EINDHOVEN, Netherlands -- NXP Semiconductors is bracing for a future of self-driving cars and increased competition, said Maurice Geraets, vice president of innovation.
Chips capable of powerful computing will be required for higher levels of autonomous functions, vehicle networking and cybersecurity. This week, NXP introduced a multipurpose processing chip that can be used to control automotive functions from infotainment to powertrain to sensor fusion.
Competitors such as Nvidia Corp. and Intel are developing high-performance processors for specific functions, such as Level 4 and Level 5 self-driving, and reports indicate that Tesla Inc. is developing its own custom artificial intelligence chip.
Meanwhile, NXP is waiting on the European Commission to approve a $47 billion offer by Qualcomm, the U.S. mobile chip and telecommunications company. The combined operation would offer products for a broad array of uses related to self-driving cars and smart cities.
"When we talk to our automotive customers, there's still a lot of car-centric thinking," Geraets said. "The new world will move to mobility, to completely new systems, to completely new interactions for the car."
Geraets, 49, spoke with Shiraz Ahmed, assistant editor, mobility, last month at NXP's headquarters here.
Q: Do you think consolidation is necessary to achieve the economies of scale that will drive down the cost of some of this technology?
A: I don't necessarily see one company that needs to address everything. On the other hand, if you look to our customers, they do value that we are the No. 1 auto chipset supplier because we can have a complete portfolio. It does help, but it's not necessarily required to achieve the economies of scale in a specific technology. You can be No. 1 in vehicle networking, and another company can be No. 1 at radar chipsets. However, for the customer approach, for the market approach, it helps if your company is a one-stop shop.
Do you think chips will get to a point where NXP's chipsets aren't markedly different from the No. 2 in the market? Like how you can get any smartphone now, and they're all good, but you're buying it for the operating system.
I don't think it will be standardized that quickly. There's a lot of development still required to optimize the hardware and processing part of the cameras. In sensors, there is still some work to do.
Secondly, software requires dedicated accelerators and specific processing which is optimized for the purpose it's built for. You will see, for many, many years, dedicated processors being developed to optimize the sensing of the car, both in the sensors itself but also combining cameras in one sensor fusion box.
So better software requires better hardware -- they drive each other.
Yes, absolutely. It's not just standard hardware with new software, it's a combination of both.
When will the ordinary car have vehicle-to-vehicle communications?
I've seen predictions that in 10 years, in 2027, 50 percent of cars will have the technology. Brands will probably be shipping the technology to customers in a few years, across the range.
Many people have the idea that dedicated short-range communications would only have advantages if you have a higher penetration rate. But if every ambulance and every fire truck has DSRC on board, they could influence traffic lights. You have huge benefit on traffic safety and on the time that it takes for an ambulance to reach an accident.
In the Netherlands, we live in quite a populated, dense country where more than 30 percent of our traffic jams are phantom traffic jams -- those which don't originate from an accident or real structural bottleneck. Those traffic jams would be smoothed out if you would have car-to-car communication. It's not just with the car in front of you, but with the cars six, seven or 50 cars ahead.
If 10 percent of cars would have the V2X technology to smoothen out their speed with each other, then the [other] 90 percent of cars would follow the pattern. So with lower penetration, you would already have quite a benefit.
Do you see NXP expanding into new areas as computing becomes more important in all areas of life?
We will expand to new markets, but we will continue to be a chipset company.
We need to interact with many new players in the value network around us. In automotive 10 years ago, we had less contact with governments. Now that's relevant. They are road operators; they are regulators; they do trial approval, and it's all relevant for our chipsets.
So we'll have more interaction than ever with strange companies compared to the past. But, in the end, our core business will be the chips that we provide. So it will not be the case that in 10 years, we'll have more software engineers than chip engineers. We're in a fast-moving industry, but we should stay to our core. That's what we're good at.