Cerebras closes first day on Wall Street at ~$100B market cap — 68% pop validates the wafer-scale inference bet against NVIDIA
Cerebras stock soared 68% on its first day of public trading on May 15, closing with a market cap just below $100 billion. The IPO arrives one month after the company signed a $20 billion / 750 megawatt compute deal with OpenAI that includes a potential 11% equity stake — a transaction structure that effectively reprices the inference market away from NVIDIA's training-rack model.
The Cerebras pitch — wafer-scale chips with less inter-chip data movement, optimized for inference rather than training — was a contrarian bet through 2023-2024 when NVIDIA's H100/B100 stack was the only practical choice. The thesis turned on inference becoming the dominant workload and wafer-scale becoming the dominant token-economics architecture. The Day-1 mcap at $100B is the public market endorsing both legs of that bet inside one quarter.
The corollary for NVIDIA: the Groq $20B acqui-hire two months earlier was the same admission from the other direction — that the inference market is a separate product than the training market and that wafer-scale and LPU-class accelerators are going to take share. NVIDIA's Q1 data-center revenue at $75.2B (92% of total) shows the absolute dominance is intact; the question is whether the inference half of that pie has a different competitive shape going forward. Public markets seem to have decided yes.
CNBC — Nvidia competitor Cerebras after wild IPO → · eWeek — Cerebras targets $33B IPO challenging Nvidia → · Motley Fool — Cerebras vs Nvidia which AI stock is smarter buy →