Inworld CEO cuts AI prices for consumer startups

The cost of running AI models has become a serious barrier for consumer startups, and one major AI voice company is trying to help by cutting prices by more than half.

Kylan Gibbs, CEO of Inworld, says that many young companies are stuck spending 70% to 90% of their operating budgets on inference costs — the computing power needed every time a user talks to a chatbot or voice assistant. At the same time, those startups often charge users just $5 to $10 a month, making profitability nearly impossible as engagement grows.

“Cost is the single number one problem,” he said in a recent interview.

Consumer AI startups face a unique economic trap. In a typical software business, unit costs flatten or drop as usage scales. But with AI models, each new interaction requires fresh computing power. The more users love a product, the more money it burns.

“Their profitability goes through the floor every time they hit success,” the CEO said. “We see that across lots of AI consumer applications.”

Why consumer AI startups can’t catch a break

The problem centers on AI inference. Large AI companies like Google or OpenAI own their own infrastructure, can negotiate better chip pricing, and spread costs across massive businesses. These young firms don’t have that leverage. They end up paying more per unit of intelligence for access to models.

As a result, many hit a growth ceiling. Some stop marketing. Others pivot to selling to businesses instead of consumers.

Some simply disappear.

The dynamic ultimately benefits the biggest players, which can copy popular features and drop them into existing products with no extra overhead.

“The irony is that users love these products,” Gibbs said. “Time spent keeps growing. But every time they become successful, their profitability falls.”

Inworld, one of the leading providers of voice-based AI models, is now trying to change that. He announced that the company will cut prices by more than 50% and offer deeper discounts as customers’ usage scales.

Some industry observers caution that lowering model prices alone won’t erase the structural cost advantage that large cloud providers and AI labs hold. But the firm is betting that cheaper infrastructure will unlock a new wave of consumer applications — in areas like education, therapy, health, and fitness — that couldn’t survive under the old pricing.

Why inference pricing is still out of whack

The broader market, Gibbs said, is running on inflated prices. Inference providers and AI model developers often set their rates relative to what competitors charge rather than what the compute actually costs.

That can generate revenue many times the underlying expense.

“If we want AI benefits to reach everyday people, the economics have to work for the companies building those products,” he said.

Inworld has raised more than $117 million and reports that revenue has grown fivefold since the start of 2026. Its bet is that by making its voice models more affordable, it can help a generation of startups avoid the spiral that has forced so many to slow down or switch directions.

For now, the economics of consumer AI remain lopsided. A corporate customer might pay $1,000 a month for an AI tool and compare that cost with hiring a person. Consumers are far more price-sensitive and sometimes more fickle about which apps they stick with. Inworld’s price cuts won’t fix that gap entirely — but they could give a few more startups room to breathe before they hit the ceiling.