dailynewsnblog
The auto industry has been experimenting with AI-based features for years with voice assistants, predictive features, and connected services. The problem is that most of it doesn’t pay.
That was the main finding at a webinar hosted by research firm SBD Automotive, with the theme “The Profitability Path of Automotive AI,” that featured panelists like cloud providers, AI specialists, and automotive strategists.
SBD reported that live polling during the session found that the vast majority of attendees said fewer than 20% of their AI features are profitable, a shockingly low figure given how aggressively the industry has invested in the technology.
“AI in automotive is nothing new,” moderator Robert Fisher of SBD Automotive noted. “But making AI pay for itself is still very difficult.”
The issue isn’t technical capability, as automakers have demonstrated they can build AI systems. The problem is the economic structure.
“This is not a technology problem,” Andy Qiu, senior manager at SBD Automotive, said. “It’s a P&L problem.”
For decades, automotive economics was simple: spend upfront on designing a feature, ship the car, lock in costs, and eventually profit after sale. AI, in particular, cloud-based AI, doesn’t follow those rules. Features don’t have a fixed cost of goods; costs keep increasing.
“Every time a user interacts with an AI feature, your cloud meter is running,” Qiu explained. “That’s not capex anymore. That’s ongoing opex every day, forever.”
The more successful a feature becomes, the more expensive it is to operate. And most automakers lack the per-feature cost visibility to even know which features are bleeding money, Qiu said.
The result is what he called a portfolio full of “zombie features,” AI capabilities that look impressive in a press release but generate minimal usage while quietly draining margins.
Qiu categorized AI features into four types: Heroes (high value, profitable, worth scaling), Utilities (valuable but expected for free), Zombies (costly and rarely used), and Grudges (poor experiences that actively frustrate users). His unsettling conclusion: Most OEM portfolios skew toward the bottom two categories.
“The biggest opportunity isn’t building more features,” he said. “It’s killing the wrong ones.”
Dani Cherkassky, CEO of Kardome, located the root of the profitability problem in user experience. Cloud-dependent voice assistants are slow, context-blind, and disconnected from natural conversation, and users won’t pay for things that frustrate them.
