Beyond the Models: Where Investors Are Finding AI’s Next Wave
Getting in on the ground floor of the next AI wave can feel particularly elusive for investors. With billions continuing to pour into frontier models and the infrastructure powering them, much of that opportunity is already spoken for. For investors looking beyond hyperscalers and data centers, the question is no longer whether AI will reshape industries but where they can still find meaningful value.
To unpack where opportunity still exists, The Information’s Stephanie Palazzolo spoke with:
- Alex Baker, partner and U.S. TMT deals leader, PwC
- Alexa von Tobel, founder and managing partner, Inspired Capital
Surviving in a World of Giants
While it may seem like every company is an AI company, most of the growth is concentrated in a handful of tech giants. For Alexa von Tobel, founder of early-stage fund Inspired Capital, this underscores the importance of defensibility and building something Google or Microsoft can’t easily replicate.
“Those companies have over a billion users. You don’t want something where they can deploy even a halfway-decent version of what you’ve built and eradicate it overnight,” she said, adding that in a world of elephants, “you don’t want to be the ant.”
PwC’s Alex Baker, who works with clients from venture capital all the way through initial public offerings and megafunds, agreed, expanding on the main factors that enable companies to hold their own.
“Defensibility is about entrenchment—how embedded is an asset to an enterprise environment; replicability—what is the underlying tech infrastructure; and trust—what do customers want and how do they use the technology we’re talking about,” he said.
The Limits of Vertical AI
Last month, AI-powered software company Cursor gave SpaceX an option to acquire it for $60 billion. For Baker, deals of this kind point to a shrinking moat in the vertical AI space.
“For a lot of very highly funded startups, their differentiation has been, ‘We train our own models, and it’s vertical specific already.’ I would argue that differentiation is eroding,” he said.
Baker argued the real moat is the orchestration layer within a vertical. Von Tobel, meanwhile, has mostly avoided vertical AI companies with one exception: Frontera Health, an autism-focused large language model.
“They’re working with a very specific, HIPAA-compliant dataset, where the data needs to be collected safely, and Anthropic and others likely weren’t going to run after [it],” she said.
Beyond Software: AI Meets the Real World
One of the biggest areas of AI promise, agreed Baker and von Tobel, is in the physical world, which offers a far more durable moat than software’s fragile barrier. For one, said von Tobel, physical AI models are harder to replicate than software ones. One example, she said, is BrightAI, a company in her fund she describes as Cursor for infrastructure workers. It puts chips on everything from water pipes to telephone poles.
“Talking about moats, taking off hundreds of thousands of stickers is a real pain,” she said. Unlike software, these systems are difficult—and costly—to replace once deployed. The company is also capturing data no one else is collecting, another defensible advantage.
“I think physical AI is going to be one of the most important trends over the next four years,” said Baker, adding that it’s reinvigorating industries like physical security.
“Video cameras, badge readers, door locks—this used to be really old technology, but now it’s a technological space, and it’s early in the curve,” he said.
Is Software Dead?
As tools like Cursor and Claude make it easier to vibe code custom software, some investors now question whether traditional software still has a place. Baker, for one, is optimistic. While AI is introducing real disruption, he said the risk of wholesale replacement is often overstated—particularly in regulated, mission-critical workflows with high switching costs.
Moreover, in a future where AI agents—not humans—are the primary users of software, operating 24/7, the most valuable platforms could become even more central.
“There’s a little semantics here, right? What’s a software versus a vertical-specific, AI-enabled orchestration layer?” he asked. “Our view is that for many software players, there is as much, if not more, opportunity right now than downside.”