Broadcom reported strong momentum in artificial intelligence infrastructure during its fiscal Q1 2026 earnings call, highlighting growth in custom AI accelerators, high-performance networking, and strategic partnerships with hyperscale technology companies.
AI Infrastructure Strategy Accelerates Growth
Broadcom’s AI strategy focuses on two core technologies – custom AI accelerators (XPUs) and advanced AI networking solutions designed for hyperscale data centers.
Revenue from custom accelerators surged 140 percent year-over-year in the first quarter. The company now works with six major customers including Google, Meta, Anthropic, OpenAI and two undisclosed partners.
Broadcom expects large-scale deployments from these customers over the next several years. OpenAI is projected to deploy more than 1 gigawatt of capacity using Broadcom-designed chips by 2027.
In AI networking, revenue increased 60 percent, supported by next-generation switching platforms such as the Tomahawk 6 switch, which delivers 100 Tbps performance, and the upcoming Tomahawk 7 architecture. The company’s 200G and 400G SerDes technologies are also playing a major role in scaling high-performance AI clusters.
Broadcom is also emphasizing Direct Attached Copper (DAC) connectivity for large AI clusters, which the company says offers significantly lower costs and improved power efficiency compared with optical networking solutions.
Strategic Partnerships with Hyperscale Customers
Chief Executive Officer Hock Tan said Broadcom has built deep, multi-year partnerships with its core AI customers to support large-scale infrastructure deployments.
To ensure supply continuity, the company has secured manufacturing capacity for leading-edge wafers, high-bandwidth memory, and advanced substrates from 2026 through 2028.
Broadcom is also integrating its infrastructure software portfolio following its acquisition of VMware. The company is positioning VMware Cloud Foundation as a foundational software layer that enables enterprises to run private cloud AI workloads, including generative AI and emerging agentic AI applications.
Tan also addressed industry speculation about Meta’s in-house chip development, confirming that the company’s MTIA accelerator roadmap remains active and that Broadcom is currently shipping these custom processors.
Strong Financial Performance in Q1 2026
Broadcom delivered strong financial results for the first quarter of fiscal 2026.
Total revenue reached $19.3 billion, representing 29 percent year-over-year growth and exceeding company guidance.
The semiconductor solutions segment generated $12.5 billion in revenue, up 52 percent from the previous year. AI semiconductor revenue alone reached $8.4 billion, more than doubling compared with the same period last year.
Infrastructure software revenue totaled $6.8 billion, reflecting modest growth of 1 percent, while VMware-related revenue increased 13 percent.
Profitability also remained strong, with adjusted EBITDA reaching a record $13.1 billion, representing a 68 percent margin.
Bullish Outlook for AI Chip Market
Broadcom provided an optimistic outlook for the coming quarters and beyond as global technology companies accelerate investments in AI infrastructure.
The company expects second-quarter revenue of approximately $22 billion, representing projected growth of 47 percent. AI semiconductor revenue in the quarter is expected to reach $10.7 billion, up around 140 percent year-over-year.
According to Hock Tan, Broadcom now has clear visibility into AI chip demand over the next several years.
The company expects AI chip revenue to exceed $100 billion in 2027 as demand from hyperscale cloud providers continues to expand.
Major technology companies including Alphabet, Microsoft, Amazon and Meta are collectively expected to invest more than $630 billion this year to build AI infrastructure, driving demand for semiconductors, servers, storage and high-speed networking technologies.
While Nvidia currently dominates the AI accelerator market, Broadcom is increasingly positioning itself as a major contender through its strategy of developing custom silicon tailored to the needs of large hyperscale customers.
RAJANI BABURAJAN
