Elon Musk's latest interview: Next year, Optimus robot production may reach a maximum of 100,000, and the target of cutting the budget by 2 trillion may only be half completed

Wallstreetcn
2025.01.09 10:09
portai
I'm PortAI, I can summarize articles.

Elon Musk also stated that by the second quarter of 2025, autonomous driving will be safer than human drivers; humanoid robots will be the largest product in history to date, with a ratio of at least 3 to 1 compared to humans; currently, three patients have implanted the brain-machine interface chip Neuralink…

On January 8th local time, Elon Musk and Stagwell CEO discussed the new developments in AI and autonomous driving technology, humanoid robots, the feasibility of the $2 trillion reduction plan, and the future positioning of X in a video conversation.

Here are the highlights of the interview:

  1. Essentially, some accumulated human knowledge has been exhausted in the AI training that occurred last year. The only way to supplement this is by using synthetic data, where AI will use synthetic data to write a paper, or it will come up with an argument and then try to grade itself, learning through synthetic data.

  2. If everything goes smoothly, our humanoid robot production will increase tenfold by 2026, so our goal is to produce 50,000 to 100,000 humanoid robots next year, and then increase that tenfold the following year.

  3. The pace of improvement in autonomous driving is exponential, and by the second quarter of 2025, autonomous driving will be safer than human drivers. Ultimately, it could be 100 times safer than human driving, with no collisions occurring.

  4. Humanoid robots will be the largest product in history so far... What is the ratio of humanoid robots to humans? I guess at least 3 to 1, 4 to 1, maybe 5 to 1. So we are talking about 2 to 3 billion humanoid robots, and I am not even clear what money will mean at that time, or if the economy will have any meaningful limits.

  5. Currently, three patients have implanted the brain-machine interface chip Neuralink, and we have upgraded the devices, increasing the number of electrodes, improving bandwidth and battery life. We plan to implant upgraded Neuralink devices for 20 to 30 patients within this year and next.

  6. We will strive to achieve the $2 trillion reduction target, which includes some buffers; if we aim for $2 trillion, we have a great chance of reaching $1 trillion. If we can reduce the budget deficit from $2 trillion to $1 trillion, allowing the economy to grow additionally, and keeping the output of goods and services in line with the growth of the money supply, there will be no inflation. I think that would be an epic achievement.

  7. If AI and robotics technology improve, ultimately AI will be able to do everything that humans can do. If computers and robots can do things better than we can, will our lives have meaning? That is a real question. Perhaps that is why we need Neuralink, so we can enhance human capabilities to keep pace with machines.

  8. I hope X can become the best source of truth. If you are trying to understand what is happening in the world, it can provide the most accurate and up-to-date information on anything big or small. It allows you to best understand what is happening in the world anytime, anywhere.

Here is the full text of the interview:

Mark Penn:

Chairman of Stagwell, Lindy Ocarino, and a group of about 25 chief marketing officers from large companies gathered at CES, where we truly saw the latest technology. We were fortunate to meet Elon Musk, who agreed to give us some time to answer some real hot questions about technology, how it will develop and change our lives, and how his interactions with the government may also change our lives So if I can start with a big overarching question, I would ask a few questions, and then we would get some questions from the group. I have to say, I have actually been a Tesla owner for many years.

Musk:

Thank you. Clearly, you have good taste.

Mark Penn:

I drove from Miami to Fort Lauderdale for dinner with my brother, about 40 miles. I only touched the steering wheel, but Tesla drove itself the whole way.

Musk:

It drove itself, you didn't need to intervene.

Mark Penn:

The entire trip required no input from me.

Musk:

Amazing. People might not believe it, but it really happened.

Mark Penn:

Yes, I can hardly believe it myself. I have always been skeptical about how far it could go, and your latest version of the software is truly incredible. What do you think will be the biggest technological advancement that will impact people's lives in the next decade? What should people expect to see from technology?

