Andrew Feldman, CEO of Cerebras, shares the company’s journey to a $63 billion IPO, highlighting their innovative wafer-scale chip technology that dramatically accelerates AI inference and the strategic partnerships that facilitated market adoption. He emphasizes the importance of resilience, timing, and leadership in pioneering AI hardware, and envisions Cerebras playing a key role in transforming industries through AI-driven productivity and new business models.
In this insightful interview, Andrew Feldman, co-founder and CEO of Cerebras, discusses the remarkable journey of his company, which recently went public with a market valuation of approximately $63 billion. Cerebras specializes in building AI-optimized computers that significantly accelerate AI workloads, particularly excelling in inference speed—achieving performance 15 to 20 times faster than traditional GPUs. Feldman highlights how the evolution of AI models reaching practical utility around 2025 triggered a surge in demand for faster inference, positioning Cerebras at the forefront of this transformative wave.
Feldman shares the technical and market challenges Cerebras faced, emphasizing their bold decision to pursue a radically different architecture using wafer-scale chips—massive chips the size of a dinner plate, unlike the conventional postage-stamp-sized chips. Despite skepticism and initial failures, the company persevered through years of development and financial strain, ultimately proving the feasibility of their approach in 2019. However, widespread market adoption lagged until AI became integral to daily workflows, underscoring the importance of timing and resilience in innovation.
The conversation also delves into the strategic partnerships that propelled Cerebras forward, including a landmark $20 billion deal with OpenAI and a deployment agreement with AWS. Feldman credits early adoption by supercomputing centers and strategic partners like G42 for helping the company bridge the gap between niche use and mainstream demand. He reflects on the complexities of scaling hardware manufacturing and software development, noting the decade-long effort required to build a robust compiler and the challenges of rapidly expanding production capacity.
Feldman offers candid insights into leadership and company culture, stressing the necessity of maintaining a fearless engineering mindset even as the company grows. He discusses the psychological challenges of leading a pioneering venture through periods of uncertainty and the importance of loving the journey rather than focusing solely on financial rewards. On the topic of going public, Feldman explains that the decision was driven by the desire to access a broader investor base, gain legitimacy, and provide liquidity to employees, while also marking Cerebras’ transition from a startup to a mature corporation.
Looking ahead, Feldman envisions AI’s acceleration enabling entirely new business models and productivity leaps, much like how faster internet transformed Netflix from a DVD delivery service into a major movie studio. He believes that while current AI applications improve existing tasks like coding and design, the true revolution will come from fundamentally reorganizing work around AI capabilities. Feldman expresses excitement about the future, anticipating that Cerebras’ technology will be central to unlocking these transformative possibilities.