Masayoshi Son, founder and CEO of SoftBank, told shareholders that building data centers in space will not meaningfully reduce costs and will arrive too late to matter.
Speaking at a recent meeting, Son argued that the next few years of the AI battle are far more critical than developments a decade away.
On the latest episode of TechCrunch’s Equity podcast, Kirsten Korosec, Sean O’Kane, and I discussed Son’s remarks alongside OpenAI’s custom chip plans and Groq’s $650 million funding round.
Kirsten noted the irony of Son acting as a skeptic, given SoftBank’s long history of wild bets.
Sean pointed out that Musk’s plan to replace satellites every few years to build an “orbital data center” simply guarantees more business for SpaceX.
The following is a preview of our conversation, edited for length and clarity.
Neo-clouds and pivoting businesses
Sean O’Kane: Neo-clouds are the new oil. Everybody who wants to make money is pivoting to them. TechCrunch is now a neo-cloud. Give us all your money.
There are so many compute-constrained players. Anyone with a shot at leasing out compute is taking it. Groq is one. It was semi-hollowed out by Nvidia.
Allbirds is another. It went into bankruptcy and emerged as a new neo-cloud provider instead of selling shoes. Tim Fernholz did an interview with the new CEO that I would recommend.
Then there is SpaceX. The idea was to build an AI platform with an addressable market the size of US GDP. Before that, they would rent out their compute.
SpaceX is not as big with deals like those struck with Google or Anthropic, but they signed another deal. It is their first post-IPO deal to rent compute to a smaller player. They are continuing down that road.
I can see this being a business for Groq in the near term. The question is how durable it is in the long term.
Space is not an immediate solution
Anthony Ha: If we are talking about SpaceX and their AI and data center business, we must also discuss Masayoshi Son’s recent comments. He asked: What is the point of data centers in space?
It speaks to the industry’s sense of being compute constrained. They need to build as many data centers as possible. There are reasons why that is proving challenging on Earth, so maybe space is the answer.
Son makes fair points. Even if it all works, the costs will be very serious. This is not happening for years. It is not a solution to any immediate problem regarding the current need for data centers.
SoftBank’s track record
Kirsten Korosec: SoftBank has a long history of making wild bets. It says something when Son asks the question that a lot of people have asked.
Many VCs and founders have been swept up in the idea of orbital data centers. Suddenly everyone is on board. A couple of years ago, mentioning this would get slapped down.
It is important that someone with a high profile is asking the question. But it is very ironic that he is the one asking it. Look at his pitch deck. They have thrown money at bold ideas.
Engineering and economic challenges
Sean: WeWork. We will be saying this for a lot over the next couple of years. Putting these things in space is an interesting engineering challenge. It is certainly an interesting economic challenge.
Elon Musk hates red tape. There are no NIMBYs in space. Of course he is going to try and do that.
It comes down to this: The business for SpaceX, especially its launch business, is overwhelmingly reliant on Starlink. They are 80 or 90 per cent of the global launch market.
This is not just because they have done better than other providers. It is because Starlink is driving up that number. If you remove Starlink from the equation, they would be closer to 20 per cent, 30 per cent, or maybe 40 per cent of the launch market. It certainly would not be 90 per cent.
When you talk about making a constellation of satellites that need replacing every few years, you are just guaranteeing much more business for your launch business. I cannot stop coming back to that point.
Renting compute
Kirsten: I want to say that SpaceX’s other big business is renting out their compute. Back to the chip conversation. We have come full circle.
Talking your own book
Anthony Ha: One theme running through this episode is talking your own book. This is not a new phenomenon. Executives at tech companies predict a future that is advantageous to their business.
It is worth remembering when having conversations about big AI companies. This is a moment of incredible uncertainty. We are wondering what the job market looks like, what effect this will have on the environment, and what skills we need to learn.
All these AI CEOs and investors have thoughts on that. It is not that they are wrong or being deliberately misleading. In each case, there is an asterisk to these predictions.
In Musk’s case, he is talking about something very good for SpaceX’s business. In SoftBank’s case, they are very heavily invested in data center projects on Earth. Sam Altman is the other notable figure who has rolled his eyes at the orbital data center idea.
He and Elon Musk obviously have a long and complicated history together.
All of which is to say there are no objective, impartial observers here. It is all these people with baggage and tremendous amounts of money at stake.
