clock menu more-arrow no yes

Filed under:

Nvidia's plan for the car isn't as crazy as it sounds

New, 14 comments

An excess of ambition and a poor CES presentation are clouding what's actually a sensible strategy for the leader in computer graphics

The company that coined the term Graphics Processing Unit (GPU) at the turn of the century has been at the forefront of PC graphics cards for almost as long as they’ve existed. But at CES this week, Nvidia isn’t selling LED-embroidered, heatpipe-cooled GPUs; it’s pitching an all-encompassing vision of the future connected car. It doesn’t seem to make sense — and in its full, blooming, overreaching glory, it really doesn’t — but a more tempered, less ambitious version of the plan announced by Nvidia at CES actually makes perfectly good sense.

Nvidia is selling the same thing it always has: massive parallel processing

Firstly, let’s be clear about what Nvidia is doing. In spite of the deluge of technical jargon and the painfully academic presentation on Sunday, the physical thing that Jen-Hsun Huang held up on the CES stage was still a logic board with chips on it. Nvidia isn’t straying away from its area of expertise. It makes awesomely fast processors that excel at handling multiple tasks in parallel: exactly the sort of brute power required to make a dozen HD video feeds from cameras outside the car and as many high-res displays inside the car work. Nvidia’s pitch technique can certainly be improved, but the company is pursuing the right strategy by essentially manufacturing workloads for the stuff it already makes.

nvidia

Like any other publicly traded company, the thing Nvidia needs most is growth, which it isn’t going to get from its traditional business of making graphics cards for the PC. There will always be a subset of young people willing to spend outrageous amounts to get the best gaming performance, but their number is unlikely to grow — certainly not at a rate acceptable to investors looking for quick returns. Nvidia lost out to AMD in the battle to provide the graphics chips in the new generation of consoles, it miscalculated by investing heavily in Windows RT, and it’s been muscled out of the smartphone market by Qualcomm’s more integrated and efficient Snapdragon chips. Every effort that Nvidia has made to diversify away from graphics cards and PC gaming has been thwarted by one antagonist or another.

There's only one hype train left to ride: cars

Without an ultra-efficient mobile processor to put into wearables or a wireless radio to connect up the Internet of Things, there’s only one other hype train for Nvidia to hitch a ride on: cars. Automakers and electronics companies are using the Consumer Electronics Show to announce new alliances, collaborations, and partnerships to help accelerate the pace at which technology moves into the car and smartens up the experience of driving (or, eventually, drives the car itself). Without exception, they are all convinced that cars are primed for a massive upgrade in their situational awareness, automation, and convenience. Nvidia can be the executor of this collective vision, the provider of the brute force required to move us into the future. The new Tegra X1 chip is certainly capable enough for the task.

The craziest thing here is Nvidia's ambition

Nvidia’s misstep is in trying to convince companies that it can provide the software as well as the hardware required for smartening up the car. The company is pushing for the use of deep neural networking — a sophisticated, but extremely computationally intensive form of machine learning — to compile a shared dataset among all cars with its Drive PX platform and thereby smarten them up through a sort of hive-mind operation facilitated by over-the-air software updates. If that sounds extravagantly ambitious, that’s because it is.

You can have faith in Nvidia delivering raw computational power, but what’s the basis for trusting this company to come up with the proper framework for harnessing that power and revolutionizing driving as we know it? And even if Nvidia could pull it off, it'd still clash with the reluctance of car companies that would prefer to be in charge of the software running on their cars.

Converting companies like Audi from fans into buyers will be the biggest challenge

Notably absent from Audi’s endorsement of Nvidia’s technology at CES was any announcement of Audi actually buying the products and services Nvidia’s advertising. It wouldn’t be a problem for Nvidia that its CES event has been panned as overly technical if it helps the company reach its intended audience of half a dozen executives. The chief technology officers of Toyota, Daimler, General Motors, and the Volkswagen and Fiat groups are the ones that Nvidia was truly speaking to on Sunday. It’s not hugely surprising that they haven’t responded by opening up their considerable wallets and embracing Nvidia’s vision: each of those conglomerates and each of their individual car brands already have smart car initiatives under way.

sony cars

There is definitely an Nvidia-shaped niche in the market for future car technology. Automakers at CES aren’t in universal agreement yet, but most are looking at cameras as the main way to add environmental awareness to a car (as opposed to LIDAR, for example), and Sony’s also pushing the same message in its effort to sell more imaging sensors. Someone needs to do all the heavy lifting of processing those pixels and Nvidia’s proven it has the muscle for the job. If Nvidia finds the humility to limit its role in the connected car to that of a powerful but dumb computation engine, it may well find the automotive success it's looking for.

See all the latest CES 2015 news here ›