Written by: Techub News

Introduction

Recently, Elon Musk appeared on the well-known tech podcast All-In Podcast, engaging in an information-dense conversation lasting 44 minutes. This comes as several technologies from his companies reach critical milestones: Tesla's Optimus humanoid robot design is nearing completion, SpaceX's Starship is moving toward full reusability, xAI's Grok model is brewing the next generation breakthrough, and the Starlink project aims for direct mobile connections through new spectrums. As a leader deeply involved in multiple disruptive fields such as space exploration, artificial intelligence, electric vehicles, and robotics, Musk's focused articulation in this dialogue provides a rare firsthand perspective for the outside world to understand the future landscape of his tech empire.

Summary

• Optimus Robot: Musk describes it as “the greatest product in human history,” currently tackling three major challenges in mimicking human hands, AI cognition, and large-scale production supply chains, with expected production costs potentially dropping to around $20,000.
• AI Evolution Speed: Musk predicts that by next year (2025), AI may surpass the smartest humans in any single domain, and by 2030, its combined intelligence could exceed that of all humanity.
• Tesla AI Chip Leap: The AI5 chip currently in final design will have a 40-fold performance improvement over AI4 in certain key metrics, significantly enhancing the autonomous driving experience.
• Starlink Direct Mobile Connection: Through newly acquired spectrums, SpaceX aims to achieve high-bandwidth direct connections from satellites to mobile phones within about two years, potentially providing global communication services.
• Multi-Planet Species Timeline: Musk believes that if Starship's capacity can grow exponentially according to Mars launch windows (once every two years), humanity could establish a self-sustaining civilization on Mars in about 25 to 30 years.

Optimus: Overcoming the “Human Hand,” Aiming for a Million Units Annual Production

One of the core focuses of the conversation was Tesla's humanoid robot Optimus. Elon Musk did not hide his ambitions, stating, “I believe Optimus will become the greatest product ever created by humanity.” He revealed that the team is finalizing the design for the third version of Optimus, which will be “an exceptionally advanced robot.”

Musk pointed out that achieving a universal humanoid robot requires overcoming three major challenges: hands that rival human dexterity, AI cognition capable of navigating and understanding reality, and the ability to mass-produce at scale. Currently, other robot companies in the market lack these three elements.

Among them, the mechanical design of the robotic hand is viewed by Musk as the biggest engineering challenge. He praised the human hand as a refined instrument evolved over time, with about 27-28 degrees of freedom, capable of performing an extremely wide array of tasks from threading a needle to disassembling a car and playing the piano. To replicate this ability, Tesla must design every motor, gearbox, and drive electronics from scratch, because “you can't purchase ready-made actuators.”

“Currently, the effort I am putting into Optimus is likely more than any other single project,” Musk remarked. This involves solving real-world AI issues, all electromechanical problems, and building supply chains and production challenges from the ground up. He estimates that once production reaches one million units per year, the marginal production cost of the robots will drop to between $20,000 and $25,000, of which the AI chips may account for $5,000 to $6,000. The final price will depend on market demand.

Tesla AI Chip: AI5 Performance Leap of 40 Times, Empowering Autonomous Driving and Robotics

In addition to the robot hardware, its “brain” is equally vital. Musk shared the latest developments on Tesla's AI chips. The company currently has two chip lines: Dojo for AI training and AI chips for inference (currently shipping AI4).

He highlighted the soon-to-be-completed AI5 chip, describing it as “a huge leap.” In certain core metrics, AI5 will have up to a 40-fold performance improvement over AI4. This is due to the highly granular collaboration between the software and hardware teams, directly optimizing for the bottlenecks of AI4. Specifically, the raw computing power of AI5 improves around eight times, memory increases by about nine times, and memory bandwidth enhances roughly five times, combined with hardware-level optimizations for specific operations (such as softmax), achieving a 40-fold composite improvement.

Musk confidently stated that even based on the current AI4 chip vehicles, the software version 14 (the largest upgrade since version 12) set to be released in the coming months can achieve autonomous driving capabilities that are safer than humans by 2-3 times or even 10 times. The new software will significantly increase the number of parameters and use reinforcement learning to improve the information loss issue during the “reality compression” process. “By the end of this year, your car will feel like it has consciousness,” he added.

SpaceX Dual-Line Progress: Starlink Direct Mobile Connection and Fully Reusable Starship

The discussion turned to SpaceX, where Musk revealed two important developments. The first is the Starlink satellite direct mobile connection service. Thanks to newly acquired spectrums, SpaceX will be able to provide high-bandwidth connections from satellites to mobile phones. However, this requires mobile hardware to support the new frequency bands, so it will take about two years for phones equipped with the corresponding chips to hit the market. At the same time, SpaceX is also building satellites that can use these frequencies.

