Masayoshi Son Dismisses Musk's Space Data Center Vision, Says AI Race Hinges on Ground-Based Computing Power
Taylor Wilson
SoftBank founder Masayoshi Son told shareholders that space data centers save on electricity — but power is a small slice of total cost. Chips are the real expense, and the AI race will be decided on Earth.
Space data centers sound futuristic — what's the catch?
Son's core argument: the main selling point of orbital facilities is cheap solar power, but electricity is a small share of total data-center operating costs.
This means → even if power dropped to zero, the savings would be modest — chips and hardware are where the real money goes.
On top of that, there are costs the electricity pitch ignores: launch expenses to get equipment into orbit, in-orbit maintenance, and unavoidable communication latency between ground and space.
In plain terms = space data centers save a small line item and add several large ones — the math doesn't work.
Why does Son say "first movers win"?
He believes the critical window in the AI race is the next few years, not a decade from now.
This means → rather than betting on a far-future technology path, the smarter play is to pour capital into ground-based compute that works today and seize the lead now.
SoftBank has committed roughly $65 billion to OpenAI and plans to channel hundreds of billions into global data centers and related infrastructure.
This reflects a clear strategy: trade money for time, trade scale for first-mover advantage.
Musk and Bezos are going orbital — how does Son respond?
SpaceX and Blue Origin have both announced orbital data-center plans, aiming to bypass Earth's energy and space constraints.
Son praised Musk as "an extraordinary change-maker" — but his position is firm: SoftBank will not follow the space route.
In plain terms = he respects the person, but vetoes the plan — SoftBank's money stays on Earth.
How much room is left in the AI market?
Son acknowledged that competition is intensifying, but argued that OpenAI and its main rivals Anthropic and Google still have ample room to grow.
He sees AI as still in its early stages, with market potential for ten-fold or even hundred-fold expansion ahead.
This means → in his view, it is far too early to call a winner-take-all outcome — the pie itself is still growing fast.
Content is for reference only, not financial advice.