Supermicro reported $12.7 billion in Q2 2026 revenue, up from $5.7 billion a year earlier and $5 billion in Q1. GPU-based AI systems delivered 84% of sales, growing 151% year-over-year.
The concentration risk is notable: one customer accounted for 63% of quarterly revenue. CEO Charles Liang says the company has signed similar-scale buyers and isn't concerned. "We are very happy that now we have many more large-scale customers," he told analysts.
Gross margins fell to 6.3%, which Liang attributed partly to expedited shipping costs for Nvidia Blackwell components. He expects those expenses to ease as supply chains normalize. Tariff concerns have also diminished, according to the CEO.
The company is betting on its Data Center Building Block Solutions (DCBBS), the modular rack-scale offering announced last week that packages servers, storage, networking, cooling, and power management into turnkey deployments. DCBBS generated just 4% of Q2 revenue but represents Supermicro's primary growth vector. The company is developing new modules including transformers, next-generation power systems, and grid power alternatives.
Four new factories are ramping production to meet DCBBS demand and reduce costs. Liang forecast Q3 revenue of $12.3 billion and full-year revenue of "at least $40 billion", suggesting Q4 could dip to around $10 billion. He didn't explain the projected sequential decline.
This marks a welcome shift for Supermicro after years of regulatory drama: Nasdaq delisting threats, delayed financial reports, and accounting questions. The company resolved its compliance issues in 2024.
The trade-offs are clear. Supermicro is winning large AI infrastructure deals, but customer concentration and thin margins present execution risk. Analysts continue to flag margin pressure as competition intensifies from Dell and HPE in the AI datacenter market. The question is whether DCBBS and expanded manufacturing capacity can sustain growth when that dominant customer diversifies its supply chain.