Builder Daily

2026-05-08

Cerebras hikes IPO range to $125–$135 as orders hit ~$10B against $3.5B on offer

Cerebras hikes IPO range to $125–$135 as orders hit ~$10B vs $3.5B on offer. Implied valuation comfortably above $26.6B; CBRS prices May 13.

Bloomberg reported on May 8, 2026 that Cerebras Systems is raising its Nasdaq IPO price range from the original $115–$125 to $125–$135 per share. Banks are fielding roughly $10 billion in orders against just $3.5 billion on offer — over 20× oversubscribed. The deal pushes implied valuation comfortably past $26.6 billion when CBRS prices on May 13.

What’s in the offering

The base deal is 28 million Class A shares, plus a 4.2 million underwriter option (greenshoe). The S-1/A discloses ~$510M in 2025 revenue (up from $290M in 2024) and a multi-year, $20B+ compute commitment from OpenAI for 750 megawatts of inference capacity running through 2028.

The book is dominated by long-only mutual funds and sovereign wealth funds — typically a sign that pricing will land at the high end and traders should expect a tight float on day one.

What changed since the May 4 filing

Three things drove the price hike:

  1. The OpenAI compute deal disclosure expanded from “multi-year” to a quantified $20B+ commitment with named MW capacity.
  2. Anthropic-SpaceX Colossus deal (May 6) validated the thesis that hyperscaler-scale dedicated inference capacity is now a real product category with bookable demand.
  3. NVIDIA’s H200/B200 supply remained tight through Q1 2026 earnings, leaving room for an alternative inference vendor with binding hyperscaler commitments.

What to watch at pricing

Practitioner note

For infrastructure decisions: this changes Cerebras from “interesting research bet” to “vendor with public-market permanence and a $20B OpenAI backlog.” If you route latency-sensitive inference (real-time voice, live coding agents, interactive translation), the case for benchmarking your actual workload on Cerebras Inference cloud just got stronger — the company isn’t going anywhere for at least three years. Action item: spend $5–$10 of API credit comparing Cerebras Inference vs your current H100/H200 baseline on your top three latency-sensitive endpoints, before post-IPO hype makes the data noisier.


Sources

Tags

Tip