00:00It's been so interesting to read your research in the month of September.
00:03I think that the question that I kept coming back to is how you see this as being so economically
00:08consequential for the United States, right, this move to physical AI.
00:13Yeah, thanks, Ed.
00:14We think historians of the future are going to write about this time now.
00:18Where were you when the robots came?
00:21Where were you when we really started to industrialize and make use of infrastructure in space?
00:27So the robotics, when you combine AI and robotics, we think that that's economically multiplicative, like an 8x or 10x
00:37global GDP before you and I retire.
00:41It's militarily deterministic and potentially socially destabilizing.
00:47So we have to get this right.
00:48In the process, we think the story of physical AI and robotics is the story of on-shoring and rebuilding
00:56American manufacturing resiliency for the first time in many, many decades.
01:00And it's going to bring a lot, I think, millions of human jobs along with it.
01:05So to make this artificial life, Ed, we need a lot of vocational and even mid and high-skilled labor
01:12in the U.S. and with our economic partners.
01:15We're talking about this from a thematic research point of view.
01:18And when we have your industry colleagues on, they say two things.
01:22China is ahead in deployment, real-world production, be that humanoid robots or autonomous vehicles.
01:28But the supply chain element is a big point of focus.
01:32Where is your research netted out on that part of China's story?
01:36It's a fact.
01:38Credit to China for having a leadership in the manufacturing ecosystem, the supply chain, all the way up to critical
01:49minerals and materials, rare earths.
01:51I know your program has talked about those things.
01:53You can't have a robot.
01:55You can't have a data center or any sophisticated electronic product without China super, super involved in the ecosystem.
02:06And that's not going to change overnight, right?
02:08So, again, part of the paradox is in order to make robot life, we need human jobs.
02:14And in order to diversify away from China to create a little more balance, we're going to need to continue
02:21to work with China.
02:23In our simulations and scenario analysis, there's no real credible way you could build a wall and just wall off
02:32China from the U.S. ecosystem.
02:33I mean, if you think inflation is rough now, just imagine what it would be if we attempted to do
02:39that.
02:40So, I actually think the story of physical AI and on-shoring in a way can lead to a, yes,
02:46re-architected and still sensitive,
02:49but we think a very prosperous entanglement between U.S. and China relations.
02:57Okay, so what role does regulation play in this United States physical AI effort?
03:03Obviously, this week you have the U.N. General Assembly.
03:06Since the Pacing the Frontier essay from Anthropik, we've obviously looked at regulation of AI from the model side and
03:14the software side.
03:15Have you modeled for, pardon the expression, but modeled for a regulation outcome for physical AI from the United States?
03:24It's deterministic.
03:26In our base case, there's no, again, plausible scenario where we don't have physical AI very regulated.
03:34Okay, it's not binary.
03:36We can debate whether the regulation, how the regulation phases in and is sequenced in,
03:43whether it's informed by industry experts, elected officials, probably a combination of both,
03:49probably in response to things that are progressing in unpredictable ways as we're seeing over the last couple of weeks.
03:58Those same concerns do pass on into physical AI because of the, any robot with an AI brain attached to
04:06it as an independent agent,
04:08it does have a dual purpose.
04:10Like any piece of super advanced technology, there is a potential military element.
04:19You're seeing it in the campaigns in Ukraine and Russia.
04:22You're seeing it in the Straits of Hormuz as well.
04:26And the topic, again, addressed by your show of how our defense ecosystem is also in need and currently in
04:34process of being re-architected in a physical AI world.
04:37That kind of stalking horse of China on the competitive side and potential deficiencies on the military side is creating
04:46a sense of urgency that, yes, we will have regulation.
04:49But in our opinion, the regulation will change the form and the cadence, but not the destination.
04:56We are going to have tens of billions of robots in our daily lives within the next one to two
05:03decades.
05:05The company case studies of embodied AI for you, Adam, and Morgan Stanley are SpaceX and, to a degree, Tesla.
05:11And you wrote on Sunday that Tesla and SpaceX have separate but synergistically linked physical AI capabilities.
05:18And you go on to write the companies are working to gradually close the gap between them.
05:23And inevitably, the question that our audience puts to you is your scenarios and your base cases for the merger
05:31of SpaceX and Tesla under that assumption.