Musk:

AI will be the biggest advancement. Five years ago, or even fifteen years ago, if someone said there would be a deep superintelligence smarter than the smartest human, I would have thought it was absurd. Computers couldn't possibly be smarter than humans and capable of doing all these complex things.

But now, the latest AI can perform better than most people on complex tests, such as medical tests and radiology diagnostics; AI can diagnose radiology exams better than most people who have been doing it their whole lives. The advancements in AI are accelerating, Jensen Huang's presentation was fantastic, it really showcased how significant the entry of artificial intelligence is, with breakthroughs in hardware, software, and data, such as the application of synthetic data.

Because we have actually run out of all the books, we've exhausted the entire internet and all the books ever written and all the interesting videos. It's like you don't need a thousand identical cat videos, but rather all the interesting videos. You just slow it down, break it into some tags, essentially fragments of information. We have now exhausted all of that; basically, some accumulated human knowledge has been depleted in the AI training that happened last year.

So the only way to supplement this is to use synthetic data, AI will use synthetic data to write a paper, or it will come up with an argument and then try to grade itself, and self-learn through synthetic data, which is always challenging because it doesn't know whether the answers are generated by hallucinations or real answers. So finding the fundamental facts is so challenging, but it's clear that AI has exhausted all human knowledge at this point.

Mark Penn:

I know you are building the largest AI center on Earth.

Musk:

Yes, it's already up and running.

Mark Penn:

Microsoft plans to invest $80 billion in an initiative at a place where I used to work.

Musk:

By any standard, that's a huge amount of money.

Mark Penn:

We conducted a poll, and 13% of people said AI is changing their lives, while 87% expect AI to change their lives in five years Musk:

This is a huge difference.

Mark Penn:

Okay, what will it do for people? Is this something you would want?

Musk:

AI will do anything you want it to do, even suggest things you’ve never thought of. In the next few years, within 3 to 4 years at most, AI will complete almost all cognitive tasks that do not involve atoms.

Another element is robotics. AI cannot just think in data centers; it also needs robots to perform tasks, such as autonomous vehicles. The pace of improvement in autonomous driving is exponential, and we are confident that in about 3 months, it will be better than human driving. We are confident that we will achieve a level better than human driving in the second quarter of this year. Ultimately, it could be 100 times safer than human driving and may not even have collisions, as Tesla continuously improves the smart driving capabilities of its cars through software updates.

Mark Penn:

Let me try a few timelines. I want technical predictions because I’m not the youngest person anymore.

Musk:

Yes, my standard for youth is getting older every year.

Mark Penn:

I used to make computers when they were not available for purchase. So what is the timeline for autonomous vehicles to get government certification? Do you think it will happen within a year?

Musk:

There are already small autonomous vehicles operating in some areas, like Waymo, but they are limited to a few cities in the U.S. Tesla's solution is a harder path, but ultimately more powerful, and is a universal solution for autonomous driving. Tesla's software relies purely on AI vision, without any expensive sensors, no LiDAR, no radar. It doesn’t even need to know the area in advance. You can drive to places you’ve never been before, and no one tells us we’ve been there. As long as it’s not an alien planet, this method still works.

Mark Penn:

When can I buy a home robot?

Musk:

I think most people, if not everyone, would want their own personal R2-D2. I actually believe humanoid robots will be the biggest product in history so far. Everyone is likely to want one, and some want them to be in industries that manufacture and provide products and services. For example, what is the ratio of humanoid robots to humans? I guess at least 3 to 1, 4 to 1, maybe 5 to 1. So we are talking about 2 to 3 billion humanoid robots. I’m not even clear what money will mean at that time, or if the economy will have any meaningful limits. Assuming things go well, in a good AI scenario, I think we will not have universal basic income; we will have universal high income.

Mark Penn:

So do you think I have to wait five more years for my first robot?

