Investor predicts 85% of AI startups will die in three years

The generative AI gold rush is in full swing, but one veteran venture capitalist is already measuring the coffins.

Ken Smythe, founder and chief executive of New York-based Next Round Capital Partners, just issued a brutal reality check for the booming sector. His verdict: 85% of artificial intelligence startups will be out of business within three years.

According to Smythe, the overwhelming majority will fail for two predictable reasons: they will simply run out of cash, or they will be swallowed up by tech giants looking for easy acqui-hires.

No balance sheets, just pedigree gambling

Next Round Capital has already made its bets in conversational AI, including backing virtual call-center agent developer Uniphore. Yet Smythe admits that regular venture investing playbook does not apply right now.

Traditional financial metrics are virtually nonexistent. Most of these new startups have no commercial track record and zero balance sheets for analysts to pore over.

Instead, VCs are forced to gamble almost entirely on the founder's resume. Smythe noted he was recently pitched by a co-founder of Apple's Siri. When you cannot audit the business, you end up betting on whether the engineer who built frontier tech elsewhere can actually turn it into a profitable commercial product.

Far bigger than the dot-com boom

While critics frequently compare current AI excitement to the late 1990s tech bubble, Smythe sees a much deeper transformation.

The dot-com boom mostly introduced easier ways to do existing tasks, like shopping online or booking travel. AI, by contrast, is positioned to directly replace everyday human labor and displace entire industries outright.

Between bots replacing call centers and autonomous robots working 24-hour construction shifts, Smythe argues the technology will create massive efficiency and slash operating expenses across every sector.

And while 85% of contenders might crash, he expects clear category leaders to emerge across verticals like healthcare, marketing, and autonomous driving—much like e-commerce spawned vertical winners alongside mega-retailers.

Why it matters

Hype cycles always make it hard to tell true paradigm shifts from vaporware, and Smythe's warning cuts through the noise with cold math.

For builders and enterprise buyers, relying heavily on a niche AI wrapper right now carries huge platform risk. If a tool you integrate into your daily workflow lacks a sustainable business model, there is an 85% chance it will either go dark or get gutted inside an acquiring tech conglomerate before your multi-year roadmap wraps up.

Across the industry, this underscores how dangerously disconnected early-stage valuations are from real cash flow. The days of raising rounds on demo videos and prestigious ex-Big Tech resumes will soon hit a wall when follow-on capital demands actual customer retention.

We think Smythe is right that AI's macro staying power is real, but the bridge between raw capability and sustainable unit economics is far narrower than current excitement suggests. Surviving the upcoming shakeout will matter a lot more than winning early headline buzz.