Summary of the Core Debate
The argument between Masayoshi Son and Elon Musk regarding the cost of AI computing power is not about who has miscalculated; it reflects a difference in time scales and strategic approaches: Son focuses on the tangible benefits of short-term infrastructure development and hardware ecosystems (scalable revenue within three to five years), while Musk is investing in long-term space-based computing solutions (aiming to overcome Earth’s energy and regulatory constraints in twenty to thirty years). Their disagreement highlights the tension between short-term quick profits and long-term transformative opportunities in the development of AI computing power, which could shape the wealth landscape of the next decade.
Detailed Analysis
1. Why is Son's “7%” figure merely theoretical?
Son mentions that electricity costs account for 7% of AI computing power expenses, but this assumes electricity availability. When Musk built a supercomputer center in Tennessee, the local grid had only 8MW of capacity (sufficient for tens of thousands of households), which was insufficient. He was forced to build his own natural gas power plant but faced community complaints about pollution, delaying the project.
How slow is infrastructure development in the US? Building power plants and transmission lines takes an average of 10 years, and transformers take even longer. Additionally, there are environmental reviews, community hearings, and legal disputes, which often derail data center projects. For Musk, electricity is not just a cost factor but a critical prerequisite; without sufficient power, the computing center would be useless.
Son’s calculations are based on an ideal scenario, whereas Musk deals with real-world infrastructure limitations that cannot keep up with the exponential growth of AI computing power.
2. The game of time: Quick profits in three to five years vs. a bet on the future?
Son’s strategy is to see immediate results: He invested in NVIDIA in 2017 and doubled his investment by selling it in 2019; later, he heavily invested in OpenAI and Arm, becoming Asia’s richest person this year. His approach focuses on “ground infrastructure and hardware ecosystems” to achieve scalability and cash flow within three to five years—similar to short-term stock trading.
Musk, on the other hand, adopts a long-term leveraged strategy: SpaceX plans to send millions of tons of equipment into orbit to utilize space-based solar energy (operating 24/7) and a vacuum environment, bypassing Earth’s power grids, land constraints, and regulatory hurdles. He is betting on the future (2035–2045), like building a new, uncharted path with potential for exclusive profits.
3. Breaking the rules: If ground-based solutions fail, try another space-based approach!
Musk’s approach is often rebellious: If traditional rules hinder progress, find a different physical environment to rewrite them:
- Using methane in rockets (which can be produced on Mars, eliminating the need for Earth’s kerosene);
- Tesla’s use of electricity (replacing oil, aligning with planetary energy structures);
- Moving AI computing power into space (avoiding Earth’s licensing processes).
It’s like opening a coffee shop in a residential area that gets rejected by the management—just move to space, where you can generate electricity from sunlight 24/7 without any restrictions.
4. The turning point: Whose logic will be proven wrong by reality?
Two key factors will determine the outcome:
- The speed of grid upgrades: If the US accelerates its power grid improvements, Musk’s vision of overcoming energy constraints may become obsolete;
- The frequency of data center delays due to electricity issues: If more projects are delayed, Son’s 7% figure will be meaningless if there’s no available electricity.
Interestingly, if Son’s short-term strategy proves successful and everyone invests in ground-based computing power, it could further strain the grid, giving Musk’s long-term vision even more momentum.
5. The dilemma of capital: Short-term sentiment vs. long-term value?
The market behaves like a voting machine in the short term (investing based on trends) and a weighing machine in the long term (evaluating actual value), but the most profitable periods often lie in between.
Currently, capital has not yet made a choice: Supporting Son is like betting on three-to-five-year returns, while backing Musk is like betting on a twenty- to thirty-year transformation. The real question is whether this tension between short-term profits and long-term growth will guide capital towards the right opportunities in the next decade.
Final Conclusion
The debate between Son and Musk is not about right or wrong, but about choosing between current wealth and future potential: Do you want to make quick money in three to five years or invest in a transformative future? This is not just their decision; it’s a question every investor and professional must consider.