At its recent shareholder meeting, SoftBank's Masayoshi Son declared a clear strategic direction for the company's AI ambitions: focus on building overwhelming data center capacity on Earth, not in space.
This decision lands squarely in the middle of the biggest challenge facing the AI industry today—an insatiable demand for computing power running headfirst into the physical limits of our planet's energy grids. The AI arms race is real, but so are the bottlenecks in electricity supply. Son is making a bet on solving the immediate problem rather than chasing a long-term vision.
Several key factors drive this 'Earth-first' strategy. First is the urgency of demand. The AI revolution is happening now, and companies need massive computational power immediately. Waiting a decade or more for speculative space-based solutions is not a viable option. Second is the critical role of energy. Recent moves by U.S. regulators to fast-track grid connections for data centers make terrestrial builds more attractive than ever. This is also why SoftBank is investing up to €75 billion in France, which offers a stable and abundant supply of nuclear power. It's a pragmatic choice to build where the power is reliable and accessible.
This pivot also serves as a direct response to competitors like Elon Musk, who have floated the idea of 'orbital data centers.' Son is effectively drawing a line in the sand, arguing that while space may be the future, the decisive battles of the next decade's AI race will be won on the ground. The focus must be on securing land, power, and essential hardware like GPUs and HBM memory here and now.
In Japan, this strategy is particularly relevant. While the country is ramping up its AI infrastructure, as seen with Microsoft's $10 billion investment, it still faces challenges with power availability. SoftBank's commitment to leading a domestic build-out addresses this gap directly, positioning the company at the forefront of Japan's AI catch-up effort.
Ultimately, Son's declaration is not an abandonment of futuristic ideas but a disciplined capital allocation choice. The highest and most certain return on investment over the next decade lies in solving the tangible, terrestrial challenges of power and supply. For SoftBank, winning the AI race means conquering Earth first.
- Data Center (DC): A facility used to house computer systems and associated components. They are the physical backbone of the internet and AI.
- Power Purchase Agreement (PPA): A long-term contract between an electricity generator and a customer to buy electricity at a pre-agreed price, crucial for securing stable energy costs for large projects.
- HBM (High Bandwidth Memory): A high-performance type of computer memory essential for training and running large AI models.