Self-driving cars were supposed to be NVIDIA's (NVDA 3.60%) big play, but it seems to have lost the initiative in this space as the competition has intensified.

The graphics specialist had created a lot of hype about how its GPUs (graphics processing units) could power the cars of tomorrow by computing tons of data taken from cameras mounted on vehicles. And it showed early promise by scoring as a customer electric-vehicle maker Tesla, which decided to use NVIDIA chips to power infotainment systems and digital instrument clusters. Tesla also promised that it would put NVIDIA's supercomputers into its cars to power self-driving capabilities, but things went south soon after.

Person reading a book while sitting behind the wheel of a self-driving car.

Image Source: Getty Images

What went wrong?

Intel (INTC 4.08%) took away the Tesla business from NVIDIA last year. At the same time, it emerged that Intel is the chip supplier of choice for Alphabet's self-driving car division, Waymo, which is leading self-driving car development by a wide margin. So, its use of Intel chips was a big endorsement for Chipzilla's self-driving capabilities, and NVIDIA failed to hold on to its early mover advantage.

Not surprisingly, the company's automotive growth started flat-lining as Intel hogged the limelight and went about striking crucial partnerships to tap this space. NVIDIA generated $145 million in automotive revenue during the first quarter of fiscal 2019, a shade above the $140 million it generated in the year-ago period.

In hindsight, NVIDIA's failure to land Waymo has dealt a blow to the chipmaker as Alphabet's self-driving car division is getting ready for a commercial launch and is planning to attack the ride-sharing market. This win could have made a big difference for NVIDIA's automotive business, but it isn't too late just yet as self-driving cars are yet to gain critical mass.

NVIDIA's new automotive strategy

It is still early days in the self-driving car race, and NVIDIA has decided to change its approach to tap this market. Earlier, the company was focusing on supplying solutions for vehicle infotainment systems, but it has now realized that this space is getting commoditized.

NVIDIA is now focused on selling end-to-end systems, integrating both its hardware and software into a single package that will power the car as well as the systems inside the cockpit. The good part: this strategy seems to have struck the right chord with automotive companies and component suppliers, as evident from the company's latest partnerships.

Mercedes-Benz parent Daimler and automotive component supplier Bosch have decided to tap NVIDIA's DRIVE Pegasus artificial intelligence (AI) computing platform to develop self-driving cars. More specifically, Bosch and Daimler will use NVIDIA's processor and GPUs for managing a network of electronic control units (ECUs) inside an autonomous car, along with using the graphics specialist's AI software to run the autonomous algorithm and help the car drive on its own.

The alliance intends to start testing autonomous S-Class sedans on California roads in the second half of 2019, and aims to launch cars with Level 4 and Level 5 automation capabilities in the next five years. What's more, Bosch and Daimler revealed in April that they plan to tap the ride-sharing market with their fully autonomous vehicles.

This is good news for NVIDIA as this partnership could be its ticket to a fast-growing industry that's expected to grow eight-fold by 2030, according to Goldman Sachs. But this isn't the only way NVIDIA could make a dent in the burgeoning ride-sharing market, as its self-driving platform was selected by Uber earlier this year.

Uber has placed an order for 24,000 Volvo SUVs to create a fleet of autonomous cabs, with the cars expected to be delivered between 2019 and 2021. It is likely that they will contain NVIDIA's self-driving system .

NVIDIA is also partnering with other automakers such as Audi with plans of putting a self-driving car on the road by 2020, though there has been no recent update on this front. Daimler and Bosch could take five years to achieve their ambition of getting a Level 4 or Level 5 capable car on to the roads, provided there are no glitches, while Uber's development won't materialize before 2021 at least.

By comparison, Intel seems better-placed to realize its self-driving dreams since Waymo plans a commercial launch before the end of 2018 and to scale up its fleet by 2020. As such, NVIDIA's automotive struggles will continue until one of its customers hits gold and starts producing self-driving cars on a mass scale, and that's unlikely to happen within the next two or three years.