Research · The Verge ·

Nvidia plans N2X and N3X chips in aggressive GPU roadmap

Nvidia is accelerating its GPU roadmap with next-generation chips N2X and N3X already in development, signaling aggressive competition in AI accelerators.

Based on reporting by The Verge — analysis by dalili

Nvidia disclosed that it is actively developing its next two generations of AI accelerator chips, N2X and N3X, building on the current N100 architecture. The aggressive roadmap signals Nvidia's commitment to maintaining hardware leadership as AI training and inference costs dominate capital allocation for cloud providers.

The pace of chip development reflects the rapid iteration cycle now required in AI infrastructure. With competitors like AMD and startups closing capability gaps, Nvidia must deliver generational improvements every 18–24 months to sustain premium pricing and market share.

Industry observers note that the roadmap also addresses a critical constraint: training large language models requires proportional increases in GPU memory and compute, and N2X/N3X are designed to push both boundaries simultaneously.

Key takeaways

  • Nvidia N2X and N3X in active development, sustaining hardware leadership
  • GPU compute remains the primary constraint for AI scale through 2027–2028
  • Competitive pressure forces 18–24 month generational improvement cycles

Why it matters

Nvidia's roadmap aggression reflects a single constraint: compute will remain the bottleneck for AI at scale through 2027–2028. Whoever owns the accelerator wins.