What We Learned About AI From 7 Active Startup Investors This Year

Editor’s note: In 2024, Crunchbase News interviewed active startup investors in artificial intelligence. Below, we publish highlights from those interviews. Read the full interviews with Felicis, Battery Ventures, Bain Capital Ventures, Menlo Ventures, Scale Venture Partners, Costanoa and Citi Ventures as well as highlights from more interviews we shared in 2023.

This was the breakout year for funding to AI companies. Venture funding through mid-December reached $99.6 billion — up 80% year over year — Crunchbase data shows. Close to a third of that funding went to foundation model companies.

The other two-thirds of funding went to sectors impacted by these new models. Separate from the hardware and data provisioning to manage and operate AI, leading sectors included autonomous driving, healthcare, robotics, professional services, and marketing and sales, Crunchbase data shows.

With billions of dollars flowing to these new technologies, we spoke with more than a half-dozen earliest- and latest-stage investors this year to understand where they are placing their bets when it comes to AI.

We heard AI is the focus for the majority of these firms’ new investments, as cloud computing was in the prior era. Scale Venture Partners’ Rory O’Driscoll estimated that some 80% of the firm’s new investments this year were oriented around AI.

Applied AI

In 2023, the picks and shovels for AI — the companies in the data infrastructure layer to set up and manage these models — were a dominant part of the conversation. In 2024, the aperture for many investors — while still investing in infrastructure — noticeably shifted to thinking through AI applications that significantly impact a sector.

“I think the real money is going to be made by figuring out which apps are going to get done first,” O’Driscoll said. “Part of that is to understand how enterprises judge success.”

And with that, “how willing they are to live with the intrinsically probabilistic nature of AI — in other words, how catastrophic is failure?”

“There’s a whole range of human tasks that AI can now do at some level of merit, and I think a lot of the success will be about judging when it’s good enough and when it’s not,” said O’Driscoll. He has invested in applied AI companies Regie.ai, Tavus, Bland AI and Klarity.

Assisted AI

The concern about accuracy was echoed by Felicis 1 general partner Sundeep Peechu, who predicted “a lot of services work that might be semi-automatable and with a human at the very end, basically adding a level of oversight.”

Multimodal AI is coming next, Peechu said in an interview. That technology integrates text, image, sound and video or some combination of those into one interface.

A new type of fund

Menlo Ventures partnered with Anthropic, one of the biggest generative AI startups in the world, creating a $100 million AI dedicated fund named Anthology to invest in early-stage startups using AI.

“People want to be a part of Anthropic right now. It’s playing the hot hand,” Tim Tully, a partner at Menlo Ventures, told us in an interview.

Startups in the fund get an allocation of free credits to Anthropic’s models — and possibly even more valuable, said Tully, is access to Anthropic leadership.

The Anthology fund has announced it made 18 investments, so far, eight of which are still in stealth.

Increased productivity

AI isn’t a feature. It’s not even a sector, if you ask Dharmesh Thakker, a general partner at storied venture firm Battery Ventures.

“I look at AI as a fabric,” Thakker told us.

“Just in front of our eyes, we’re seeing companies that are growing 20% to 30%, can now be cash-flow break-even almost immediately, or within 12 months, by applying generative AI to sales and marketing, R&D, recruiting and other areas of the business,” Thakker said. “And so, we’re seeing this innovation fully impact productivity in the tech sector.”

Another large cost center, research and development, is materially impacted. “You can generate output from expensive software engineers making $300,000 to $400,000 a year using Copilot. You don’t need so many software engineers,” Thakker said.

In other sectors such as healthcare, for instance, the impact might be a lot more profound, he said.

Increased revenue

In software, more dollars are spent on services integrating technology solutions than on products, Rak Garg, a partner at Bain Capital Ventures, said in our interview with him.

Generative AI presents an opportunity to turn service revenue into product revenue.

“These services are taking data from one place, molding it in different ways, putting it in a different place,” he said, citing Unstructured as an example. “Models are really, really good at that.”

AI application companies are beginning to break out with real revenue, said Matt Carbonara, managing director at Citi Ventures, the venture investment arm of banking giant Citigroup.

“They’re pre-built and just easier. Enterprises are still learning. It’s a new skillset,” Carbonara told us. “How do you deploy it? How do you get the accuracy you want? And it’s an iterative process, versus somebody coming to you with a pre-baked pie.”

Going vertical

SaaS has been the dominant tech force of the past 20 years, but is now giving way to AI, several investors told us.

“We have literally solved most of the problems that could be solved by that generation of technology,” said Greg Sands, founder of Costanoa Ventures.

The firm believes that generative AI has now opened up vertical opportunities that might have been too small in the past, Sands told us.

“The knock on vertical SaaS has always been: it’s a smaller total addressable market because you limit yourself to one vertical, rather than building something that can apply to every vertical,” John Cowgill, a general partner at Costanoa, said in our interview.

The firm is looking closely at each vertical to understand how big it can be with the rise of AI.

Agentic is coming

The conversation on agentic AI is less than six months old, but points to the creation of new native AI companies.

“The real value in AI comes when you can own an end-to-end workflow,” said Cowgill. “This is the idea of an agent moving from just applying AI, pointing at something and saying, ‘search across it with AI, summarize with the AI’ — to own a job to be done with AI. Getting agents to work is incredibly difficult.”

It takes a decade

While investment dollars to AI are up in 2024 relative to 2023, it will take more investment cycles for the AI revolution to fully take hold.

“There is probably going to be a period in the next two or three years where we all get a little disappointed at the traction, and there’s a little bit of a correction,” said O’Driscoll. But, “zoom out 10 years, no one’s going to be building software without AI at the core.”