Broadcom sees a runway to extraordinary AI-chip growth as frontier labs race to build computing capacity. Investors see the same demand surge, but question whether financing, customer concentration and real-world infrastructure can support the promise.

Broadcom’s fiscal third-quarter results gave management fresh ammunition on Wednesday: revenue rose 86% to $29.59 billion and adjusted earnings of $3.32 a share topped expectations. Semiconductor revenue more than tripled to $16.7 billion, evidence that custom AI processors and networking gear are already powering the company’s expansion.¹

On the earnings call, CEO Hock Tan laid out the bigger prize. Broadcom expects AI revenue to reach $115 billion in fiscal 2027 and then double to $230 billion in 2028, driven by deployments for Google, Anthropic, OpenAI and Meta. “Shipments of Jalapeno for OpenAI will continue,” Tan said, while Meta’s custom accelerator would move into production for inference and recommendations.¹

The company’s confidence rests not only on chip orders but on customers’ ability to fund enormous buildouts. Finance chief Amie Thuener said Broadcom was helping leading AI labs “bridge the gap between their current cash flow and the significant upfront investments required,” including potential residual-value guarantees.¹

That is where the market’s enthusiasm turns conditional. Broadcom’s $230 billion 2028 target beat outside expectations and helped lift the shares after hours, but its near-term fourth-quarter revenue outlook of $34.8 billion came in slightly below consensus. Analysts also noted that the forecast already assumes constraints around land, power and data-center shells—the physical bottlenecks behind the AI boom.²

The cautious view is not that demand has vanished; it is that demand alone does not settle the investment case. A financing platform involving Apollo and Blackstone may help support more than 20 gigawatts of infrastructure, yet it has sharpened concerns about circular financing. Alongside political resistance to data centers and reliance on a handful of AI labs, those risks prompted a hold stance and a lower price target despite the blockbuster long-term numbers.²