Broadcom's AI Growth Fueled by Oracle's Cloud Expansion
Broadcom's trajectory in artificial intelligence (AI) acceleration and high-speed networking is set for substantial growth, largely propelled by Oracle's ambitious cloud expansion. Oracle's staggering $455 billion cloud backlog signals a robust demand for the advanced technologies that Broadcom specializes in. This dynamic partnership underscores the critical need for sophisticated infrastructure to support the burgeoning AI landscape, especially as inferencing workloads become more prevalent and complex. The synergy between these two tech giants is creating a powerful momentum, positioning Broadcom for sustained expansion in the coming years.
The strong potential for Broadcom was recently highlighted during Oracle's first-quarter fiscal 2026 earnings call, which concluded on August 31. During this call, Oracle's leadership revealed aggressive plans to significantly expand its global AI infrastructure. This expansion is supported by an impressive $455 billion in remaining performance obligations (RPO), indicating a massive backlog of signed contracts, a remarkable 359% increase year over year. This substantial commitment from Oracle translates directly into increased demand for Broadcom’s offerings, including custom AI accelerators, high-speed Ethernet switches, and optical interconnects. These components are vital for building the next generation of AI data centers.
A critical aspect of this development is the reported five-year, $300 billion agreement between OpenAI and Oracle for computing capacity. Such a monumental deal necessitates a rapid and extensive expansion of Oracle’s data center footprint. Consequently, Oracle plans to allocate nearly $35 billion in capital expenditures in fiscal 2026, primarily for data center equipment. This accelerated deployment strategy aims to monetize capacity sooner, which in turn promises a faster and more consistent demand for Broadcom's specialized products. As AI workloads, particularly inferencing, grow in complexity and volume, the need for ultra-low latency and high-bandwidth networking solutions becomes paramount. Broadcom's Ethernet-based switches and routers are ideally suited to meet these demanding requirements.
Broadcom's recent financial performance further validates its strong position in the market. In the third quarter of fiscal 2025, ending August 4, the company reported a 22% year-over-year revenue increase to $16 billion. Adjusted earnings before interest, taxes, depreciation, and amortization (EBITDA) also saw a significant surge of 30% year over year, reaching $10.7 billion. The company also announced a record backlog of approximately $110 billion. AI continues to be the primary catalyst for this growth, with AI revenue soaring 63% year over year to $5.2 billion in the third quarter, 65% of which came from its custom accelerator business. This robust growth is largely attributed to strong demand from its major hyperscaler clients, with Broadcom having secured an additional $10 billion in XPU-based AI rack orders from a fourth client. Management anticipates even faster AI revenue growth in fiscal 2026 compared to fiscal 2025, fueled by Oracle's aggressive data center expansion and increasing inference workloads.
The growing intelligence of large language models and the expanding size of AI clusters have led to an explosion in demand for Ethernet-based networking. When AI clusters exceed 100,000 GPUs or XPUs, ultra-high bandwidth networking becomes indispensable for sharing memory across nodes. Broadcom’s Tomahawk switches and Jericho Ethernet fabric routers offer exceptional efficiency in connecting these powerful processing units within racks, across data centers, and even between geographically dispersed data centers. Oracle’s strategy to establish more data center regions and enhance inference capabilities on enterprise data aligns perfectly with this trend, creating a higher demand for networking components. Hyperscalers favor Ethernet due to its open-source nature, proven reliability, and widespread familiarity among architects and engineers, as well as the flexibility it offers by avoiding vendor lock-in. Thus, as AI clusters continue to scale, the demand for Ethernet-based networking solutions is projected to skyrocket in the coming years.
While Broadcom's valuation, trading at 36 times forward earnings, might appear high, its substantial $110 billion backlog and accelerating demand for AI-optimized accelerators and Ethernet networking paint a compelling picture. Oracle’s massive AI infrastructure build-out is set to drive sustained spending in the sector, positioning Broadcom for multi-year growth. Long-term investors may find opportunities to acquire stakes in Broadcom stock, particularly during any market pullbacks, as the company is strategically placed to capitalize on the ongoing AI revolution.