Menlo Ventures has been an early backer of AI powerhouses including security startup Abnormal, LLM developer Anthropic, and vector database Pinecone. The firm invests more broadly across the consumer, enterprise, and healthcare sectors and is known for historic bets on Chime, Poshmark, Siri, Tumblr, and Warby Parker.

“AI itself is not new, I think it’s just how mainstream it’s become,” Matt Murphy, a Menlo Ventures partner who invests in SaaS and robotics startups, told Business Insider in an interview.

“The Big Bang all started when OpenAI launched ChatGPT, which was the rebirth of AI,” he said.

AI’s “picks and shovels”

Murphy and fellow Menlo Ventures partner Tim Tully say that 2024 will be a big year for the “picks and shovels” of AI. Though they operate in the background, these startups will play an integral role in making AI more powerful and easier for consumers to use.

“With a new wave of technology comes a whole need for new surrounding and supporting infrastructure of tools and capabilities to make the most of it,” Murphy said. He likened the current AI revolution to the data-stack rebirth of a decade ago that ushered in the era of cloud computing and catapulted companies like Snowflake and Databricks to juggernaut status.

“It’s analogous to what’s really going on here in the AI space,” Murphy said. “We have these LLMs, but probably what’s much bigger is a whole set of surrounding tools to help people use these capabilities in a very distinctive way compared to what they had before.”

Tully — a partner focusing on AI, machine learning, the new data stack, and next-generation cloud — added that the volume and speed with which Menlo has seen new companies in the space has “multiplied rapidly,” meaning there are more opportunities than ever before to invest in infrastructure startups that could one day become the next Cisco, Oracle, or AWS of the AI age.

The firm has backed Anthropic in the foundation model layer; Pinecone, Clarifai, and Truera in the tooling layer; and Typeface, Sana, and Lindy in the application layer.

“Fortunately, we’ve been lucky enough to build out a really great AI portfolio that lets us see a lot of stuff both at the application layer and the ‘pick and shovel’ layer,” Tully said.

There’s room in the game for more LLMs

Sam Altman’s OpenAI has been at the forefront of the generative AI revolution. But it’s far from the only LLM in the game — and Murphy and Tully say there’s room for challengers. Menlo Ventures is an investor in Anthropic, which raised $750 million across two funding rounds last year and, at the end of 2023, was in talks to raise an additional $750 million at an $18.4 billion valuation.

Along with smaller startups like Cohere and big-tech LLMs run by Meta, Google, and others, Anthropic is working to give OpenAI a run for its money.

“Our belief is that it’s not going to be a winner-take-all model, and that’s more clear now than when we started,” Murphy said of the LLM arms race, adding that he believes there’s an appetite for a second big LLM to share the stage with ChatGPT.

“You want to be in one of the top two,” Tully said. “The ones we saw most commonly were OpenAI and Anthropic, and between seeing the frequency with which Anthropic is used, and then loving what they’re doing with trust and safety, obviously, the team, it was pretty clear.”

Vector databases

Investors have long been drawn to the buzzy area of vector databases, a growing sector in the AI ecosystem where startups store and catalog vast amounts of otherwise unstructured data such as text, images, and videos.

These vector databases capture the essence of a particular piece of data that a machine-learning program can then pull from, rather than relying on a human-generated tag or label.

The process is integral to the AI programs being used today, Tully explained.

“You literally cannot use OpenAI on its own,” he said. “You have to have a vector database because something has to push context into the query to OpenAI. And where does that context come from? Always a vector database.”

Tully is on the board of Pinecone, a leading vector database startup that raised a $100 million Series B funding round last year at a $750 post-money valuation. Other vector database startups including Chorma and Weaviate have also raised millions from VCs.

“No one talks about it because you just don’t see it, and it’s not front and center,” Tully said. “It’s another ‘picks and shovels’ thing, which sometimes isn’t sexy, but it’s actually the thing that’s driving it all.”