$60B AI chip darling Cerebras almost died early on, burning $8M a month

· Source: TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, short

Summary

Cerebras Systems, a company selling AI chips for inference, recently achieved a blockbuster IPO, reaching a valuation of approximately $60 billion with its co-founders becoming billionaires. However, in 2019, the company faced near-failure, burning about $8 million monthly and incinerating nearly $200 million while attempting to solve a complex technical problem: creating a wafer-scale chip. This involved designing and manufacturing a single, giant chip from an entire silicon wafer, a feat previously deemed impossible in the semiconductor industry. The primary challenge was "packaging" – integrating the massive, power-intensive chip with cooling, power delivery, and data transfer systems, which required inventing new manufacturing processes and tools, such as a machine to simultaneously bolt 40 screws. After extensive trial and error, the team achieved a breakthrough in July 2019, leading to a functional wafer-scale chip.

Key takeaway

For investors evaluating high-risk, high-reward hardware startups, Cerebras Systems' journey highlights that significant capital burn and prolonged technical challenges can precede massive market success if a truly novel engineering problem is solved. Your due diligence should focus on the team's ability to innovate beyond existing industry solutions and their resilience in overcoming seemingly insurmountable technical hurdles, as this indicates potential for disruptive market entry and substantial returns.

Key insights

Cerebras Systems overcame unprecedented wafer-scale chip packaging challenges to achieve a $60 billion IPO.

Principles

Method

Cerebras developed a wafer-scale chip by designing a mega-chip, manufacturing it with TSMC, and then inventing custom packaging solutions for cooling, power, and data transfer, including specialized assembly machinery.

In practice

Topics

Best for: Entrepreneur, Investor, AI Hardware Engineer

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by TechCrunch.