Software Recurring RevenueNarrow moat

By acquiring large enterprise-software businesses, Broadcom gave itself a second engine entirely distinct from its chips: recurring, subscription-like revenue drawn from customers whose critical systems run on that software and who would face real pain in switching away from it. This was a genuinely surprising move for a semiconductor company, and it has proven shrewd, because it grafts the steady, predictable cash flow of enterprise software onto the more volatile body of the chip business.

The stickiness of the software is of the same species as the stickiness of the chips, merely expressed in a different medium. When a large enterprise has built its operations around a particular piece of software — its infrastructure, its security, its core systems — replacing that software is a costly, risky, and disruptive project that few undertake without overwhelming cause. The customer is embedded, and embeddedness is the whole game, whether the thing embedded is a networking chip or a mainframe software suite.

One of the most valuable properties of the software revenue is the way it smooths the cyclicality that has always plagued the semiconductor industry. Chip demand rises and falls with the broader economy and the inventory cycles of Broadcom's customers, producing lumpy, unpredictable results; the software subscriptions, by contrast, renew steadily through good times and bad. Blending the two gives the whole enterprise a steadier and more predictable profile than a pure chip company could ever achieve, which the market rightly values.

The strategy does lean heavily on raising prices for captive customers, and this is where candor is required. Broadcom's approach to its acquired software franchises has often involved significant price increases on customers who find switching too painful to contemplate — a tactic that works precisely as long as the switching cost holds. It is effective, but it must be managed with some care, because a captive customer pushed too hard eventually devotes real resources to finding the exit, however difficult that exit may be.

In the end, the software business rests on the same foundation as the whole company: systems woven so deeply into a customer's operations that unwinding them is scarcely worth contemplating. It is a less romantic moat than a beloved consumer brand, built not on affection but on the sheer inconvenience of leaving — but inconvenience, at sufficient scale and depth, is a perfectly durable thing to build a business upon, and Broadcom has built a very large one indeed — software is now roughly two-fifths of a ~$64B company¹.