Broadcom targets $29.4bn quarter as AI chip expectations surge

Broadcom is expected to report quarterly revenue of US$29.4 billion when it publishes its latest results, representing an 85% year-on-year rise, with adjusted earnings forecast to reach US$3.23 a share. Yet analysts warn that even a record quarter may not be sufficient to move the stock, as investor expectations for forward guidance have outpaced near-term delivery.

The semiconductor group has been one of the clearest corporate beneficiaries of the AI infrastructure build-out, with demand for its custom silicon products running at almost three times shipments in the prior quarter. That order-book backlog points to constrained supply rather than weakening demand as the primary bottleneck, a distinction that will weigh heavily on how investors receive management's guidance for the coming year.

The guidance problem

Josh Gilbert, lead analyst for APAC and the Middle East at trading platform eToro, said the bar has risen materially since Broadcom's previous earnings call. "Last time out, Broadcom posted record AI revenue, and its shares still fell 12.6% after the Q3 AI forecast missed elevated expectations and management left its longer-term target unchanged," he noted. "Another conservative target won't cut it with shareholders."

Broadcom's management has guided for more than US$100 billion in AI semiconductor revenue for the following fiscal year. Gilbert argues the market is pricing in something closer to US$120 billion, and that with bookings significantly outpacing shipments, investors expect supply constraints to ease and guidance to reflect that. Nvidia's most recent results, published the prior week, reinforced the picture of robust infrastructure demand across hyperscaler and enterprise customers, raising the threshold further for Broadcom's own commentary.

The Google question

A secondary concern centres on Broadcom's relationship with Alphabet's Google, which is understood to be its largest custom silicon customer. A reported chip agreement between Google and rival Marvell Technology has raised market speculation that Broadcom could face share loss at its most important account.

Gilbert said Broadcom needs to demonstrate clearly that "Google is adding a supplier rather than swapping one out." Custom silicon, or application-specific integrated circuits designed for specific AI workloads, is a high-margin, high-dependency business: design cycles are long, switching costs are substantial, and hyperscaler relationships typically span multi-year commitments. A reassuring commentary on Google's trajectory would carry significant weight in investor sentiment.

Market and competitive context

Broadcom sits alongside Nvidia and Marvell as one of a small number of established players with the design capability and manufacturing relationships to serve hyperscaler-scale custom AI silicon programmes. While Nvidia dominates the merchant GPU market, Broadcom's strength lies in custom ASICs and high-speed networking silicon, areas that are growing as large cloud providers seek workload-optimised alternatives to general-purpose accelerators.

The broader semiconductor sector is also navigating US export controls under the Bureau of Industry and Security's AI chip rules, which restrict the sale of advanced accelerators to a number of markets. Broadcom has not disclosed in public filings the proportion of its AI revenue that is subject to those controls, a detail that could become material if restrictions tighten further.

The results are expected to confirm that AI infrastructure spending remains elevated. The harder task for management will be translating record demand signals into guidance that persuades investors US$100 billion represents a floor, not a ceiling.