Broadcom's AI Surge Is Real, but Its Valuation Still Needs Proof

Broadcom's fiscal third quarter delivered the kind of growth usually associated with a much smaller company. Revenue reached $29.6 billion, up 86% from a year earlier, while non-GAAP operating income was $20.1 billion and non-GAAP diluted earnings were $3.32 per share.

The more revealing number is cash. Broadcom generated $14.2 billion from operations and spent about $500 million on capital expenditures, leaving $13.7 billion of free cash flow. That equals roughly 46% of quarterly revenue and gives the company financial flexibility that many AI infrastructure suppliers lack.

Management attributed much of the acceleration to demand for custom AI accelerators and networking products. These chips are designed with large cloud customers for specific workloads, offering Broadcom a different position from Nvidia's more general-purpose accelerator platform. The model can produce large, durable programs, but it also concentrates revenue among a small number of buyers.

The original report paired Broadcom's results with proprietary share-price targets. Those targets are not necessary to understand the business and cannot be independently reproduced from the disclosed assumptions, so they are excluded here. The official release supports the operating figures, not a guaranteed valuation outcome.

Broadcom's results also contain an important counterweight to industry warnings about excessive AI capital spending. Customers are paying for deployed silicon and networking equipment now, and the company is converting a large share of revenue into cash. That is stronger evidence than announcements about future capacity alone.

Still, current growth cannot be projected mechanically. Custom-chip programs can shift with customer design choices, data-center schedules, power availability and memory supply. Broadcom also combines semiconductor operations with infrastructure software, so investors should separate the sources of growth and margin rather than treating the company as a pure AI-chip vendor.

The next test is whether AI revenue keeps expanding without eroding companywide margins or increasing dependence on a handful of hyperscalers. Fiscal third-quarter execution supports the positive case, but the valuation depends on how long that growth remains both concentrated and unusually fast.