Venture capital investors have been clamoring to get in on artificial intelligence deals for months, and many of the biggest, priciest bets have been on so-called “foundation model” companies, like OpenAI and Anthropic. But that’s not where some of the best opportunities could be moving forward.
For Seth Rosenberg, an investor at VC firm Greylock Partners, there’s one underrated area that he’s eyeing for A.I. investments. “People talk about, 'Is the best opportunity for A.I. in startups or incumbents?' I think there's opportunities in both, but I think there's a third category that's maybe a little bit overlooked, which is companies that are maybe one to two years old, that have an amazing product or some amazing value that can be significantly accelerated with A.I.,” he told me. “In some ways, they're really at a sweet spot, because they're more nimble than an incumbent to…totally change their business model, totally change how the product is built to be A.I.-optimized, but they have a head start versus a brand new seed-stage company.”
Rosenberg reasons it’s easier to incorporate A.I. technology into an established (but still young) business for something like, say, mortgage origination, than the other way around. A fledgling startup “would need to spend the next two years figuring out underwriting, getting capital markets partnerships, getting servicing contracts, and...two years from now, then they would start adding on A.I. to automate the process,” he said. But Greylock’s portfolio company Pine, a Canada-based digital mortgage lender, has “already done all that.” While A.I. wasn’t part of the investment thesis when the firm invested in Pine in 2022, Rosenberg says it's now a big part of their strategy moving forward.
It all plays into Rosenberg’s thesis around competitive advantage in A.I. He says he views it in two ways: Is the main value of the business derived 80% from the large language model the startup is using, or 80% from something else? For companies like Pine, Rosenberg argues that 80% of their value comes from their fintech business versus what A.I. model or technology they use. That formula—80% fintech + 20% A.I.—can make these companies hard to compete with, he thinks.
But I wonder, what’s to stop the incumbents—be it Big Tech or in financial services—from just buying up these startups or A.I. teams to do it themselves? And aren’t pretty much all companies talking about using A.I. now?
When I asked Rosenberg, he theorized that even if these incumbents are able to acquire A.I. talent, it’s harder for them to totally overhaul (in his words, “destroy”) how their business currently works. But in a larger sense, Rosenberg believes investors “over intellectualize” the competition question. “If you think of the massive consumer technology platforms that have been built over the last 10 years, most of them don't have, like, a crazy [intellectual property] moat,” he says.
Though they haven’t made any recent investments in these types of companies, Rosenberg says, he’s been looking into it in addition to helping Greylock’s current portfolio A.I.-ify themselves. Whether or not Rosenberg’s strategy would pay off, it could have one major benefit: Investing in startups that are still branded as a “fintech” or “healthcare” company could help the firm avoid paying an inflated A.I. premium.
See you tomorrow,
Anne Sraders
Twitter: @AnneSraders
Email: anne.sraders@fortune.com
Submit a deal for the Term Sheet newsletter here.
Jackson Fordyce curated the deals section of today’s newsletter.
