Elon Musk on Self-Driving, AI Compute and Electricity Constraints

In an interview at Bosch ConnectedWorld shown in this March 2024 video, Elon Musk discusses self-driving, AI infrastructure, humanoid robots and vehicle electrical systems. Matthew Berman then comments on the interview and offers his own predictions. The timelines and projections are the speakers' views from that period.

Asked about AI over the next five to ten years, Musk points first to autonomous driving. He says Tesla has focused on making the technology work in the United States before expanding to more complex conditions elsewhere, and predicts fully autonomous cars are close. That is a prediction, not evidence that vehicles were generally driverless.

Musk describes a hypothetical trip to the Austin airport without intervention using cameras and neural networks. He compares cameras and digital networks with human eyes and biological neural networks. He argues that the car must interpret signs and infer the intentions of drivers and pedestrians, calling this a limited form of general intelligence.

Musk estimates that a passenger car is used around ten hours per week and guesses that autonomy could increase its use to 50–60 hours, as a vehicle could operate like a taxi. This fivefold-utilization scenario depends on future deployment and demand; it is his estimate rather than a measured result.

Turning to online assistants and image tools, Musk says he has never seen a technology develop as fast. He claims AI compute coming online is increasing tenfold every six months, while acknowledging that this rate cannot continue indefinitely. The figure is his observation, not an independently established growth rate.

Musk contrasts traditional explicitly programmed industrial robots with Tesla's planned Optimus humanoid robot. He anticipates broad capabilities and jokes that he hopes robots will be nice to people. The discussion describes an ambition, not a demonstration that the robot can perform almost any human task.

On manufacturing and productivity, Musk predicts major gains across manual labor, logistics and customer service. He calls AI an unusually large technological revolution and says concern about whether humans might become obsolete had unsettled him, before joking that he would rather be alive to witness whatever happens.

Asked about constraints on exponential growth, Musk says chip shortages were followed by shortages of voltage step-down transformers, and predicts electricity supply will become the next bottleneck. His joke is that AI transformers need electrical transformers to run. His forecast does not establish that every project will face the same constraint.

A Bosch interviewer asks about the Cybertruck's 48-volt electrical architecture. Musk contrasts it with longstanding 12-volt vehicle systems and says higher voltage can reduce the amount of copper needed for a given power load. He estimates roughly a quarter as much conductor in the comparison and predicts broader adoption; the actual saving depends on design.

The interviewer frames infrastructure as a race to keep up with technology, then thanks Musk. Berman explains that he has edited the Bosch interview and omitted some vehicle material to focus on AI-related points.

Berman notes that Musk has repeatedly predicted imminent full self-driving and has often been wrong about timing. He nevertheless describes himself as a technology optimist and believes driving without human intervention on roads worldwide may be closer than many expect. That is Berman's own forecast, not a demonstrated worldwide capability.

Berman explores how autonomous vehicles might be used more: instead of sitting parked most of the time, they could run errands or be shared among families. These benefits are hypothetical and depend on reliable autonomy, regulations and the economics of sharing.

Berman returns to Musk's rapid-compute-growth claim and wonders whether infrastructure spending will be a one-time buildout or continue as new chips and architectures appear. He expects continued investment, but presents this as his view about an uncertain market, not a verified forecast.

Berman is enthusiastic about humanoid robots and says he had predicted they would become prominent in 2024. He interprets Musk's comment about robots being nice as partly humorous and partly reflecting concern about AI. He also discusses people pushing robots in demonstrations and speculates about how sentient robots might respond; sentience is not established by these examples.

Berman agrees that AI could increase productivity. He acknowledges concerns about job losses but expects many jobs to change and new ones to appear, drawing an analogy with earlier technologies. The balance between displacement and creation is a prediction, not an outcome the interview establishes.

Berman discusses Musk's concern that a future AGI might dominate humanity. Berman says he does not personally expect an AI apocalypse but admits the future is uncertain. Both the existence of such a system and its consequences are speculative in this discussion.

On electricity demand, Berman repeats Musk's predicted shortage and notes Tesla's interests in charging, batteries and solar products. He speculates that the company could become a major energy business, while recognizing that Musk has a commercial interest in the topic.

Berman looks into step-down transformers, explaining their role in reducing high incoming voltage to levels suitable for equipment. Musk's view that supplies will run short is a forecast. Berman says he plans to research manufacturers and explicitly says his discussion is not investment advice.