Vertiv vs Broadcom: AI revenue hits $16.7 billion

Vertiv and Broadcom sit at different points of the AI data center buildout. Vertiv's business is physical: power, cooling and infrastructure, and it has expanded through projects such as Taiwan's first AI data center with NVIDIA GB300 and the first AI data center adopting 800V DC architectures. Its proposed acquisition of UtilityInnovation Group would add microgrid controls and energy-storage orchestration, capabilities aimed at helping operators secure power faster.

Broadcom's position is in the compute and network layers. Its custom accelerators, including the Ironwood TPU v7 and Google's next-generation TPU v8i, are shipped to customers like Google, Anthropic, OpenAI and Meta, while its Tomahawk 6 and Tomahawk Ultra switches are used across AI hyperscalers. The company has provided long-range guidance, saying it secured supply for roughly $115 billion in AI semiconductor revenue in fiscal 2027 and sees a path to roughly $230 billion in fiscal 2028.

The comparison hinges on execution: Vertiv's UtilityInnovation acquisition still needs regulatory approvals, and its growth depends on data center capacity expansion. Broadcom's scale is larger but rests on continued demand for custom accelerators and networking. The figures show both companies riding the same spending wave from different angles, and the coming quarters will test whether their guidance holds amid supply and power constraints.