NVIDIA relies on AI for chip design as research costs climb

NVIDIA increasingly relies on AI to develop chips, with automated systems exploring alternative designs overnight to improve performance and energy efficiency. CEO Jensen Huang described the approach during a 2024 interview with Patrick Collison at Stripe Sessions. AI also helps identify software bugs, investigate their causes and route issues to the responsible engineers, while engineering teams evaluate results and develop new technologies.

The company reported $96.2 billion in second-quarter fiscal 2027 revenue, representing 106% annual growth. Quarterly research and development expenses rose from $4.29 billion to $7.05 billion, an increase of approximately 64%. GAAP gross margin improved from 72.4% to 75%, primarily because of a stronger product mix involving Blackwell Ultra rather than a disclosed benefit from AI-assisted engineering. NVIDIA has not quantified how much automation reduces development costs or accelerates product launches.

Competition from AMD and cloud companies developing custom processors is increasing pressure to improve energy efficiency, memory capacity, networking and operating costs. AI-assisted design could reduce unnecessary iterations and improve performance per watt, but rivals could adopt similar techniques. Rising memory prices also show the limits of design improvements in offsetting supply-chain costs. NVIDIA expects third-quarter gross margins of approximately 74%, below its latest reported result.