SoftBank’s CEO isn’t the only one with questions about Elon Musk’s orbital data center hype

Not everyone seems to be shopping for Elon Musk’s imaginative and prescient for orbital information facilities. 

Masayoshi Son, the founder and CEO of Softbank, argued at a latest shareholder assembly that constructing information facilities in house gained’t do a lot to chop prices and can take too lengthy when “within the battle for AI, the subsequent few years will probably be way more vital than what may occur a decade or so from now.”

On the newest episode of Trendster’s Fairness podcast, Kirsten Korosec, Sean O’Kane, and I mentioned Son’s remarks as a part of a broader dialogue that included OpenAI’s plans for customized chips, chipmaker Groq’s new $650 million funding, and far more.

Kirsten famous that it’s “very ironic” that Son is taking part in the skeptic right here, given SoftBank’s “lengthy historical past of untamed bets.”

Sean, in the meantime, mentioned that when Musk talks about “making a constellation of satellites — satellites that should be changed each few years as nicely —  to make up an ‘orbital information middle,’” he’s simply “guaranteeing that rather more enterprise” for SpaceX.

Maintain studying for a preview of our dialog, edited for size and readability.

Sean O’Kane: Pay attention, neo-clouds are the brand new oil, and all people who desires to generate profits is pivoting to a neo-cloud. I’m proud to announce that Trendster is now a neo-cloud, give us all of your cash.

I imply, that is the factor you do. It looks as if there are such a lot of gamers which might be compute constrained, so anyone who has a shot at having the ability to lease out that compute is taking it, whether or not that’s Groq, an organization that was semi-hollowed out by Nvidia, or Allbirds, which went into chapter 11 and and emerged from it as a brand new neo-cloud supplier as an alternative of promoting sneakers — Tim Fernholz did an interview with the brand new CEO of of that new effort that I might positively suggest individuals go learn. 

Or whether or not you’re SpaceX, the place your concept was: I’m gonna construct an AI platform that’s gonna have an addressable market the dimensions of U.S. GDP, however earlier than we get there, we’ll simply lease out our compute.  And we noticed this proceed to occur with SpaceX, the place it’s not as large because the offers that they’ve struck with Google or Anthropic, however they simply signed one other deal, [their] first submit IPO deal, to lease out compute to a different smaller participant. They’re persevering with down that highway. 

You already know, I can see this being a enterprise for Groq within the close to time period. The query with all of those is how sturdy is it in the long run.

Anthony Ha: If we’re speaking about SpaceX and their AI enterprise and information middle enterprise, we even have to speak about these feedback that Masayoshi Son, the CEO of SoftBank, made not too long ago, the place he mainly mentioned: What’s the level of knowledge facilities in house? Which is a query we’ve requested on this present. 

And it speaks to, once more, this sense within the business of being actually, actually compute constrained — they should construct as many information facilities as attainable, [and] there’s every kind of the reason why that’s proving to be difficult right here on Earth, so possibly house is the reply. However I believe Son makes some fairly honest factors about: All these items we’re speaking about, even when all of it works — and the prices are going to be very, very severe to make it work — this isn’t occurring for years and years and years, so this isn’t an answer to any quick downside, as far the present want for information facilities goes.

Kirsten Korosec: I simply need to level out that SoftBank has a protracted historical past of creating wild bets. I believe it says one thing when Son comes up and asks the query that lots of people have requested. 

I imply, there are a number of VCs and founders [who] have been swept up into the thought of orbital information facilities and it looks as if all of the sudden everybody’s on board. When simply a few years in the past, I believe, if somebody had talked about that, it will get slapped down a bit bit. So I do suppose it’s an vital a part of the method that somebody who has a fairly excessive profile is asking that query. However it is vitally ironic to me that he is the one asking it, as a result of should you take a look at his pitch deck, they’ve thrown some huge cash at some fairly daring concepts.

Sean: WeWork! Pay attention, we’re going to be saying this for lots over the subsequent couple years. The concept of placing these items in house goes to be an attention-grabbing engineering problem and positively an attention-grabbing financial problem. 

Anthony, what you mentioned is certainly proper to a sure extent. Elon Musk is an individual who hates purple tape and , there aren’t any NIMBYs in house so after all he’s going to try to try this. 

To me, it comes right down to: The enterprise because it stands now for SpaceX, particularly its launch enterprise, is simply overwhelmingly reliant on Starlink. The rationale that they’re 80 or 90% of the launch market globally isn’t just as a result of they’ve achieved all these items which might be higher than just about each different launch supplier across the globe, it’s additionally as a result of they’ve Starlink that’s driving up that quantity. For those who take away Starlink from the equation, they’d be nearer to — I don’t know, possibly 20% or 30% of the launch market, or 40%, but it surely actually wouldn’t be 90%. 

And if you speak about making a constellation of satellites — satellites that should be changed each few years as nicely —  to make up an “orbital information middle,” quote unquote, you’re simply guaranteeing that rather more enterprise on your launch enterprise. And I simply can’t cease myself from coming again to that time.

Kirsten: I need to actually shortly say that [SpaceX’s] different large enterprise is renting out their compute, by the way in which. So again to the chip dialog. We’ve come full circle.

Anthony: One of many different themes which will run by way of this episode is this concept of speaking your personal ebook. This isn’t a brand new phenomenon. Executives at tech firms, or every other firm, what they’re predicting for the longer term is in the end the longer term that’s going to be advantageous to their enterprise. 

However I believe it’s one thing that’s simply all the time value remembering once we’re having these conversations about large AI firms, as a result of it’s this second of unimaginable uncertainty, and we’re all questioning: What does the job market appear to be sooner or later? What impact is that this going to have on the setting? What are the abilities I must be taught? 

All these AI CEOs or AI traders, all of them have ideas on that. And it’s not that they’re mistaken or that they’re being intentionally deceptive, however in every case, there’s an asterisk to those predictions. In Musk’s case, he’s speaking about one thing that might be excellent for SpaceX’s enterprise. In SoftBank’s case, they’re very, very closely invested in information middle tasks right here on Earth. Sam Altman is the opposite notable determine who’s rolled his eyes a bit on the orbital information middle concept — and once more, he and Elon Musk clearly have a protracted and complex historical past collectively.

All of which is to say that there’s simply no goal, neutral observers right here. It’s all these individuals with baggage and super quantities of cash at stake.