The mechanics are legal but the objective is the same which is buy your own output by manufacturing a demand. I am sure they can buy good lawyers to keep it completely legal.
There is nothing illicit or illegal in anyway about what NVidia is doing. It's reasonable business practice, and people on HN are simply ignorant to think otherwise.
NVidia is very aware of the risks it entails, but has the money to cover those risks.
Which is exactly the point of the person you're responding to. What part of “but completely legal” isn't clear enough?
(Also note that illegal != illicit.)
The OP clearly is implying that it should be illegal for some reason. This is wrong - not only is it nothing like Enron (!?) but it's a great way for both NVidia and the companies building on them to build what they want.
1. The equity investments Nvidia has made in its customers.
2. The guarantees/backstops it has extended to some of its customers.
3. Vendor financing.
The vendor financing is the least interesting of the bunch. Nvidia has already disclosed that when it provides vendor financing, the average customer pays in less than 60 days. These are not long-term financing arrangements and virtually every big company sells on these type of terms (net-30, net-60, etc.).
The equity investments and guarantees are where there is room for legitimate debate.
The guarantees dwarf the equity investments. If there is a shenanigan, it's going to be there.
A problem: the line between the guarantees and traditional vendor financing is blurry–one could argue use commitments are no different from repurchase commitments.
I agree that the guarantees are where the risk is.
First, under US GAAP accounting rules (ASC 606), these are absolutely not repurchase agreements. The customer takes title to the asset (the chips) and Nvidia does not have a contingent obligation to repurchase the asset. Providing a contingent guarantee to purchase services is not a repurchase under the accounting rules, but of course you can have legal accounting and still have a problem.
As I've made public market investments in and traded in this space, I've done some math on the guarantees and what came out was this: relative to Nvidia's current earnings, the guarantees amount to approximately a quarter of a year's revenue at the guarantee cap.
That's my own analysis and I'd encourage anyone who cares to run the numbers themselves. It's easy enough as Nvidia is publicly traded.
The thing that differentiates Nvidia from previous vendor financing examples like Lucent in the 1990s/early 2000s is that its margins are huge. So what I see, based on the current numbers, is that in the really ugly scenario, Nvidia has to live with depressed earnings, no stock buybacks and a weaker (but still comparatively strong) balance sheet for a number of years. This is a stock problem, not a solvency issue.
The wildcard is if Nvidia keeps extending big guarantees, or starts using debt to do so, to the point where the commitments are expanding faster than its cash flow. At that point, the risk obviously compounds accordingly.
It can be manufacturing demand that wouldn't otherwise exist. It can be facilitating demand to come online sooner and smoother. You can't tell which it is by only looking at the transaction; you need to know how many dollars are going into the ecosystem as a whole for purchases of goods and services. Until Anthropic's S-1 lands tomorrow-ish, we won't have that publicly.
I wish we still lived in an age where things were sold to cover their costs.
My money's on their product having 10% of its current user base if they charged 50 cents per month.
How much would the average normie pay if Google suddenly charged? Sure the user-base would drop.
And what? Did Google suddenly become a bad business?
I find it funny how people have these strange and hypocritical viewpoints when it comes to OpenAI and Anthropic when the hyperscalars like Google, Amazon etc, followed these exact same kinds of playbooks for years and years.
Google Searches, once the infra setup was finished, were ridiculously cheap. Not the case for OpenAI & co.
Inflation adjusted, launching most of Google's core services (Search, Gmail, Maps, etc) cost less than launching ChatGPT 5. Running them at similar scales is also much cheaper per user.
There is a point at which it's just too much.
Oh, Google and Amazon never had these kinds of humongous losses and they reached profitability much sooner.
Yes, I'm against all predatory economic behavior. Selling user data, price dumping, platforms/walled gardens/monopolies/oligopolies/cartels/etc.
Google+ was the fastest-growing social network of all time not that long ago.
When you're one of a pair of VC darlings that have effectively-infinite money because of -in part- handwavy promises to cure cancer and eliminate 90% of payroll everywhere, or if you're an established company that has total control over very widely used consumer products, you can do all sorts of things to manufacture amazing growth numbers.
