Skip to main content
Peter Ludwig

Peter Ludwig

German art collector and entrepreneur

Nov 13, 2026

44:01
Mm-hmm
44:02
...
44:02
so there's no perfect answer to, to those questions.
44:05
Um, undoubtedly, there's still, and there will forever be, I think, tremendous value in people who are experts in their fields.
44:12
I think, um, there is, uh, uh...
44:17
The, the, the biggest challenge, though, is changing the way of working.
44:22
Um, and oftentimes what we see is, in organizations, that, um, the, at the leadership level, there's very strong support for accelerating, uh, progress and development using AI, and oftentimes for, let's say, the, um, the, the, the new graduates, the, the younger, uh, the younger people within an organization, there's also a lot of enthusiasm.
47:02
What do you expect a leading car maker to look like in 2035?
7:50
Mm-hmm.
7:50
So digital AI is what we think about when we are using a smartphone app like ChatGPT or Claude, and, uh, and we're interacting with a phone or a computer and, and we're getting responses that live on the screen.
8:03
Um, physical AI is when, uh, we're interacting with intelligent systems where they're actually controlling something in the real world, right? This could be, this could be a car, this c- a self-driving car.
8:15
This could be a, a humanoid robot or an intelligent, uh, factory machine.
8:20
I think anything that has to do with, with real, real moving bits-

26 MINS LATER

34:46
Yeah.
34:46
Um, the, the next era of, uh, of autonomy is really what we view as, uh, a closed-loop, uh, simulation with reinforcement learning.
34:56
This is a, a key technical unlock, uh, which I think is frankly going to solve, uh, many of the unsolved, uh, aspects of physical AI, and this is all going to play out, uh, over the next few years.
15:45
Peter, maybe it's helpful for you to walk us through the history of that evolution and how it all plays together.
15:50
Firstly, I would say something that we, we knew when we started, uh, almost 10 years ago was just that the technology is still going to change a lot.
15:58
There's a lot of advanced engineering and, and research that works in this entire field.
16:02
A way that you can be part of that but not be, let's say, overly exposed to any specific implementation is to think more horizontally.
16:11
And so for us, that meant initially really focusing on tools, and then building tools in such a way that we could continue adding onto the platform and just recognizing that the technology itself, when we talk about physical AI and advanced autonomous systems, almost every two years, there's some sort of breakthrough that changes how you have to think about these things.
16:30
If you're not dynamic enough in, in understanding how to adapt that, that latest technique, that latest breakthrough, then you can become almost obsolete in that sense.
16:40
And that, that's been really baked into Applied's DNA.

15 MINS LATER

31:57
Anything that's around the actual technology that you're building that worries you?
12:44
Um, and I think that's, that's, that's, uh, that's, that's, that's something that's understated in this, in this, uh, conversation.
12:50
Yeah.
12:50
A few other things on, on digital versus physical AI.
12:53
So in, in digital AI, the state-of-the-art is you, you can train models effectively on the entirety of the internet, and then maybe augment that with additional data that's been collected and refined with some hired experts, right? This is sort of a hot field right now.
13:06
Uh, but generally, you're talking about a foundation model that's built on internet data.
13:10
In physical AI, the internet data is, is useful too.
13:14
However, to actually build a foundation model in physical AI, there's also a lot of private data collection.

49 MINS LATER

62:19
You know, now you've generated, uh, technol-- the same thing is happening in our universe as well.
speaker_5UNKNOWN
30:05
Mm
30:05
... that the vehicle has to pass.
30:07
And that was how the industry worked, let's say, uh, un- until 10-ish years ago.
30:11
But what's changed now is with these models everything is stati- statistics, right? Like you, you no longer have a black-and-white answer, but it's like, well, how many orders of magnitude or how many nines of reliability can, can I get in the system, and how can I, how can I prove that to be true? Um, and, uh, and, and the big unlock honestly for physical AI as, as an industry is that these models are just becoming much more reliable.Right? Things like things actually work a lot better.
30:39
It's like the number of nines you can get out of these systems are, are now good enough that it actually becomes cost effective to, to really deploy these things.
30:46
And so the, the big shift in, in, so verification and validation has been from a little bit more of a...
30:52
Again, in the past it was strictly requirements, and are you meeting or not? And, and now it's more of a statistical, uh, verification and validation case where it's all about how many nines of reliability and, and meantime between failures, that sort of thing.

