The Cerebras Systems initial public offering was a resounding success, generating billions of dollars for major investors including Benchmark. Among the biggest winners is Eric Vishria, a venture capitalist who almost skipped the first meeting with the startup.

Recalling his initial reluctance to meet with Cerebras, Vishria admitted, “It was five founders and a deck, and it was our first hardware investment in 10 years.” His grumpy attitude vanished by the third slide when co-founder Andrew Feldman laid out plans for AI training chips. The light bulb moment came as he realised GPUs were inadequate for deep learning.

Over eight years of development, Cerebras faced numerous challenges, including new cooling methods and manufacturing techniques, all while Vishria questioned their decision. Hardware investments are risky, but Benchmark’s persistence paid off when Cerebras’ chips proved invaluable in AI inference, leading to significant revenue growth and a profitable year.

The journey from a skeptical first meeting to multibillion-dollar success showcases the unpredictable nature of venture capital. Vishria credits the team for their “adaptiveness” despite his initial skepticism. For Benchmark, it’s another feather in their cap, demonstrating that sometimes, even grumpy meetings can yield the best outcomes.