Musk:

Our Optimus robot is the most complex humanoid robot in the world. Its hands have 22 degrees of freedom and look and feel like human hands. Our goal is to build thousands of such robots this year. Initially, we will test them at Tesla factories, but assuming everything goes well, next year we will achieve a tenfold increase in production. So our goal is to produce 50,000 to 100,000 humanoid robots next year Then in the following year, achieve tenfold growth. It's like having 500,000 robots in three years. I think we should view them as Roman legions. How many robot legions will we have? Like a Roman legion, that's 5,000.

Mark Penn:

So when will we establish a colony on Mars?

Elon Musk:

I think we will be able to send the first rudimentary spacecraft to Mars within two years. Earth and Mars synchronize every two years. So we are currently at a synchronization point. The next one is about two years from now. Then there will be another two years later.

For the first trip, we need to ensure that we can land the Starship without crashing. We need to prove that we can land the spacecraft on Mars without increasing the number of craters. If those spacecraft land safely, then perhaps on the next trip we will send people up, and hopefully this will grow exponentially. So eventually there will be thousands of Starships going to Mars. There could be a very cool visual effect, like a battlefield. So the colony ships completely leave, like bright spots in space. I think it looks cool.

But beyond that, I think the goal must be to reach a self-sustaining level on Mars. The point of Martian self-sufficiency, truly defined as if the supply from Earth is interrupted for any reason, Mars will not perish, and Mars can continue to grow. If something happens on Earth, it could be a third world war or a natural disaster, or other reasons that cause supply interruptions. If Mars can still continue to survive, then the potential lifespan of civilization will be greatly extended. If you take a step back and ask how you would evaluate any civilization, you would ask, is that civilization still stuck on its own planet? Or are they a multi-planet civilization? We do not want to be one of those boring single-planet civilizations. There is a decent outcome here, even if you can't go beyond our solar system, at least reach another planet.

Mark Penn:

Okay, finally I want to talk about the communication between the brain and technology. This seems quite promising to me, can I learn more?

Elon Musk:

Yes, we now have Neuralink technology, which has been implanted in three patients and is functioning well. We have upgraded the devices, increased the number of electrodes, and improved bandwidth and battery life. We plan to implant upgraded Neuralink devices in 20-30 patients within this year and next.

Our first product is designed to help those who desire a brain-body connection, such as patients with limb paralysis or aphasia. If we can enable them to communicate as quickly as normal people, or even faster, it would be transformative. This product can read signals from the brain's motor cortex, allowing them to control the cursor on a computer or phone through thought, enabling people to control their computers or phones with their minds.

Next, we will develop products to help the blind. Even if a person has lost their eyes or optic nerves, or has never seen anything, we can connect directly to the brain's visual cortex, allowing them to see. We have been experimenting on monkeys for two years, essentially enabling people to control devices Ultimately, we believe that if a second Neuralink device is implanted below the site of spinal cord injury, which exceeds the point of injury, we can transmit brain signals to allow paralyzed patients to walk again. This would clearly be a significant breakthrough.

But I am confident that we can achieve this as much as possible; Neuralink's long-term goal is to increase bandwidth. Currently, when we talk about bandwidth, the sustained bandwidth of humans is very low, averaging less than 1 bit per second. So in a day of 86,400 seconds, the average human input and output is far less than 86,400 bits. If someone is a writer, they might exceed this number, but most people do not output more than the number of seconds in a day. With Neuralink, you could increase output capacity by 1,000 times or even 1 million times. This would be a completely different experience, essentially superhuman.

Mark Penn:

I want to be an early adopter; please put me on the list.

Musk:

I believe you will really like this chip. I can guarantee that.

Mark Penn:

Let's get back to reality; I want to ask a question about spending cuts. In fact, I worked closely with President Clinton in the 90s when we did reshape the government. We balanced the budget in two years. But it didn't last long because it collapsed quickly. Have you identified some cuts that you think will be successful? Do you think $2 trillion is a realistic number?

