NVIDIA’s CEO remains unfazed by the buzz surrounding custom chip projects from tech giants like Microsoft and Google. In a recent talk at GTC Paris 2025, he confidently asserted that most custom chip ventures would likely be abandoned before making any significant impact.
NVIDIA continues to dominate the AI landscape, consistently breaking new ground with innovative architectures. The company sees itself as indispensable to the thriving computational segment. Despite Big Tech’s dependency on NVIDIA’s chips, many companies have embarked on developing ASICs, such as Microsoft’s Maia and Amazon’s Graviton, to challenge NVIDIA’s dominance. Yet, NVIDIA’s CEO dismisses these efforts as unlikely threats.
According to the CEO, many ASIC initiatives are initiated but ultimately scrapped. The key question remains: why build an ASIC if it isn’t superior to what’s already available? While companies like Google, Microsoft, and Broadcom are eager to create their own custom chips to potentially break free from NVIDIA’s market stronghold, the CEO believes these endeavors may fall short in practice.
Developing custom AI chips could theoretically lead to more efficient performance and financial benefits by tailoring chips to specific tasks. Having a proprietary supply could also shield companies from market fluctuations and delays. Nevertheless, the CEO argues that deployment challenges are where these efforts falter. Custom chip projects demand significant financial and technical resources, but deploying them at scale requires expertise that NVIDIA currently excels in.
In summary, NVIDIA’s leadership is not concerned about its chip technology being replaced by ASICs in the AI sector. They are confident in their continued dominance and technological superiority, suggesting that custom chip efforts may simply not measure up.
