StockLife
View plans
All analyses

Companies· 4 min read

Broadcom and AI revenue, from 8.4 to 10.7 billion dollars

Broadcom expects 10.7 billion dollars of AI revenue in the current quarter. Why custom chips are becoming the second pillar of the boom.

Broadcom and AI revenue, from 8.4 to 10.7 billion dollars
Photo: Michael Dziedzic on Unsplash

The short version

  • Revenue in the first quarter of financial year 2026 was 19.3 billion dollars, a gain of 29 percent.
  • The area with an AI connection reached 8.4 billion dollars, a gain of 106 percent, and 10.7 billion dollars are indicated for the current quarter.
  • Order books for custom chips now reach into 2028 according to market reports.

The figures of the quarter

Broadcom achieved revenue of 19.3 billion dollars in the first quarter of financial year 2026, a gain of 29 percent. The adjusted margin on the basis of operating profit before depreciation was 68 percent.

The area with an AI connection grew considerably faster. It reached 8.4 billion dollars, a gain of 106 percent. For the current quarter the company indicated 10.7 billion dollars. Observers consider it possible that this area grows to 100 billion dollars by 2027.

Of 19.3 billion dollars of quarterly revenue, 8.4 billion therefore come from the area with an AI connection. If this part keeps growing at triple digit rates and the rest of the business at 29 percent, the weight inside the group shifts from quarter to quarter without any single decision having to stand behind it.

Custom chips instead of standard accelerators

The difference to the best known supplier in the industry lies in the business model.

Broadcom therefore stands for a different thesis than the market leader. Not raw computing power but efficiency, network technology and tailored silicon.

For the buyer this route has two reasons, and both are economic. A chip that only has to serve its own applications carries no circuits for tasks this customer never sets, and needs correspondingly less power to run. And anyone who has a say in the design is less dependent on what a single supplier delivers next year and in what volumes.

How far demand is already fixed

A second point concerns the visibility of demand. Order books for custom chips now reach into 2028 according to market reports. That follows from the matter itself, because a jointly developed chip is not made for stock but to order, and the order stands at the beginning and not at the end of development.

The background is the investment boom of the large cloud providers. Together they are likely to spend more than 600 billion dollars on AI infrastructure this year. They are the same companies that increasingly commission their own designs in order to cut costs and dependence.

An order book reaching into 2028 means in this environment that part of this spending is already allocated before the money flows. For judging the business that is the more valuable information than a single quarterly figure, because it shows not what has been sold but what stays sold.

Assessment

An area that more than doubles while the whole company grows by 29 percent describes a shift inside the group.

What I find interesting is the strategic dividing line. Anyone who bets on standard accelerators sells computing power. Anyone who builds custom chips sells efficiency. Both are benefiting at the moment, but they hang on different phases of the same cycle.

Should the focus shift from raw performance towards running costs, the second model gains weight.

Frequently asked questions

How strongly is the area with an AI connection growing

It reached 8.4 billion dollars in the first quarter of financial year 2026, a gain of 106 percent. For the current quarter the company indicated 10.7 billion dollars. The company as a whole grew by 29 percent to 19.3 billion dollars over the same period.

What distinguishes a custom chip from a standard accelerator

A standard accelerator is manufactured in large volumes and sold to many customers. A custom chip is created together with a single buyer for that buyer's applications. It is less flexible but more efficient at exactly this task and cheaper to run.

What do order books reaching into 2028 mean

They describe the visibility of demand, that is how far the company already knows its future revenue. Long order books reduce the risk of a sudden collapse because customers are tied in. But they also mean that barely anything extra can be delivered at short notice when demand surges.

This text is not investment advice. It reports verifiable figures and puts them in context.

This analysis is for information only and is not investment advice.

More analyses

All analyses

↑↓ to move↵ to openesc to close