Supercharging the Next Era of Vehicle Intelligence
Magna is doubling down on next-generation automated driving with a new partnership built around NVIDIA’s powerful Blackwell GPUs—opening the door to richer sensor fusion, higher-fidelity perception, and far fewer compromises in real-time processing. With ever-growing streams of camera, radar, lidar, and thermal data, the company sees these high-performance SoCs as the key to interpreting complex environments without cutting corners. Their approach to software is intentionally flexible: blending in-house modules, customer-developed stacks, and NVIDIA’s ecosystem to meet diverse ADAS and AV needs. The message is clear—Magna’s innovation engine isn’t slowing down, and this collaboration aims to push the boundaries of what next-gen automated systems can see, process, and intelligently decide.
Transcript
00:00:00 There's one thing that Magnet does extremely well. I know you do a lot of things well, but there's one thing to me that you do extremely well. You never stop innovating. You never stop building for the future. You're building all these great components for some of the world's leading manufacturers. You have a great contract manufacturing business.
00:00:16 You're always innovating and always looking to the future. And recently, the company just announced in a partnership with Nvidia. How did that partnership come together? And what is the goal of that partnership? we have a very good alignment on our needs and our requirements and our desire to to change things uh going forward. I mean it's not a uh something that we take lightly to
00:00:37 to work with a a partner or to work with a company that that that helps to gain advantage in the future. Um but we see for example that the ecosystem, the partner network, the tooling that I mean software tools, hardware tools, everything is very well in line with what we're trying to do. So it made very good sense for us. As part of that partnership, you're going to be able to
00:00:56 leverage the Blackwell GPU. What is that going to allow your engineers to build? Are they going to build something really cool, really new? What's that going to allow them to build? I mean, it's a platform. It's it's one of those ones where there's an answer I can give you today of things we know, and then there's an answer in two or three years time, which are things we don't know.
00:01:12 And of course, those generations will keep on coming out of the of the uh tooling plant as well. You we know there's going to be a next generation, a next generation as well. I would say I'd answer the question more generic and say all of these new generations of SOC's that they get more and more powerful just open up a completely new world for us. It means where we had to compromise
00:01:33 before on processing the data in the world cuz the world's a complicated place. There's a lot happening. You you see a lot and you add more cameras, more radars, more LAR, more sensors to the to the mix, you know, thermal and everything. You see more and more and more and you don't want to compromise and have to cut off the data. And so this is what those architectures and
00:01:52 these very high performance SOC's and ECUs help us to do is to be able to interpret the world around us in a much better way.
>> So the thermal is interesting. So you have you have the thermal, you have you have the radar, you have the the LAR, you have the camera, and now you have the Nvidia chipset. Now then the piece next piece of the pie is the software.
00:02:11 Is that are you developing that in-house Magnus solely? Are you working with partners? Are you perhaps running on the Nvidia CUDA? How how is that working?
>> We do the same. We do we do a mix of everything and it's very dependent on customer uh and also what we are having in-house experience and we're not having in-house experience. I mean it's important to remember that you should
00:02:32 focus on your strengths and and you should leverage other strengths and you should let you should help and cover others weaknesses right so in some cases customers they have their own software in house they want to use that that makes a lot of sense they've invested in it we help them to put that in a platform where where we are part of the system right so maybe they have a
00:02:51 perception stack or drive policy stack that that navigates and we'll help them with a middleware and making sure the system is up and running and it's functional and it's producable and it's working. All the all the like the things the housekeeping like making sure that it's uh it's still alive and there's nothing that's crashed and all those kind of things or maybe the over there
00:03:10 updates or maybe the security module to communicate with other units and so on so forth. So that's one instance could happen if a if a customer has their own stack they're preferring. They could also say they want a third party that they've chosen or it could be that they want us to do it and and we have our own in-house software as well for drive policy, parking and so on and so forth.
00:03:29 Um or we could leverage the NVIDIA network if there's some hybrid version. But software is such a broad thing. It's it's especially in ADAS today long time ago it was very easy very well contained. You had your radar, it had its software, you were done. You had your camera, it had its software, you were done. Now they're all intertwined. you have these big machine learning
00:03:49 models, you have you know extreme amounts of data and processing and understanding and real training and so on. So I would say we we we do a mix of everything depending on the customer, depending on the scenario.