Although it didn't last long because it collapsed quickly. Have you identified some cuts that you think will be successful? Do you think $2 trillion is a realistic target?

Musk:

Yes, we will strive to achieve the $2 trillion target. I think that is the best outcome. But we need some buffer. If we aim for $2 trillion, we have a good chance of reaching $1 trillion. If we can reduce the budget deficit from $2 trillion to $1 trillion, allowing the economy to grow additionally and the output of goods and services to keep pace with the growth of the money supply, then there won't be inflation. I think that would be an epic achievement.

In terms of government spending cuts, this is an environment with many opportunities. I feel like I'm in a room full of targets, like you can't miss them when you close your eyes. There is a lot of waste in government, especially in the federal government, because they have unlimited sources of funding, and the people spending the money are not spending their own. Even if some government officials want to use funds effectively in principle, the system prevents them from doing so. For example, at the end of the budget cycle, they are required to spend all their budget, or else the budget for the next year will be reduced. This is actually a reverse incentive for wasting funds, and then they are punished for not wasting money.

Mark Penn:

I agree. I did an analysis comparing how the government used to do things. Considering inflation, the infrastructure bill should actually allocate $4,000 for the Brooklyn Bridge or the Lincoln Tunnel. But the government did not execute it in the past way.

Musk: That's right, we have accumulated a large number of laws and regulations, making any large project basically illegal. Even if you try to do it, you have to spend more money on paperwork than on the project itself, and then the project will be delayed. So the key is to focus on regulations and abolish those that do more harm than good. Every regulation has its pros and cons, and we need to assess the balance and get rid of those that are completely meaningless and do not serve the public interest.

Mark Penn:

Linda O'Kallina received a huge round of applause when she mentioned the Dodge issue in her keynote speech. I think this country is really waiting to see such efforts. They are supportive and optimistic. Before I raise other questions, let me talk about a topic that Mark Zuckerberg has recently sparked a lot of discussion about. What do you think about his acknowledgment that the government or his company, or both, are censoring certain content?

Musk:

There is no doubt that the government is indeed censoring content. We learned this from the Twitter files. Some practices are quite illegal, such as the FBI having a portal into Twitter that can monitor and delete any content, with a two-week automatic deletion period. We don't even know exactly what they did, but they clearly have enormous power, which sounds very illegal. At the same time, there is a lot of self-censorship. I strongly believe that for a democracy to function normally, you must have freedom of speech. Without freedom of speech, how do you know what is really happening? If you can't make informed votes, then there is no real democracy.

I think we should listen to the wisdom of the founders of this country and think about why they made freedom of speech the First Amendment. They did this for a reason, because they came from a place where speech was strictly censored, and expressing one's thoughts could lead to imprisonment or execution. They wanted America not to repeat that mistake.

Mark Penn:

You're right, citizen journalism is becoming increasingly important today. We do need citizen reporting to tell people the truth. Thank you for your insights.

Musk:

Yes, before the internet, we had to rely on traditional media. Journalists would go out to interview, then return to the office to write articles, which would then be professionally printed into newspapers and distributed. That was the only way to know what was happening, but it was very slow. For example, the news of Lincoln's assassination took three weeks to reach Asia. In the past, you might not even know that your country had gone to war for a month because it took a long time for the news to reach your village.

Mark Penn:

Yes, we actually learned about Pearl Harbor because we decoded it. We actually learned about Pearl Harbor because we decoded the Japanese network system. That's how we really understood it. I only have three minutes.

Musk:

If you have more time, if you want. Okay.

