Sonya Huang

I’m a Silicon Valley kid at heart. I was born in Mountain View and grew up in the gravity well of innovation that was 1990s Silicon Valley. It was a fascinating time, from learning to code in third grade to witnessing blockbuster IPOs and the dotcom bubble burst.

Growing up in the Valley made me believe deeply in the American Dream and technology’s role in creating a better future. My parents are immigrant engineers, and technology and entrepreneurship gave them the vehicle to create something lasting. I see venture capital as a conduit to amplify the best founders and the best ideas and give them every possible unfair advantage.

I worked at Goldman Sachs and in private equity before joining Sequoia. But I don’t really think of myself as a “finance person.” I’m much more comfortable reading a technical paper or talking to a researcher than going to an industry conference. I like technical people who are precise and can ship.

I’ve always been very interested in AI, dating back to my college days where I trained computer vision neural nets on brain scans and astrophysics data. It’s awesome, exhilarating, and disorienting just how rapidly the field is accelerating. (PS, I’m very AGI-pilled).

I’m lucky to count some of the top founders and researchers in AI as friends. I’m documenting many of these conversations through Sequoia’s AI podcast (Training Data) and our annual AI event (AI Ascent).  I believe we are living through a historical moment in time that deserves to be explored and preserved with great curiosity and care.

We’re approaching the event horizon that technologists have dreamed of for decades. We recently passed the Turing Test, and nobody even noticed. AGI is within reach. Things are about to change profoundly and rapidly.

I met the OpenAI team more than a year before ChatGPT launched, when they were still a pre-product research lab. What struck me wasn’t just their ambition—it was their conviction and deep-held belief in how scaling laws would unfold, and the guts to just go for it.

Glean came out of a different instinct. Arvind and I spent time together in the early pandemic, when “search at work” was still a neglected problem. He talked about enterprise knowledge the way a systems engineer talks about memory access: fast, invisible, and vital. Glean wasn’t just a better search box—it was a bet on how AI could shift the speed and shape of knowledge work–and that as intelligence becomes almost too cheap to meter, pipes and context become king.

LangChain and Fireworks are building the scaffolding of AI-native software, empowering the next wave of applications. LangChain’s insight was that chains of reasoning and tool use, not just single calls to a model, are what unlock complex behavior. Fireworks is the underlying infra to make that reliable, fast, and deployable at scale, responsible for powering 5 trillion inference tokens a day including beloved AI apps like Cursor and Perplexity. Together, they form the LLM stack that developers will build on for years.

Much of my work today centers on one belief: we’re living through a singular moment in time. We’re entering a fast takeoff moment–a period where systems that learn continuously will outstrip those that don’t. I’m looking for teams who know how to wield that power in building the companies of the future.

Get in touch with Sonya