In the case of those VC darlings, we're seeing their shift towards providing their products that are most expensive to create exclusively to B2B customers and also the shift towards justifying the elimination of most of their R&D expenditure. Every company performs belt-tightening in advance of their IPO, [0] and those two are no exception.
[0] ...which is when their finances will be scrutinized by the public and regulators...
The world is fundamentally and momentously being rewired.
I'd argue that in this case, if the flywheel never reaches critical velocity, they are manufacturing demand.
Maybe I'm in the middle of the road :)
Seems like economy, scale, velocity, and even "critical mass" are related in some way, but not the same at all.
I would say it's quite possible that economies of scale can go from positive to negative without much warning.
I think it's most sustainable financially when the underlying "economy" is what drives the resulting scale-up, which usually does occur in phases or stages where each successful milestone informs the next campaign more realistically than you can get any other way.
As market demand grows beyond baseline sustainability it becomes less costly to serve each additional customer this way.
The opposite effect could occur if meeting lofty scaling goals requires an ever increasing cost of customer acquisition beyond the point of unmet initial pending demand.
When the scaling process itself is what drives the activity without being limited by the actual buying power of the ultimate consumers at any one point, things can really get ahead of themselves. Accounting practices can be so diverse that the only way to be sure whether scaling ahead of the curve was actually "economical" is after liquidation ends up occurring.
Unfortunately, liquidation of one kind or another is more likely when the scale is based on hyperbolic dreams rather than more reasonably optimistic estimates. But who's to say which is which, and the continuum between them is blurry enough without any highly interested parties trying to muddy the waters even further. Who even knows if they've given it as much thought as it deserves, or if more clear-headed thinking could be the primary factor given what there is to work with :\
Hence the designation "Hyper-Scalers".
Do you mean economies of scale?
Eh, I think it's an open question whether OpenAI and Anthropic would be buying GPUs like they are with or without Nvidia's financing. Financing customers' purchases isn't proof per se of demand creation versus demand inducement. Anyone who claims they've seen a certain fact in these financings is deluded or lying.
You need to be more specific, because there are absolutely sections of the AI economy that are clearly and presently profitable.
> Ai farms and much of Ai software world are money sucking machines
If AI farms refers to datacenters, plenty of existing ones are currently profitable.
Whose assets? Where are you getting this from to be able to state it with this level of certainty?
But here a reference; https://www.tomshardware.com/pc-components/gpus/datacenter-g...
This is just extremely unbelievable to me. I am certainly not operating at a level the hyperscalers are and have much more limited direct experience. But I do actually put various GPUs inside datacenters (and much harsher locations) and have operated them at balls-to-the-wall 100% utilization for over a decade now.
You get the typical bathtub curve of failures. Unless the hyperscalers are operating these things even more overclocked and beyond thermal specification limits than early GPU crypto miners used to do, I simply cannot believe that the average hardware life is less than the useful life of the whole chip generation itself.
I have plenty of decade old GPUs that operate today just fine. Both consumer and datacenter form factors. The failures tend to be board-level like capacitors and such, so if you are operating at a massive scale partnering with someone who can fix those relatively cheaply is not all that difficult either.
It could be that these H200 and above class sort of stuff is engineered extremely fragile, but I seriously doubt it. The prevailing "common knowledge" pre-AI for GPUs were that they'd burn out in a year or two of heavy use, and that was simply untrue. I saved a ton of money buying batches of used units because everyone was terrified of this - and had no more early failures than I did buying brand new after basic refurb of re-pasting and putting a new fan on them.
I've been pitched data-center deals. They depreciate on an 18- to 24-month schedule, well under the Tom's Hardware terms. The ones who went online a year or two ago aren't losing their chips like ducklings through a storm gate; if anything, their resale value has remained remarkably stable because compute production is the bottleneck.
You've given a source (a great one, btw) for depreciation but not revenue. If you can name a company, I can look if I have a public source that confirms what I know. But broadly speaking, no, unit economics in the AI economy is weirdly sound, though I suspect it's because every non-AI CEO is blowing out their budgets on frivolous spending.