Musk Points Out AI's Second Major Bottleneck is Ground Power Shortage, Promotes Space Computing and Starlink Upgrades
Musk clearly stated that the second core bottleneck for AI development is the shortage of ground power. In many regions, the approval of new power plants has been hindered due to residents' environmental concerns, forcing many AI computing factories and chip plants to delay production.
He believes that the upper limit of civilization depends on the total amount of available energy, with the goal of expanding the current scale of human energy use by a million times. The solution is to build space AI computing centers in orbit, utilizing the endless solar energy in space to break through ground power limitations.
Musk also anticipates the next major bottleneck: the speed of human information transmission is too slow (hundreds of bits per second), while the bandwidth required for AI and robots is at least ten million times higher. SpaceX's new generation Starlink V3 will increase bandwidth by about 100 times and reduce latency by half, potentially becoming the underlying connection method for most networks, replacing existing fiber optics.
SpaceX's next potential plan for a massive increase in market value is a lunar base, which, relying on the mass production of Starship, could deliver 1,000 gigawatts of computing power to space each year. In the long term, AI satellites could be launched from the moon using low-gravity electromagnetic catapults, with annual computing power output expected to reach 1 million gigawatts.
Source: Public Information
ABAB AI Insight
Musk has previously emphasized the synergy between energy and computing power through xAI, Tesla's energy business, and SpaceX. This public acknowledgment of the ground power bottleneck continues his long-term "first principles" thinking, having predicted multiple times between 2024-2025 that AI training would face a power ceiling, and pushing Starlink towards higher bandwidth evolution.
On the capital front, SpaceX is leveraging Starship's mass production capabilities and Starlink V3 as dual engines, addressing current ground power constraints while preparing the infrastructure for AI/robotics era with a thousand-fold bandwidth. Funding is accelerating towards vertically integrated aerospace-energy-computing companies, and the lunar base plan further amplifies long-term growth potential.
This strategy is similar to Tesla's early simultaneous layout of batteries, cars, and energy to solve the entire electrification chain bottleneck. Currently, AI infrastructure is at a critical transformation stage from ground centralization to space distribution.
Essentially, this is a reconstruction of the industrial chain: AI computing power and network infrastructure are migrating from ground resource limitations to infinite solar energy and low-gravity launches in space, as Starlink V3 and the lunar base can simultaneously address both power and bandwidth bottlenecks, shifting pricing power from traditional ground data center operators to companies with full-stack aerospace capabilities.
ABAB News · Law of Cognition
Energy and bandwidth are always the true limits of civilization; whoever moves computing power to space first locks in a million-fold growth window early. Taking action to solve issues before ground bottlenecks arise is the Musk way of leading; true experts never wait for problems to hit them. Starlink is not just a satellite, but the framework of the future internet; the lunar base is not a dream, but the launchpad for the next generation of computing factories.