NVIDIA CorporationAnalyst argues chipmakers benefit from AI model competition, driving demand for GPUs and infrastructure.

Bank of America analyst Vivek Arya said semiconductor companies stand to gain regardless of whether open-source or proprietary artificial intelligence models ultimately dominate. In a research note, Arya noted that closed models such as Claude Opus 5, GPT-5.6 Sol and Gemini 3.1 Pro still lead on key enterprise metrics like agentic tool-use, coding and complex reasoning, with the best closed model scoring 55.3 against 43.1 for the top open-weight alternative on tool-use. He argued that if proprietary frontier models remain dominant, training them will require massive GPU clusters and heavy infrastructure spending, while broader adoption of open models could expand overall demand through the Jevons paradox, where lower costs increase total consumption. Arya compared the dynamic to the long coexistence of Microsoft Windows and Linux, suggesting AI infrastructure spending will become more diversified rather than disappear.
NVIDIA CorporationAnalyst argues chipmakers benefit from AI model competition, driving demand for GPUs and infrastructure.
Bank of America Corp