> We have not decided on timing yet; it may be a while because there are things we want to do that are likely easier as a private company.
Presumably those things were harder as a charity/non-profit.
Presumably those things were harder as a charity/non-profit.
Perhaps Larry Ellison can cut them a nice quid pro quo for a few months to make OpenAI look profitable (like the SpaceX/Anthropic deal), although that's probably unlikely given the debt Oracle is taking on to build it's infra.
I understand the scepticism around Google's deal with SpaceX, given the former holds a stake in the latter. But Anthropic buying SpaceX's compute doesn't have any related-party smell to it. That genuinely looks like SpaceX having cornered some valuable compute.
This is a reasonable accusation! It doesn't make a lot of sense–the Journal article is worth a hell of lot more than SpaceX referencing Anthropic's profitability. And we have zero evidence for it–one could raise this accusation against any compute partner Anthropic were to buy from.
Reasonable. The influencers who just learned the term circular financing are mostly idiots. The ones pointing out the conflict of interest with Google are technically correct, but the conspiracy takes so many moving parts to yield such little gain that it would have to be particularly stupid in vision yet competent in execution to pull off.
But asking if there is a quid pro quo between Anthropic and SpaceX? Like, there could be. We have no evidence of it. The S-1 mention doesn't make any sense. But they're both going public and if I were a journalist I'd look into it.
The base case, that there is commercial value to xAI's datacenters that folks in the frontier-model space are competing to get access to, does seem to be one folks here are actively rejecting.
That's nice way to say "invested in AI that turned out to be flop nobody wants to pay for so they are selling spare capacity"
Both takes are true. xAI invested in capacity that was supposed to yield frontier-model-maker margins. Grok failed to generate enough interest. So now they're selling it.
That's absolutely a good business, in a way that's more certain than the frontier-model one. But it's also lower margin, which doesn't support the sort of valuation SpaceX is going for.
Data centers (before recently) are low margin businesses because all the inputs are commodities: you buy power (joules), power (PDU), cooling hardware, physical racks, etc.. from the same vendors as everyone else. Worse, your biggest potential clients have the scale to just build it on their own and cut you out because of their scale and because you don't bring anything unique (outside of maybe physical proximity to an interesting market)
xAI has all the same commodity inputs plus another huge upfront capital expense (GPU/storage/networking), and their customer base is exclusively the well-funded companies who would normally just build it on their own.
I assume that they can't get better deals from nvidia than (e.g.) Microsoft because of their scale, so the unit cost of their inputs is the same or worse than their clients.
So the whole game is hoping that they hope to charge more now because people can't build fast enough and try to recoup their upfront costs before either a) other capacity comes online and b) the installed hardware becomes obsolete.
I'm being earnest -- it seems like they're trading one tiny margin service (datacenter) for another tiny margin service, with the added difficulty that there's an additional 10 figures of upfront expenditures and their viability depends solely on paying everything off before the price floor drops. Maybe it's staunching the bleeding, but it seems like not a great move
You're right that long term it should stabilize into a low margin business.
Elon is also much less risk averse than others, which helps to build stuff fast, possibly cheaper, pushing legality to the limit. Colossus was definitely built much faster than anything else. I think building datacenters suits him better than a pure software play, where "move fast break things" is already the norm.
WRT SpaceX building data centers: I think there's a natural tension between a "low margin business" and "being risk adverse". SpaceX (the rocket business) did well because it was high risk and high reward. Building a 10b datacenter to hope to get a slice of a low-margin industry is high risk and low reward and just seems fundamentally like a losing strategy.
Also I think stuff like Hetzner is a commodity. But are gigawatt scale data centers a commodity? You need those for AI training.
Anyways their goal is datacenters in space, not traditional data centers. Although I think that's only viable for inference.
AI compute hardware is not a commodity. And in a shortage, commodities can command high margins.
xAI has lots of NVIDIA GPUs and HBM. It also has permits and power hook-ups, both things that are getting harder to come by day by day in the U.S. Natural gas is a commodity. Doesn't make having lots of right now bad business.
