The AI ​​startup market is vast, from companies looking to create new chips, to those using AI to create robots, to others looking to use AI to create solutions to niche for industry-specific workflows. There are many potential areas for venture capitalists to invest in, but there are clearly some subsectors that they are more excited about than others.

TechCrunch recently survey 20 VCs investing in startups looking to sell businesses about their predictions for 2025.

Mark Rostick, a vice president and senior managing director of Intel Capital, told TechCrunch that now that the big foundational models have been established – at least in his opinion – the next interesting area to invest in is AI solutions. for specific tasks.

“I find models that excel at specific functions to be particularly interesting, especially when combined with agents built on top of them,” Rostick said. “As AI adoption accelerates, application-focused companies will take center stage, as CEOs increasingly seek ways to use AI in specific areas that will deliver tangible, transformative effect.”

This was echoed by Mike Hayes, a managing director at Insight Partners. He added that he would look at supporting companies that build products that use AI to reduce friction in business.

“I look for solutions that solve unique, orthogonal challenges for businesses – places where traditional solutions fall short,” Hayes said. “This includes vertical and persona-specific workflows reimagined with GenAI or agentic automation and security innovations that not only identify and alert, but also heal.”

VCs interested in finding companies that focus on specific business use cases should make sure that these startup solutions are real companies, as opposed to just parts. If not, we will see a repeat of the SaaS boom in 2021, when many companies that were only a one-note segment raised large amounts of venture capital in the past. for being abandoned in favor of companies that offer platform solutions when business budgets contract in 2023.

There are of course tasks important enough to warrant a partial solution. For SaaS, we hear that businesses will still pay for companies that offer specific cybersecurity solutions. For AI, at what point solutions businesses are willing to pay for are still unclear. Ed Sim, the founder and general partner of Boldstart Ventures, recognizes this challenge.

“The trick is to skate where the puck is and also think about whether it’s a part, or a product, or a business,” Sim said.

Another area that VCs are excited about is reliability and stability. Jason Mendel, an investor at Battery Ventures, said he is looking to invest in companies in the observation and reliability space. Liran Grinberg, the co-founder and managing partner of Team8, also focused on what he called “the strength of the business.”

“The Crowdstrike software update incident shows how vulnerable our digital world is, not only to cyber attackers but also errors,” Grinberg said. “We need a more resilient, anti-vulnerable digital infrastructure by design.”

AI infrastructure will also remain a hot investment area in 2025. VCs cite that with advances in AI agents, they are looking at the infrastructure required for businesses to adopt the technology in addition to companies that help determine the price for AI agents as well.

“It’s very early innings here, and I believe the momentum for AI infrastructure will continue until 2025, especially as agent frameworks proliferate, new modeling paradigms (including reasoning) are developed, “-AI content is evolving, and the UI/UX of AI applications is evolving (including computer usage),” Janelle Teng, a vice president at Bessemer Venture Partners, said.