The auto industry is speeding up the expansion of artificial intelligence (AI) features, but cases that translate into actual revenue remain limited, an analysis showed. Automakers have invested for years in voice assistants, connected cars and predictive systems, but higher usage is adding to operating cost burdens, making profitability a key challenge.

On May 24, local time, blockchain media outlet Cryptopolitan reported that a live survey conducted during an SBD Automotive webinar found most participants said only about 18 percent of current in-car AI features generate revenue.

Automakers have applied various AI technologies to vehicles, including voice recognition tools, driver prediction systems, digital shopping functions and connected services. But building the technology and generating stable revenue from it are entirely different issues, the analysis said. Robert Fisher of SBD Automotive said, "Car AI itself is not a new concept," but added, "Making AI pay for itself is still very difficult."

The industry’s biggest burden is operating costs. In-vehicle hardware involves relatively little additional cost after initial installation, but AI features incur cloud computing costs each time they are used. Costs rise as voice command processing, route recommendations, predictive functions and connected services are repeatedly called.

Andy Chiu of SBD Automotive said, "This is not a simple technical issue but an income statement issue," and explained that "the core task for car AI is ultimately profitability management." He pointed in particular to a structure in which the more successful AI features are, the larger the cost burden becomes. Chiu said, "Every time users interact with AI features, the cloud meter runs," adding, "This is not a one-off capital investment but operating costs that occur every day."

If AI features fail, only research and development costs remain, and even if they succeed and usage rises, operating cost burdens can surge, the analysis said. In the end, manufacturers are in a position where they must prove that the features generate sufficient returns through subscription fee revenue, higher vehicle sales or stronger customer loyalty, it said.

The problem is that many automakers are not managing the profit-and-loss structure of individual AI features in detail. In many cases it is unclear which features contribute to actual revenue and which erode margins.

SBD Automotive classified car AI features into four types: a "hero" type that secures both customer value and profitability; a "utility" type with high consumer satisfaction but an expectation of free provision; a "zombie" type with low usage but high maintenance costs; and a "grudge" type that harms the user experience. The industry sees the classification as showing that the focus of the car AI race is shifting from simple feature expansion to a stage of verifying sustainable revenue structures.

This profitability pressure is also intersecting with the recent slowdown in the electric vehicle market. Jatco, a parts unit of Nissan Motor, scrapped a plan to produce EV powertrains in Sunderland, Britain, after EV demand in Europe weakened more than expected. Jatco had planned to invest 48.7 million pounds and produce up to 340,000 EV powertrain units a year.

Consumer use of AI is gradually increasing, but trust issues remain. A Cars.com survey found that 44 percent of consumers considering buying a vehicle or who recently bought a new car said they had used an AI-based car search tool. Some 71 percent of respondents said they trusted AI to some extent when it comes to providing vehicle information.

But they have not reached the point of accepting AI recommendations as they are. Even among consumers who use AI regularly, only about half said they were comfortable accepting vehicle and price recommendations. Another 63 percent said they were concerned AI search tools could provide biased recommendations.

In the final verification stage, consumers still prefer vehicle sales platforms and review sites, the survey showed. About two-thirds of respondents said they trusted car sales sites and professional review sites more, and 41 percent said they would visit dealerships or automaker websites in addition even after using AI search tools.

The industry sees the car AI race as now moving beyond simply installing features to a stage of determining which functions lead to actual sales and customer inflows. For automakers, a key task is how to control a structure in which higher consumer convenience and greater usage also mean higher costs.