Sequoia sees more money in AI-related software than in models

(Bloomberg) — Venture capital giant Sequoia Capital believes most billion-dollar artificial intelligence companies will be built by building applications rather than models, though the company is investing in both sectors, partner Pat Grady said at a conference on Wednesday.

Speaking at a Goldman Sachs Group Inc. event in San Francisco, Grady said the firm has invested about $150 million in companies that build foundational models, such as Sam Altman’s OpenAI, Ilya Sutskever’s Safe Superintelligence Inc. and Elon Musk’s xAI. These models, which are the building blocks of many of the latest AI companies’ products, are incredibly expensive to build.

Grady admitted that given the more than $55 billion Sequoia manages, the company spends relatively little on such ventures.

“We’ve invested an order of magnitude more dollars in the application layer, even though the revenue generated in the application layer is much smaller,” he said, referring to AI companies that build products that use models in novel ways. But in AI, “we think the application layer is where the largest number of billion-dollar companies are going to emerge.” Grady didn’t provide a dollar amount of Sequoia’s spending on other AI companies.

One factor driving AI overspending is what he called the “science-awe effect,” when investors tell themselves a startup will succeed because “this is so cool. These people are so smart.” Sequoia tries to avoid such assumptions, he said.

He also praised OpenAI’s upcoming version of the Strawberry model, saying it is “really good.”

Grady has invested in AI companies such as Harvey, a legal services firm, and Notion Labs Inc., a productivity tools company.

For more stories like this, visit bloomberg.com