On May 14, 2026, Cerebras listed on Nasdaq, closing its first day at $311 — a 68% surge — raising $5.55 billion and briefly touching a $100 billion intraday market cap. Early investor Zhou Nan called it "the most insane IPO demand I've seen in my entire career." The company, founded in late 2015 by five engineers who left AMD, has since 2016 bet on building a chip from an entire silicon wafer. It spent three years in stealth mode and burned through nearly $300 million before its first Wafer-Scale Engine (WSE) taped out in the summer of 2019. As AI shifted from training to inference, the competitive dimension moved to memory bandwidth, favoring wafer-scale architecture, which delivers inference speeds roughly 15 times faster than GPU-based solutions. On Christmas Eve 2025, OpenAI signed the largest AI inference deployment contract in history, committing to deploy 750 megawatts of Cerebras compute power. Post-IPO, the supply chain is locked onto TSMC's 5nm process, while Nvidia allied with Groq, cloud giants develop custom ASICs, and OpenAI unveiled its first in-house inference chip, Jalapeño. After its first earnings report, a downward revision to gross margin guidance triggered a 15% single-day stock drop, briefly pushing shares below the IPO price. CEO Andrew Feldman said the company is leasing back equipment already sold to customers, a hit of roughly