Broadcom has emerged as a frontrunner in application-specific integrated circuit (ASIC) technology, leveraging its expertise to design custom AI accelerators for hyperscalers and enterprise AI deployments. Notably, the company collaborated with Alphabet to develop Tensor Processing Units (TPUs), which now underpin large-scale AI workloads at Google Cloud. This success has attracted marquee clients including Meta Platforms and OpenAI, solidifying Broadcom’s reputation for delivering high-performance, cost-efficient alternatives to general-purpose GPUs.

Analysts at Citigroup project that Broadcom will generate just over $50 billion in AI-related revenue in fiscal 2026, rising to $100 billion in fiscal 2027. By comparison, the company reported total revenue of $63.9 billion in the last fiscal year, underscoring the magnitude of its AI opportunity. These forecasts exclude potential contributions from Apple and assume a conservative outlook on AI deployments by Anthropic, suggesting upside risk if those partnerships accelerate.

Broadcom exited fiscal 2025 with an AI switch backlog exceeding $10 billion, driven by strong demand for its Tomahawk series Ethernet switches. In total, the company reported $73 billion of AI-related backlog for delivery over the next 18 months, nearly $20 billion of which comprises non-accelerator networking components such as optical modules, digital signal processors and PCIe switches. This backlog provides multi-quarter revenue visibility and positions Broadcom to capture a significant share of the global AI infrastructure buildout.