Nvidia's AI Supercomputer Costs Heavily on Vera Rubin Chips

Nvidia's AI supercomputer costs $4 million to build, and most of that expense goes towards Vera Rubin Chips.

A recent bill of materials from HSBC shows that 70% of the cost is attributed to the Vera Rubin GPUs themselves, excluding high-bandwidth memory. This means that Nvidia's silicon takes a significant share of the revenue generated by these supercomputers.

Nvidia CEO Jensen Huang explained during a call with investors that 'Today, we’re not just selling the best chips. We’re selling a full-stack AI factory platform.'

The company's data center revenue hit $89.02 billion in Q2 FY2027, up 117% year over year. Management guided fiscal Q3 revenue to $108.0 billion plus or minus 2%, and expects Vera Rubin to account for about 20% of data center revenue in Q3.

The strong demand for Nvidia's GPUs has been driven by the growing need for AI infrastructure, with top-five hyperscalers expected to reach nearly $800 billion in 2026 and $1.3 trillion in 2027. Nvidia has already received purchase orders from every major hyperscaler, AI cloud, and system OEM, and expects Vera Rubin to be the fastest product ramp in company history.