Leaked financial docs show OpenAI is losing billions of dollars a year
- Thread starter JournalBot
- Start date
Upvote
145
(149
/
-4)
Who among us doesn't require fifteen short years for our gazillionaire unicorn company to maybe turn our first profit?
Upvote
116
(119
/
-3)
Even if R&D impossibly dropped to zero, the expenses are still larger than revenue.
There are companies that operated for years at a loss while building market share (or market dominance). Amazon is the poster child for that. But, for this industry, I see no evidence that a few years of losses are in any way an "investment" towards future profitability.
There are companies that operated for years at a loss while building market share (or market dominance). Amazon is the poster child for that. But, for this industry, I see no evidence that a few years of losses are in any way an "investment" towards future profitability.
Upvote
97
(99
/
-2)
How did they effing spend almost 6 billion in sales when this thing is pretty much selling itself? I must assume that line is where bribes are going.
Upvote
85
(86
/
-1)
It's a little disconcerting that their cost of revenue plus sales and a proportional share of admin is still more than revenue. But R&D is an expenditure against future revenue, not current, as long as the company has enough capital and credit to float itself until the R&D investment turns that future revenue around.
Upvote
-6
(13
/
-19)
I once thought that investors wouldn't keep dumping money into things that didn't make a profit for years and years. Then I watched as it happened with Amazon. Given that example, it wouldn't surprise me if OpenAI has a lot of runway left before the VC's demand a return on their investment.
Upvote
17
(18
/
-1)
I think that word in the caption in this article should be “income”, not “profits”.
Upvote
61
(61
/
0)
ubercurmudgeon
Ars Praefectus
I'm sure they'd like to think they're following the path to profitability that Amazon.com took, but they're looking more and more like the Pets.com of the AI bubble.
Upvote
48
(48
/
0)
B-b-but it's inference-profitable! But the world depends on it! But it's just about to get a boom of new paying users!
Suck it, robot cultists.
Suck it, robot cultists.
Upvote
32
(34
/
-2)
That maybe is doing all the heavy lifting. Especially with a technology that becomes more expensive the more people are using it.
Upvote
26
(28
/
-2)
Well, thanks to our well oiled capitalism, SpaceX is now buying cursor.
it’s all gonna burn down at this rate. Thanks to ignorant investors, SpaceX will just buy its way out of the grok problem. So it is with anthropic and OpenAI. there isn’t enough money in the VC space for them to succeed, and the market won’t let them succeed with this kind of cost problem.
Upvote
27
(28
/
-1)
chateauarusi
Ars Praetorian
Maybe it's just because I'm older, and I came of age in an era when companies' products and financials had to "make sense" and "be sustainable," but I look at today's business markets and I wonder if I'm going completely crazy.
Upvote
83
(84
/
-1)
My personal theory is that all of the first wave AI companies are going to be vaporized. Even if they tried to charge as much as it actually costs to build these tools, the market would not bear the price. The capital is sunk and not recoverable. Their lunch is going to be eaten by the second wave companies who did not have to front the R&D and will run sub-frontier, more efficient, possibly open-weight models.
Upvote
59
(61
/
-2)
Look at how much money (adjusted for inflation) was spent on Amazon over the years before it returned a profit (keeping in mind for a number of those years, they weren't profitable because they were reinvesting into Amazon what would have otherwise been profit). Then compare that to OpenAI's monthly spend.
I think the problem (and at least one reason for the IPO talk) is that VCs are running out of money and probably do want to see some of it back, so they need a bigger baghold..err pool of investors....
Upvote
22
(23
/
-1)
This reminds me of the early dotcom years, when investors would brag about the "burn rate" of the various companies. Naturally, it all came crashing down once those investors started asking for some returns on their investments.
The same is going to happen here, only with a louder crash and a wider splash of damage, thanks to all of the ETFs that have been forced to buy into the nonsense thanks to the rule changes.
The same is going to happen here, only with a louder crash and a wider splash of damage, thanks to all of the ETFs that have been forced to buy into the nonsense thanks to the rule changes.