> the whole game is hoping that they hope to charge more now because people can't build fast enough and try to recoup their upfront costs before either a) other capacity comes online and b) the installed hardware becomes obsolete
Correct. But charging people now generates incumbency advantages that make beating (a) and (b) easier. (From what I can tell, (b) isn't an existential issue, at least for xAI, because they've basically already recouped their investment with commited contracts they'd have to fuck up on to lose.)
I don't see the distinction you're drawing about "commodity", but I'm happy to be wrong on that. My point was that spaceX's ai division is buying all their inputs from external vendors and can't meaningfully differentiate themselves from person Y who buys all the same hardware except for the fact they bought them first. Which...
> Correct. But charging people now generates incumbency advantages
I don't see now this is an "incumbency advantage". There's nothing that sticks their clients to stay there and sign up for the next data center.
People pay markedly more for NVIDIA GPUs than they do for others. That opposes the fungibility requirement of a commodity.
I expect that Google are renting SpaceX NVIDIA GPUs so they can resell to corporate GCP customers at higher rates, but if the AI growth story remains intact then I would expect the GPU-agnostic token demand to be much higher than the NVIDIA-specific rental demand.
Which is why nobody should claim NVIDIA makes a commodity.
In the long term, hopefully the market stabilizes, new entrants can challenge Nvidia etc. But of course maybe not!
However for SpaceX, this is a dead end move. They made a good decision on buying this compute when they did but they failed to use it to create a compelling model.
So they're selling access to recoup some of their investment (maybe a profit?). But what's the plan as these chips age out over the next three to five years? Become a compute company? They claim they want to... in space!
Regardless, they bought some valuable chips, failed to use them, but can now sell access and recoup over the next few years before they become outdated.
Like they might have hired really good AI infra folks, gotten really good uptimes on their nodes, gotten folks who really know how to configure Infiniband (or whatever). But then, didn’t find the folks who knew what to run on that infrastructure. Or maybe Grok just had too much political drama around it.
EDIT: said 50 engineers at $50m/yr originally and meant 50 @ $1m/yr
There is also a question of how sustainable this datacenter rental demand is. It would seem unexpected if Anthropic and Google continue renting from SpaceX for more than a few years, and both contracts can be cancelled with 90 days notice.
When Anthropic spends on xAI, it benefits Google. When google spends on xAI, it benefits Google. When xAI spends on Google, believe it or not, that benefits Google.
This is how a Ponzi -style circular financing scheme typically works.
Unless Google is directing these transactions, this is not a novel issue. (We see a similar effect with mutual funds owning most companies [1]. It's a weak effect.)
> This is how a Ponzi -style circular financing scheme typically works
No. It's potential conflicts of interest. It's not circular financing. Circular financing follows the cash. When NVIDIA invests in OpenAI so OpenAI can buy NVIDIA chips, that is circular financing.
[1] https://insights.som.yale.edu/insights/the-rise-of-the-mutua...
Google has a fantastic balance sheet with or without these investments. None of the recent deals have uniquely enabled an IPO. So they'd be playing to increase their stakes' value by a few points ahead of a dump, a dump that would almost certainly wipe out much more than they'd stand to gain by trying to make someone else a dollar so they get nickels and dimes out of it.
This is literally true for any revenue. Treat the buyer and seller as a single company and their transaction is internal.
There is a lot of revenue dumping into this sector. If there weren’t, you’d have a point about manufactured numbers. But I don’t think anyone seriously doubts Anthropic and Google are hauling in serious dough.
The question, as you point out, is how much they are keeping. But xAI selling compute doesn’t really hide any of that. If anything, given the prices Musk is getting, it adds to the cost line. (And xAI isn’t masking compute revenue as Grok’s.)
In 10 years, we probably will have $700B/yr in productivity gains and revenue from LLMs, but we're not going to be able to sustain $700B/yr in capital spending until we get there. And the problem is much worse than the fiber buildout of the late 90s. Fiber built out in 1998 was still usable 10 years later. The GPUs that are being built out today are going to be obsolete trash in 3 years.
Eh given the quality of recent IPO proposals I think they can just say there's a couple zillion air molecules to turn into gold and be done with it.