The Story Behind Cerebras’ $63 Billion IPO with Founder and CEO Andrew Feldman

By No Priors Podcast

Categories: AI, VC

Summary

Cerebras achieved a $63B valuation by building a radically different AI chip architecture (wafer-scale at dinner-plate size vs. postage-stamp competitors) and waiting for the market inflection point—speed only mattered when AI models became useful enough for daily work in 2025, creating 15-20x GPU performance advantages across all model sizes.

Key Takeaways

  1. Radical improvement requires architectural differentiation—Cerebras built 46,000 sq mm wafer-scale chips instead of traditional postage-stamp designs, achieving 15-20x faster inference across all model sizes (small to trillion-parameter), not incremental gains.
  2. Market timing is critical: speed is worthless when AI is a novelty (2016-2024), but becomes essential once models are useful enough for daily work integration (2025 inflection). The market for slow inference is zero, like dialup internet.
  3. Demand explosion followed product-market fit, not vice versa: after 2025, Cerebras signed deals with OpenAI ($20B+) and AWS within months, demonstrating how speed becomes a forcing function once AI moves from experimental to operational use cases.
  4. Speed asymmetry creates winner-take-most dynamics: performance advantages of 15-20x are so substantial they create defensible moats. The threshold for user acceptance (like web page load times) means customers won't tolerate slow inference once they've experienced fast.
  5. New business models emerge from infrastructure speed increases: Netflix shifted from DVD rental to movie studio when internet speeds improved; AI will reorganize around fast inference, creating entirely new product categories and productivity jumps beyond current coding/design SaaS tools.

Related topics

Transcript Excerpt

Netflix used to deliver DVDs and envelopes and when the internet got fast they became a movie studio right it opened up an entirely new business something fundamentally different that's what happens with speed and I think that's what fast AI does right now we're replacing things that everybody can see like coding design the SAS tools but once we start sort of fundamentally reorganizing around this you're going to see this sort of new business models and fundamental jumps in productivity and I'm eager for that That's so cool. >> Today on No Prize, we have Andrew Feldman, the co-founder and CEO of Sarah Brass. Sarah Bros was founded in the mid 2010s to focus on new workloads for AI, particularly the machine learning world, and then has made the transition into very fast inference for the fou…

More from No Priors Podcast