Even three years after the generative AI growth began, most AI startups are nonetheless being profitable by promoting to companies, not particular person customers.
Though customers rapidly adopted general-purpose LLMs like ChatGPT, most specialised client GenAI functions have but to resonate.
“Quite a lot of early AI functions round video, audio, and photograph have been tremendous cool,” mentioned Chi-Hua Chien, co-founder and managing companion at Goodwater Capital, on stage at a TechCrunch’s StrictlyVC occasion in early December. “However then Sora and Nano Banana got here out, and the Chinese language open-sourced their video fashions. And so, plenty of these alternatives disappeared.”
Chien compares a few of these functions to the easy flashlight, which was initially a preferred third-party obtain after the iPhone launched in 2008 however was rapidly built-in into iOS itself.
He argued that, simply because it took a couple of years for the smartphone platform to solidify earlier than game-changing client apps emerged, AI platforms want an analogous interval of “stabilization” for lasting AI client merchandise to flourish.
“I feel we’re proper on the cusp of the equal to cellular of the 2009 -2010 period,” Chien mentioned. That interval was the delivery of large mobile-first client companies like Uber and Airbnb.
We might be seeing inklings of that stabilization with Google’s Gemini reaching technological parity with ChatGPT, Chien mentioned.
Techcrunch occasion
San Francisco
|
October 13-15, 2026
Elizabeth Weil, founder and companion at Scribble Ventures, echoed Chien’s sentiment concerning the early days of GenAI, describing the present state of client AI functions as being in an “awkward teenage center floor.”
What’s going to it take for client AI startups to develop up? Presumably a brand new machine past the smartphone.
“It’s unlikely {that a} machine that you simply choose up 500 instances a day however solely sees 3% to five% of what you see goes to be what in the end introduces the use circumstances that take full benefit of AI’s capabilities,” Chien mentioned.
Weil agreed {that a} smartphone could also be too limiting for reimagining client AI merchandise largely as a result of it’s not ambient. “I don’t suppose we’re going to be constructing for this in 5 years,” she mentioned, indicating her iPhone as she confirmed it to the viewers.
Startups and incumbent tech corporations have been racing to construct a brand new private machine that may supplant smartphones.
OpenAI and Apple’s former design chief, Jonny Ive, are engaged on what’s rumored to be a “screenless,” pocket-sized machine. Meta’s Ray-Ban good glasses are managed by a wristband that detects delicate gestures. In the meantime, various startups are attempting, with typically disappointing outcomes, to introduce a pin, pendant, or ring that makes use of AI in a approach totally different from how smartphones do.
Nevertheless, not each AI client product will probably be depending on a brand new machine. Chien prompt that one such providing might be a private AI monetary adviser personalized to the person’s particular wants. Equally, Weil anticipates {that a} customized, “always-on” tutor will change into ubiquitous, with its specialised tutelage delivered straight from a smartphone.
Although excited by AI’s potential, Weil and Chien expressed skepticism concerning the emergence of a number of, still-stealthy AI-powered social community startups. Chien mentioned these corporations are constructing networks the place 1000’s of AI bots are interacting with the person’s content material.
“It turns social right into a single-player sport. I’m undecided that it really works,” he mentioned. “The rationale that folks get pleasure from social networking is the understanding that there are actual people on the opposite aspect.”