Broadcom just told Wall Street it can see three years into the future, and it likes what it sees. CEO Hock Tan announced during the company’s Q2 fiscal 2026 earnings call that order visibility now extends to 2028, pushed out a full year from the 2027 timeline previously communicated to investors.

The chipmaker posted record quarterly revenue of $22.2 billion, a 48% jump year-over-year, fueled largely by AI semiconductor orders that topped $30 billion in the quarter alone.

The numbers behind the AI surge

The company designs custom AI accelerators for hyperscale clients. Google, Meta, Anthropic, and OpenAI were all named as parties involved in substantial multi-gigawatt compute infrastructure deals.

AI chip revenue is expected to exceed $100 billion in fiscal 2027. Some projections suggest that figure could reach as high as $180 billion by 2028.

The extended visibility to 2028 is backed by secured supply chains and active agreements that provide capacity assurance through 2029.

A $35 billion financing platform enters the picture

Perhaps the most intriguing announcement was a new $35 billion AI XPV compute financing platform. The platform is designed to help customers pre-plan and deploy the massive infrastructure capacity required for next-generation AI workloads.

What this means for investors

The $30 billion in quarterly AI bookings suggests that hyperscalers are not slowing their spending. AI chip revenue is projected to grow from $100 billion in fiscal 2027 to as high as $180 billion by 2028.

Nvidia continues to dominate the GPU market for AI training, and companies like Marvell Technology are competing directly with Broadcom in the custom silicon space. Amazon, Google, and Meta are all investing in their own in-house chip design capabilities as well, which could eventually reduce their dependence on third-party suppliers.