“The end result is that you should be able to watch videos on your phone anywhere,” Musk described. He envisions a future where users might subscribe to a globally applicable Starlink account to achieve comprehensive high-speed connections for homes and mobile devices, similar to choosing AT&T or Verizon. When asked if he would consider acquiring operators (like Verizon) to gain more spectrums, he replied, “It's not impossible; it could happen.”

Another major advancement is the fully reusable Starship. Musk stated that despite the recent launch resulting in a “big explosion,” the team's recovery speed is impressive. He credited the efficient work of the SpaceX team and their ability to respond to regulatory scrutiny. For the next step, he predicted: “We will achieve (booster and spacecraft) recovery next year.” At that point, Starship will be capable of fully reusing more than 100 tons of payload to be sent into orbit.

The current biggest technical challenge lies in the reusable orbital thermal protection shield. Musk noted that the Space Shuttle's thermal protection system required up to nine months of maintenance after each flight, while SpaceX's goal is to develop a thermal protection system that can withstand re-entry high heat without requiring piece-by-piece inspection and is lightweight yet strong, involving both material and engineering challenges.

xAI and Grok: Reconstructing “Truth Data” with Reasoning, AI Intelligence Explosion Imminent

When discussing his artificial intelligence company xAI and its model Grok, Musk revealed a key innovative direction: using massive reasoning computing power to purify and enhance training data. Grok is examining all human knowledge sources (such as Wikipedia, books, websites), assessing the truthfulness, completeness, and fallacies of each piece of information, and then rewriting content to correct errors, supplement missing contexts, and eliminate misinformation.

The host suggested publicly releasing this “revised” knowledge base (playfully dubbed “Grokpedia”) to counter biases and information warfare present in existing encyclopedias, to which Musk stated he would discuss with his team.

Regarding the evolution speed of AI intelligence, Musk made bold predictions. He believes the increase in intelligence is roughly logarithmic to the volume of computation; for example, tenfold computing power may double intelligence. Based on this, he estimated: “AI may be smarter than any individual human in any single task by next year (2025). And by 2030, the total intelligence of AI could surpass that of all humans combined.”

Musk views AI’s evolution as part of a cosmic scale expansion of intelligence. He compared that human intelligence has improved with population growth and increased information storage, but is now facing a plateau due to stagnation or even decline in population growth. AI, on the other hand, may continue along the curve of increasing computing power, pushing intelligence to new heights.

Concerns about “Western Suicide” and the Redemption of “Curiosity Philosophy”

In the latter part of the interview, the discussion shifted to more macro-level civilizational aspects. Musk expressed deep concern about the tendency of “Western suicide.” He pointed out several symptoms: fertility rates below replacement level, open border policies leading to societal fractures, and selective media neglect of crime issues. He believes that when people lose optimism and a sense of mission for the future, civilizations begin to decline.

“Having children is an act of optimism for the future,” Musk said. He proposed the need for a philosophy or belief that can inspire optimism and excitement to fill the void. For him personally, this is a form of “Curiosity Philosophy” – a desire to understand the essence of the universe, to expand consciousness and the limits of human existence.

It is this philosophy that drives all of his business ventures: exploring space, becoming a multi-planet species, and developing AI to expand cognitive boundaries. He cited a metaphor from The Hitchhiker's Guide to the Galaxy: the answer (the universe) is right there; the difficulty lies in asking the right questions. Humanity needs to continue growing and expanding to ask and comprehend those fundamental questions about the universe.

Timeline for Mars: Establishing a Self-Sustaining Civilization in 25-30 Years

The grandest project to put Curiosity Philosophy into practice is to make humanity a multi-planet species, especially colonizing Mars. Musk elaborated on its core logic and expected timeline.

He emphasized that the key is not in the earliest Mars missions, but in whether enough supplies and capabilities can be transported to Mars to ensure its continued prosperity even after Earth-based supplies are interrupted. This means rebuilding a complete civilizational infrastructure on Mars, including chip factories.

Musk provided a relatively specific timeframe: If all goes well, and Starship capacity can grow exponentially according to the Mars launch windows (every two years), then in about 25-30 years, humanity could establish a self-sustaining civilization on Mars. This will require sustained efforts over about 10 to 15 Mars launch windows.

“Is the arc of civilization continuing to rise, allowing us to achieve self-sufficiency on Mars, or will it decline before that?” Musk queried. He believes that humanity has for the first time in its 4.5 billion-year history a window to make life multi-planetary. Achieving this would greatly extend the possible lifespan of consciousness, as human civilization would no longer rely entirely on the intactness of a single planet (Earth). In his view, this is more fundamental and important than what form of governance Mars adopts in its early days.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。