Mark Penn:

Thank you. I have three questions from the audience. Dan Gardner, Toby Daniels, and Ivan want to know, do you think the internet is bad? What do you think we need to fix it? Elon Musk:

Are you referring to content on the internet or internet connectivity? Starlink can help people get online, especially in poorly connected areas. In fact, I believe Starlink has had a significant impact on helping people escape poverty. In many parts of the world, people have products they want to sell, but they can't do so without internet connectivity. Or if they want to learn something, basically you can learn anything for free on the internet. MIT has many lectures. There are free resources online. So once you have internet connectivity, you can get an education, and you can access the global market. So I think this is very important for how connectivity changes people's lives.

We know, perhaps the question is whether there is too much negativity on the internet? I think sometimes there is. On the X platform, I suggest adjusting the algorithm to make it a bit more positive, but people are uneasy about that.

Mark Penn:

The second question comes from Target's CEO, Zach Moffat, who asks how you make pessimism uncool again? This country used to be very positive, and now when I ask if we are on the right track, they no longer say we are on the right track.

Elon Musk:

You know, I am very optimistic about the future. I think we have the potential to enter a golden age. So it is very important for us to get rid of a lot of regulations that hinder progress. I'm not saying there aren't some good regulations, but there are too many things we can't do. Take California's wildfires as an example. Yes, we really need firebreaks, we need to clear brush around homes, and ensure reservoirs are full. These are obvious things, but due to many environmental regulations, you actually can't do that in California. So I would allow for five breaks, and then I would allow for the brush to be pushed away from the house.

They might hear something like the red-legged frog. It's like some kind of creature, usually one you've never heard of, that prevents this from happening. We have more freshwater flowing into the ocean. We really should do this because, theoretically, it helps these little fish that like a brackish freshwater mixture. If we retain more freshwater, then this fish that smells bad won't be happy. But in fact, if we drink a bit more freshwater, there is no actual evidence that the smell of the fish will make us unhappy. So we should retain more freshwater, keep the reservoirs full, and set up some reasonable firebreaks to keep the brush away from homes. This is just one example. If we do this, we save a lot of trouble and tragedy in LA. So, you know, yes, I think AI and robotics will bring people a higher standard of living beyond imagination, and AI will make amazing progress in healthcare and pharmaceuticals.

Mark Penn:

The last question comes from the Heritage Foundation: if all robots and everything froze human time, how would they utilize that time? Or what could we ask them to do? I think that definitely perfects the cycle of technology Elon Musk:

It's a bit like retirement; if AI and robotics get better, eventually AI will be able to do everything humans can do. Any task you do is optional, like a hobby. If computers and robots can do it better than us, will our lives have meaning? That's a real question. Maybe that's why we need Neuralink, so we can enhance human capabilities and keep up with machines.

Mark Penn:

Thank you for your optimistic view on technology and its direction. I feel reassured that there will be significant progress in building trust and credibility in the coming years. Thank you again for your important role in free speech.

I have been closely connected for many years, and I am very grateful for that. I think this is definitely a huge message as we are in Las Vegas, researching what technologies people can use. Really, thank you again for your tremendous role in open free speech, and thanks to X and Linda for their contributions in this regard. I know we are working hard to gain full recognition from everyone. This platform should exist because free speech is the right way.

Elon Musk:

Yes, I don't think imbalance is a good thing. I hope the X platform can become a global consciousness, like the collective conscience of humanity. You will get every aspect of humanity, the good and the bad. That's just how it happens naturally. I hope this is a good and productive thing. My wish is to maximize the time users spend without regrets.

Mark Penn:

Interestingly, when predicting future technologies, very few people foresaw the emergence of social media and its impact. One last question, where do you hope the X platform will go?

Elon Musk:

I hope X can become the best source of truth. If you are trying to understand what is happening in the world, it can provide the most accurate and up-to-date information about anything big or small. It allows you to best understand what is happening in the world at any time and place.

Mark Penn:

Thank you very much. Do you have anything else to say?

Elon Musk:

I want to encourage people to remain optimistic about the future; I believe the chances of good outcomes are much greater than those of bad outcomes. That's my prediction. Thank you