Inside Sequoia Capital’s Growing AI Portfolio

Editor’s note: This is part of a series in which we interview active investors in artificial intelligence. Previous interviews were with investors at General Catalyst, Bessemer Venture Partners, Accel, Insight Partners and Index Ventures. 

A year ago, about 16% of Sequoia Capital’s new investments were in artificial intelligence startups. So far in 2023, that number has jumped to nearly 60%.

“AI has brought new life to the investing ecosystem in the last year,” said Stephanie Zhan, a partner at the firm who invests in seed and early-stage companies.

Crunchbase News spoke with Zhan about Sequoia’s AI investments, which the firm says now span 70-plus portfolio companies.

Investing early

“We’ve seen more creation and formation of new AI companies,” said Zhan on the firm’s strategy to follow developers and talent. “We’ve seen the rise in early-stage AI investing — most notably, pre-seed and seed-stage AI companies that we’re actively investing in right now.”

“Every company will have some kind of AI angle around it, and it won’t be seen as an AI company,” Zhan said. “It will be seen as a company solving a particular domain or problem space that they’re focused on, with AI as an accelerator.”

AI in the Valley

Zhan has enjoyed a front-row seat to early developments in AI. While studying computer science at Stanford, she was advised by Andrew Ng, co-founder at Google Brain, adjunct professor at Stanford, and the current CEO and founder of Landing AI.

While a student, computer vision and self-driving cars were at the forefront of the industry, she said. Nest, where she worked in product development, acquired Dropcam, which had image detection technology to “identify the cat on the dropcam.”

Zhan joined Sequoia in 2015, the year OpenAI was founded. At the time, OpenAI rented office space from Sequoia above Dandelion Chocolate in San Francisco’s Mission District. She followed the startup’s research, tracking AI’s possibilities beyond computer vision.

Sequoia made a growth investment in OpenAI in 2021.

In 2015, it seemed AGI (artificial general intelligence) — a hypothetical form of artificial intelligence with human-like cognitive abilities — wouldn’t be possible in the span of our lifetimes, Zhan said. Now, with the advent of large language models and ever-more advanced AI systems, AGI at least seems within striking distance.

“There are interesting areas of innovation beyond the current transformer architecture,” Zhan said. “Will there be players who do the same thing, but for other very specific domains?” To that end, the Sequoia team has seen companies focused on audio, multimodal inputs and biology.

Zhan is also excited by open-source initiatives such as LLaMA, the large language model owned by Meta, and the company’s Code Llama tool for coding assistance — both comparable alternatives to OpenAI’s GPT-4.

Sequoia in AI

Sequoia says it has around 70 portfolio companies in the AI space, with 17 of those still in stealth. These span from seed to public companies across all industries from semiconductors — Nvidia is a portfolio company from the 1990s  — to biotechnology, robotics to autonomous vehicles, and creation to collaboration tools.

The firm’s 2023 announced AI investments include Harvey, AI technology for legal teams; AI assistant Dust; and Replicate, a cloud technology to help developers run machine models important for the open-source movement.

Other notable investments the firm has made in recent years include backing generative video producer Tavus, Hugging Face, which operates a hub for building art models, and Glean, which offers an AI workplace search product that allows employees to search across their company’s apps.

Zhan noted Sequoia was also an early investor in Google, co-leading its Series A in 1999. In 2011, Google founded Google Brain, its AI research effort, which was acquired by U.K.-based DeepMind in 2014. The company authored the transformer model, which revolutionized the output of large language models and ultimately led to the rise of generative AI technologies such as ChatGPT.

Shifting tides

Sequoia has made some big changes to its structure in recent years. In November 2021, it announced a change to an evergreen fund model to continue to hold investments in enduring public companies. In retrospect, it was a badly timed strategy shift announced within months of an extended downturn in technology stocks.

However, in an interview with Crunchbase News in early 2021, Sequoia partner Roelof Botha noted how “patient” the firm is with distributions, often waiting years after a company has gone public to distribute.

The new structure codified this practice of maintaining stakes in “… these handful of companies that really surprised you to the upside,” he said then, citing the tens to hundreds of billions of dollars in value creation by public technology companies Zoom, Airbnb, DoorDash, Square, PayPal and Google — all Sequoia portfolio companies.

In June of this year, amid rising U.S.-China tensions, the firm announced its split into three separate firms for China, South East Asia and its Europe/U.S. business.

Related Crunchbase Pro list

• Sequoia Capital AI Portfolio Companies

Editor’s note: This article was updated on Oct. 23, 2023, with several clarifications, including the timing of Sequoia’s investment in Replicate as well as the nature of the firm’s split into three businesses and its public company holdings.

Illustration: Dom Guzman