Upvote
20
(21
/
-1)
The difference between OpenAI and Amazon is everyday people could see the value in what Amazon was offering. The same is not true of AI, despite their user numbers. Everyone paid to "use" Amazon through buying things, Amazon generally wasn't giving away product for free. Compare that to OpenAI, where only 5.5% of users are actually paying anything.
Upvote
53
(55
/
-2)
It’s an investment in getting other investors/bag holders to give you exit liquidity.
It CAN work. SpaceX just went up to 220 for no goddamn reason at all beyond number-go-up, giving it a market cap at its peak higher than Microsoft. Businesses that make more money than they spend are SO yesterday. Nobody gives a shit about that anymore.
It CAN work. SpaceX just went up to 220 for no goddamn reason at all beyond number-go-up, giving it a market cap at its peak higher than Microsoft. Businesses that make more money than they spend are SO yesterday. Nobody gives a shit about that anymore.
Upvote
37
(38
/
-1)
Considering the current AI company in the market as of last week is now at 2T, this clearly justifies an IPO price of 3T...right? /t
I don't get the value proposition of any of these companies at all, we're at such stupid bubble levels now.
I don't get the value proposition of any of these companies at all, we're at such stupid bubble levels now.
Upvote
21
(22
/
-1)
Oh that other $20 billion we lost? Never mind that, just an accounting charge!!
Upvote
9
(10
/
-1)
My layman's understanding was that Amazon could have become profitable at any point they wanted--and they could convincingly sell investors on this promise. That is totally unclear to me in this case, even as Sam spins bullshit the size of the sun.
Upvote
61
(62
/
-1)
Close. The caption is incorrect, but it should say "revenue", not "profits".
Upvote
5
(5
/
0)
Upvote
16
(16
/
0)
Okay. Hear me out. Maybe it's because AI mostly sucks and nobody wants to use it except for tech bros, venture capitalists, and the C suite.
Now excuse me. I have to tell my AI agent to buy some NFTs using crypto.
Now excuse me. I have to tell my AI agent to buy some NFTs using crypto.
Upvote
11
(16
/
-5)
~20B for R&D seems quite a bit. Are they hiding their training costs as R&D? Just asking questions 
Upvote
8
(11
/
-3)
I had the same thought; their marketing is free when every gullible journalist publishes some version of "it's sentient / scary / better than humans / almost AGI"
(none of which is true)
Upvote
39
(39
/
0)
The Amazon comparison is superficial; Amazon was operationally profitable, just refusing to take profit in order to fuel its expansion without expanding its debt burden. This is expanding debt burden to fuel growth while being operationally UNprofitable, because non-R&D expenses still exceed revenues.
I continue to not see a path to profitability for OpenAI, given the depreciation rate of its infrastructure and the sky-high cost of development.
I continue to not see a path to profitability for OpenAI, given the depreciation rate of its infrastructure and the sky-high cost of development.
Upvote
61
(62
/
-1)
I don’t see the cost of building and maintaining data centers anywhere. What am I missing?
Upvote
9
(9
/
0)
Not free. But especially in the early years, Amazon was selling stuff below cost ... and successfully bankrupted quite a few competitors.
Upvote
20
(20
/
0)
I cant wait to see the audited statements when they become publicly accessible.
Even when its done properly accounting standards on depreciation of intangibles like the model training costs need to be reevaluated. Do they keep rolling it forward despite new models being released multiple times a year? The old training has to be impaired by now if its irrelevant to the current models. It should be taken into income in the current period, not perpetually deferred.
There might be another Arthur Anderson accounting firm collapse when the bubble pops.
Even when its done properly accounting standards on depreciation of intangibles like the model training costs need to be reevaluated. Do they keep rolling it forward despite new models being released multiple times a year? The old training has to be impaired by now if its irrelevant to the current models. It should be taken into income in the current period, not perpetually deferred.
There might be another Arthur Anderson accounting firm collapse when the bubble pops.
Upvote
11
(11
/
0)
Upvote
8
(9
/
-1)
Indeed. I can imagine spending millions on sales and advertising, but billions
Upvote
10
(10
/
0)