05:35Sure. It's a fair question.
05:37Elon gets that question a lot as well.
05:41Well, I won't be able to comment on a specific probabilities of a transaction between two publicly traded companies, but
05:48the relationship seems deterministic at this point, okay?
05:53Tesla, as a manufacturing and data collection layer, they make the robots and can scale that intelligence into the physical
06:01world.
06:02SpaceX is the connectivity in the AI layer, also having very impressive manufacturing chops themselves.
06:12And Elon and others, and some of our work as well, working with our colleagues across the auto team,
06:18Andrew Bercocco, who covers Tesla, and our internet team, led by Brian Nowak and the internet team,
06:24collectively working on what's that next iteration, regardless of strategic transaction.
06:32We think the two companies are investors should expect continued cooperation as the shared mission is essentially converting energy to
06:43intelligence at scale
06:44to achieve the most efficient intelligence per watt, per dollar, per second.
06:52That race of intelligence and density of energy, and then getting that time to power quickly,
06:58we think both companies bring something important to the table.
07:01I think a lot of people are asking themselves, what would the end result be, the net result, right?
07:07So if you take Optimus, a humanoid robot, I guess the question for you, Adam, is like, does Optimus work?
07:13Does it exist in the real world in industrial settings or in the at-home setting that they've also mentioned
07:21without Starlink connectivity, without inference at the edge powered by Orbital Data Center?
07:26Like, how much is one a precursor to the other?
07:30Well, we think any machine that can be automated will be.
07:33Any machine that can have an AI edge inference computer in it will.
07:38Any machine that does not may have a much more rapid obsolescence curve, if not really have any useful life
07:45at all.
07:47The humanoid form factor that you bring up is, it gets a lot of attention.
07:53We think it's more of a recruiting tool and a capital raising tool.
07:57In our modeling of our global robot model out to 2050, it is an important form factor.
08:02We do get into the billions of, in many of our scenarios, of humanoids.
08:06But we think it's just one of maybe thousands of different stratum of low-altitude robots, terrestrial robots,
08:12autonomous vehicles, mobile robots, industrial, and everything in between.
08:17And yes, the connectivity, connecting those robots together in a swarming intelligence
08:23that could do potentially the vast majority of the world's inference work.
08:28Elon refers to this as the distributed inference cloud connected to the distributed orbital cloud
08:34of the space-based AI infrastructure.
08:37We're still going to have data centers for the foreseeable future.
08:41But we like to say at Morgan Stanley, using a data center or a GPU for an inference task
08:46is like using a Ferrari to pick up the milk.
08:49Look, we need to start to activate more inference at the edge, kind of like neurons in your brain,
08:55so we can do the compute vastly more efficiently than me speaking into a phone
09:03and having a rather simple query generating tokens somewhere in Memphis.
09:09Your job is to be the global embodied AI and robotic strategist at Morgan Stanley.
09:14Your history and origins, like more focused on the automotive part of Tesla.
09:19I just would love to understand where Tesla, as an embodied AI company,
09:24separate from whatever happens with SpaceX, ranks in your world view of that industry
09:30against all of its peers in China, other players here in the United States.
09:34We think Tesla, not singularly, but Tesla and Tesla and SpaceX
09:40represent one of the best chances to help revive and give Western industry
09:47a chance to keep up, if not to surpass our geopolitical rivals over time
09:54in robotics and physical AI.
09:58Tesla specifically right now, I'll tell you today,
10:01I was driven from Westchester to Midtown Manhattan in my Tesla,
10:07and I did not have to touch the wheel.
10:09Interestingly, I was delivering a robot dog made by a Chinese company.
10:15I brought it back to my office to return it to the company we borrowed it from.
10:20We think autonomous cars are solved,
10:23and this was something that even Apple and a lot of the physical AI propagators have said
10:29was the world's hardest problem.
10:30In our opinion, when the historians of the future write about
10:34when did we solve autonomous cars,
10:35it was 2023 when Waymo pulled the driver from Phoenix.
10:41So is it to the level of safety?
10:44Do we have enough nines to make it as where we want it to be?
10:46No, that's going to continue.
10:47But we think that on that modality,
10:51if we solve that and we have insurance companies start to give you a discount
10:54for letting the car drive,
10:55that will open up the floodgates for other form factors to follow.
Comments