue streams that extend well beyond commodity GPU sales.
Tan first laid out the scale in March. He told analysts that visibility into 2027 had “dramatically improved.” Broadcom held line of sight to AI chip revenue “in excess of $100 billion” that year. The company had secured supply chains to support it. That projection still stands. Recent commentary from the Bloomberg Tech conference reinforced the shift. Tan indicated Broadcom would de-emphasize acquisitions. Organic AI growth simply outpaces what most deals could deliver. Bloomberg captured the executive’s arithmetic: “I’m looking around — what can I buy that would come close to that?”
The market reaction exposed tension. Broadcom’s valuation had priced in relentless acceleration. When the third-quarter AI target of $16 billion missed the highest expectations, some investors hit the sell button. Reuters noted the forecast also fell short of the $17.2 billion average some models carried. Shares slid. The broader semiconductor sector felt the pull. Yet the underlying trajectory remains intact. From $8.4 billion in AI revenue in the first quarter to $10.8 billion in the second, the ramp continues. The $16 billion guide still implies more than 200 percent year-over-year growth for the period.
Custom silicon sits at the center. Unlike Nvidia’s focus on general-purpose GPUs, Broadcom specializes in application-specific designs. These chips optimize power, latency, and cost for particular AI training or inference tasks. Hyperscalers gain differentiation. They avoid full dependence on one supplier. Broadcom captures high margins on the engineering work and subsequent volume. The model scales.
Recent reporting adds color. A June 5 Bloomberg video interview featured Tan discussing the “leap of faith” the company took with Anthropic a year earlier. That bet has paid off. Demand for these specialized accelerators shows no sign of pausing. Separate coverage from The Next Web two days ago highlighted how AI revenue now makes external deals less compelling. Organic expansion offers superior returns.
Software remains the quiet stabilizer. VMware’s performance has exceeded internal expectations in several areas. AI workloads require sophisticated orchestration, storage, and networking layers. Broadcom’s portfolio supplies them. Tan’s dismissal of the “saaspocalypse” theory carries weight because his company sits at the intersection. It sells both the chips that power the models and the infrastructure that keeps them running reliably at scale.
Wall Street has heard the $100 billion figure before. Tan repeated it on the latest call. He expects momentum to carry into fiscal 2027. Some analysts translate the projection into 10 gigawatts of AI chip shipments by that year, according to recent analysis. The power consumption alone signals the infrastructure buildout underway. Data centers are expanding at a pace few predicted even two years ago.
But execution risks persist. Supply chain constraints could reappear. Competition in custom silicon may intensify. And investor patience has limits. The post-earnings drop served as a reminder that even strong results can disappoint when expectations have run far ahead. Broadcom’s market capitalization shed roughly $380 billion in one session before partial recovery.
Tan has guided this company through multiple cycles. He built much of its scale through acquisitions. Now he bets on internal momentum. The numbers support him so far. AI semiconductor revenue has more than doubled in recent quarters. Margins have expanded. Cash generation sets records. The question is whether the street will accept steady progress or demand constant upward revisions.
For now, Tan holds the line. AI demand looks durable. Software faces no immediate threat from agentic systems. In fact, the two appear complementary. Enterprises will spend on both the specialized hardware and the management layers that make it useful. Broadcom sits in the middle, collecting on each.
That positioning explains why the company’s long-term outlook continues to draw attention even after a disappointing stock reaction. The $56 billion fiscal 2026 AI target may have missed some forecasts. The $100 billion 2027 figure still looms large. And the message on software offers a counterpoint to prevailing pessimism in parts of the market.
Investors will watch the next several quarters closely. If AI revenue continues its steep climb and VMware shows further strength, the recent selloff could prove temporary. Tan has delivered results before. The latest quarter added to that record even if the immediate price reaction suggested otherwise.
WebProNews is an iEntry Publication