Broadcom has become the market's favorite way to bet on custom AI chips, and its June 3 earnings report now has to prove that enthusiasm is more than momentum.
Broadcom is walking into earnings with very little room for a quiet quarter. The chip and software company has added roughly $280 billion in market value over four sessions, turning what might have been a standard fiscal second-quarter update into a live test of the AI infrastructure trade.
According to Broadcom's investor notice, the company is set to report fiscal Q2 results after the market closes on June 3, with management holding its call at 5 p.m. ET. That timing matters because investors are not simply waiting for a revenue number. They are waiting to hear whether the custom silicon boom is still accelerating, whether hyperscalers are still leaning harder into ASICs, and whether VMware is producing enough cash to support a valuation that has quickly become demanding.
The benchmark is already high. In fiscal Q1, Broadcom reported record revenue of $19.3 billion, up 29% year over year. AI revenue reached $8.4 billion, up 106%, helped by demand for custom AI accelerators and AI networking. Management then guided fiscal Q2 AI semiconductor revenue to about $10.7 billion and total Q2 revenue to roughly $22 billion, with adjusted EBITDA expected at about 68% of projected revenue.
Those are not small numbers. They place Broadcom in a different conversation from most chip companies, because the company is no longer being treated as a diversified semiconductor supplier with a software arm. It is being valued as one of the few credible alternatives to Nvidia for large-scale AI infrastructure.
Nvidia still owns the center of the AI chip story. Its GPUs, CUDA software ecosystem and early lead in training infrastructure give it an advantage that cannot be copied by simply announcing a new chip. But Broadcom is playing a different game. It helps major cloud and internet companies build custom AI accelerators and the networking systems that connect them inside massive data centers.
That distinction matters because hyperscalers do not want every AI workload to run on the same general-purpose hardware. Alphabet has long used Tensor Processing Units. Meta, OpenAI and other large model builders have strong incentives to explore custom silicon where they can control cost, performance and supply. Broadcom benefits when these companies decide that owning more of the chip stack is worth the engineering effort.
The networking side is just as important. AI systems are not only about raw compute. They are about moving huge volumes of data between processors without wasting time or power. Broadcom's strength in Ethernet switching, optical components and custom data center connectivity gives it exposure to the less glamorous pieces of AI infrastructure that still determine whether a cluster performs well.
This is why the market is watching the Q2 call so closely. A clean beat would reinforce the idea that Broadcom is becoming the picks-and-shovels supplier for custom AI infrastructure. A cautious guide would raise a tougher question: whether investors have raced ahead of the actual deployment curve.
VMware gives Broadcom a second lever
Broadcom's AI story is powerful, but it is not the whole company. VMware, acquired in 2023, has changed the earnings profile by adding a large infrastructure software business with subscription revenue and deep enterprise relationships. In fiscal Q1, Broadcom's infrastructure software revenue reached $6.8 billion, giving the company a cash-generating base that many pure chip names do not have.
That software cash flow helps explain why investors have been willing to pay up. AI chips can be cyclical, supply constrained and customer concentrated. VMware gives Broadcom another way to produce profit while it funds semiconductor growth, returns capital and keeps integrating a major acquisition. The risk is that customers have pushed back against VMware pricing and packaging changes, which means Broadcom still has to show that higher profitability does not come at the expense of long-term customer loyalty.
For now, Wall Street appears focused on the AI upside. Analyst target increases and pre-earnings positioning have helped push the stock toward record levels, with some market data services showing Broadcom's market value above $2 trillion in recent sessions. That is a remarkable move for a company that was still widely viewed as a disciplined acquirer and dividend compounder only a few years ago.
But a bigger market value changes the burden of proof. When a stock has already priced in extraordinary growth, good news can become normal news. Investors will want more than a confirmation that AI revenue rose. They will want details on customer demand, order visibility, networking growth, margins and whether the $10.7 billion Q2 AI semiconductor target is the start of a longer step-up rather than a one-quarter surge.
This is the uncomfortable part of AI investing right now. The demand is real, and Broadcom is one of the few companies with the relationships and technology to capture a meaningful share of it. At the same time, the market is beginning to price second-tier AI winners as if they already own the next phase of the cycle.
That makes June 3 more than an earnings date. It is a check on how much confidence investors should place in the custom silicon trade. If Broadcom delivers strong numbers and clearer demand visibility, the Nvidia alternative story becomes harder to ignore. If guidance sounds merely solid, the market may be reminded that even excellent businesses can get ahead of themselves when AI euphoria does the math first.
Also read: AI worms have moved from theory to demonstrated cyber risk • Factorial raises fresh capital to turn HR software into AI work • Apoha raises $36 million to bring AI into materials discovery