33 MINS LATER

speaker_6UNKNOWN
64:24
Uh, and then for Peter, I was just kinda curious if there's any particular tech or research problem that you would call out as very meaningful for you guys if it was solved, and unsolved, and if anyone is working on it, they should get in touch with you.
5:46
And
5:46
maybe you get a little icon on the dashboard that you don't really know what it means.
5:51
And maybe you're looking at the manual and like, OK, what's the problem here? and on a modern car you can have a much more sophisticated diagnostic system a much more sophisticated way of servicing the information to the user and then also with ai we can also then surface that to the manufacturer and they can understand across their entire fleet what are the sort of the aggregation of the problems that they're having and root cause those things and i think we won't necessarily go into the deep details but these are some of our you know R&D and showcase vehicles.
6:20
And so the middle vehicle here is a showcase vehicle for what's called our vehicle operating system.
6:24
And then this is a self-driving vehicle.
6:26
So this can drive on urban environments and highways.
6:30
It can park itself.

11 MINS LATER

17:01
Big.
8:17
us resources.
8:18
Yeah.
8:19
We just learned and learned and learned.
8:21
We really expanded out and became very horizontal.
8:24
We started working in a bunch of different industries.
8:28
We were very much on top of all of the latest AI breakthroughs at every step.
8:34
And so we could sort of see when something interesting was happening, should that actually change our strategy? And so there were a few interesting things that happened which evolved the strategy.

11 MINS LATER

20:03
Oh, Oh, you're
14:37
How do you develop the software for these different environments and different vehicle sizes?
14:41
Yeah, that's a great question.
14:42
So there's so many different parts of the overall software package, right? So there's the entire embedded software stack, which actually runs on the vehicle.
14:50
There's the infrastructure piece, which is all of the data management and model training infrastructure that's really off board.
14:57
And then there's also just the science and the true engineering of the stack itself, which is algorithms and machine learning and model training and model tuning, all those things.
15:08
And in each of these components, there's actually a lot of similarities.
15:12
And I'll remind the audience as well that the world's best vehicle dynamics package is also out of applied intuition.
15:34
Are there any sensor or hardware limitations as you build and scale this?
12:20
Could we see eventually your L4 stack go into off-roads? Is that something Department of Defense has been very public about for years, almost decades now, that they have ambitions in that?
12:30
Yes.
12:30
So just some terminology that I first want to clarify.
12:33
So level four typically refers to when you have a vehicle that's operating in a geo-fenced area.
12:40
And so, for example, the Waymo Robotaxi in San Francisco, currently it's geofenced to roughly the city boundaries of San Francisco.
12:48
And that's level four technology.
12:50
Level five is when a vehicle can effectively go anywhere that human can go.
14:57
Do you have to redesign the sensors or build special housing for the type of environment that you're operating in?
7:02
What is the big lessons from Android? Especially on hardware diversity, right? That's kind of, I think, the big innovation of their industry.
7:10
Android, it's an incredible story from where it started to where it is today.
7:15
I think it's true that Android is the number one OS in the world.
7:21
It runs on more devices than anything else by a healthy margin.
7:24
The big thing that Android figured out was just how to run applications uniformly on a huge variety of hardware.
7:32
And how to do that in a way that the user experience is actually consistent.
7:35
And there are a lot of lessons, both at the technical and non-technical level, on how to actually enforce that.

15 MINS LATER

22:53
Is there anything you can share in terms of some of the approaches you've taken there?

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.