LAS VEGAS, Nevada: After years of expensive false starts, the self-driving car industry is regrouping around artificial intelligence and partnerships, as chipmakers, tech firms, and auto suppliers look for a more practical path toward autonomy.

At CES in Las Vegas this week, a series of tie-ups underscored that shift. Rather than betting billions on building fully autonomous systems alone, companies are increasingly pooling resources, sharing platforms, and leveraging AI to cut development costs and accelerate progress, even as automakers remain cautious about demand and profitability.

The promise of cars that drive themselves has long been transformative, but delivering that technology safely on public roads has proven far more complex and more expensive than early boosters predicted. Only a handful of companies, including Alphabet's Waymo and Tesla, have persisted with largely in-house approaches. Others, such as General Motors and Ford Motor, have abandoned their own fully autonomous vehicle units after years of losses.

Now, suppliers and tech firms are stepping into the gap.

At CES, Amazon Web Services announced a deal with German supplier Aumovio to support the commercial rollout of self-driving vehicles. Autonomous trucking firm Kodiak AI said it had partnered with Bosch to scale up manufacturing of self-driving hardware and sensors. Nvidia unveiled its next-generation automotive platform, which will be used in a robotaxi alliance involving Lucid Group, Nuro and Uber.

Mercedes-Benz also said it would launch a new advanced driver-assistance system in the United States later this year, powered by Nvidia's chips, that allows vehicles to operate autonomously on city streets under driver supervision.

Artificial intelligence itself is becoming a critical development tool, not just the end product. By simulating scenarios and validating systems virtually, AI can reduce the need for expensive real-world testing.

AI and generative AI are acting as a "big accelerant" for the industry because it actually allows ... a significant amount of development and validation with significantly fewer resources," said Ozgur Tohumcu, general manager for automotive and manufacturing at Amazon Web Services.

Western automakers are also under pressure from China, where authorities last month approved two vehicles with Level 3 autonomous capabilities, allowing hands-off driving in certain conditions. The industry defines autonomy on a five-level scale, from basic driver assistance at Level 1 to fully self-driving at Level 5.

Still, many executives warn against overhyping the pace at which fully autonomous cars will become commonplace.

Jochen Hanebeck, CEO of German chipmaker Infineon, cautioned against what he called "market fantasy" around rapid progress to Level 5 autonomy. Automakers, he said, are more focused on revenue-generating Level 2 driver-assistance systems that require constant human attention.

"I don't see, really now, a tsunami flowing towards Level 5," Hanebeck said.

There has been a recent flurry of small robotaxi deployments announced across China, the United States, Europe, and the Middle East. But scaling those services to cover larger areas requires vast amounts of data, vehicles, and logistical support — "which is costly and expensive," said Jeremy McClain, head of system and software at Aumovio's autonomous mobility unit.

The industry's long history of hype still looms large. Tesla CEO Elon Musk famously predicted in 2019 that a million self-driving Teslas would be on the road within a year. Tesla only launched a limited robotaxi trial last year, six years later.

One enduring challenge is the sheer number of "edge cases" vehicles must handle. A human driver who sees a ball roll into the street instinctively slows, anticipating that a child might follow. A self-driving car reacts only once it detects the child.

Past failures have been costly. General Motors shut down its Cruise unit after years of losses, a process accelerated by an incident in which a Cruise vehicle struck and dragged a pedestrian.

Nvidia argues AI has now reached a point where it can address some of those weaknesses.

"There's some foundational pieces of technology that make us feel like we're there," said Ali Kani, general manager of Nvidia's automotive team.

Morgan Stanley analysts said Nvidia's new Alpamayo platform could give traditional automakers a boost and help them compete more effectively with Tesla, even if the electric vehicle maker remains years ahead.

Others see a familiar tech pattern emerging.

"In one way, you could almost see Apple and Android playing out," said Russell Ong, former product lead at self-driving vehicle maker Zoox, likening Tesla's proprietary approach to Nvidia's open-source strategy that could unite its rivals.