HSBC spies $207B crater in OpenAI's expansion goals OpenAI needs to secure $207 billion in new financing by 2030 to fulfill its expansion plans, according to HSBC Global Investment Research – a challenge that could ripple across Big Tech. The funding shortfall emerged after OpenAI committed $300 billion to Oracle, $250 billion to Microsoft, and $38 billion to AWS for cloud …
"HSBC predicts that OpenAI's ChatGPT consumer products will attract 3 billion regular users by 2030, up from 800 million last month, and equivalent to 44 percent of the world's population over 15 years old."
Really? They must be assuming it's going to be hard to fight off. A bit like a virus.
Not to mention that ChatGPT is not available in China, so you can drop about a billion from the total.
3 is 44% of 6.82, so if we make that 5.82, they're projecting over half the adult population of the world that *can* access ChatGPT will be using it. (And that's *also* assuming sanctions on Russia are lifted by then). And 10% of those will be subscribing to a paid plan.
Let's...let's see how that goes.
> So with 88% of the world population over 15 as subscribers, a 10% price hike, advertising, licensing and assuming their costs stay the same, OpenAI might just about break even.
So assuming they could annihilate all competition and force every single human on the planet to pay for a CrapGPT subscription from cradle to grave, how long would it take them to pay their existing debt??
I really wonder wtf their investors and lenders are smoking
(And of course, their costs would NOT stay the same - they'd need a lot more chips and a lot more 'leccy for starters.. Unless the population just pay the subscription in spite of hours-late answers or 503 errors)
I really wonder wtf their investors and lenders are smoking
I thought that was a pretty good question. So I went over to https://chatgpt.org and asked "what are open AI investors smoking that causes them to think the operation can be profitable"
And I got back 40 lines of marketing drivel which I would reproduce here if I could copy it to the clipboard. Which I can't seem to. Perhaps Chat GPT prefers not to be quoted.
The marketing drivel did not address my question of what on Earth the investors are smoking. Whatever it is, it seems to be potent. And likely illegal.
the whole bloody tech industry! since i've been put off work for the last nearly 12 months because apparently "there's a shortage of technical talent in the UK" doesn't apply to anyone in infrastructure or cybersecurity & a uk passport, i've been hitting all the events in excel & about.
AI Agents are EVERYWHERE! Every firm, everything. Yet, everyone in line one says "agents are awesome, they're going to be a $trillion industry " and then lines 2,3,4,5,6,7 are about security risks, hallucinations, "keeping a human in the loop", data leaks, prompt injections , hacking, improper use, etc etc
it's worse than that dancing disease, it's like some crazy pyschosis that everyone in the tech industry has, caught, everyone KNOWS it's bullshit but everyone is pumping it.
on the plus side; like the grift that is cloud, there will be years of work reversing management FOMO & jumping on the bandwagon. years of work repairing the damage that these things do & years of hilarity as some companies that wanted to be "first & deliver quickly" will be first to the wall
The OECD is described as having 659.9 Million people of working age 15-64, employed and unemployed.
Do HSBC believe we're going to be volunteering a median of €6 quid a month to OpenAI, and they'll make money through us not actually making queries that cost OpenAI more than this + energy costs and datacentre costs decreasing greatly?
Maybe they think that commercial users will be paying for it on our behalf? What would that require, 180 per month for each adult in sales and advertising commission?
Why would HSBC think that the market for models won't be more fragmented, with the high value queries sent to LLMs that can statistically prove they provide consistently useful results when fed somewhat standardised input?
Chinese businesses will likely drive their own, and may be of more use due to localisation.
I'm sure that there will be gaming and leisure applications but again, the likes of Epic's Unreal Engine could provide trained models that run on NPU and their own service.
Honestly, I don't think they are thinking. They've got THE NEXT BIG THING by the tail and it's going to change the world and make they and their investors wealthy beyond imagining. They think they are Microsoft in 1990 rather than Segway in 2001 (remember them? They were going to change personal transportation forever). The gods of progress are on their side and nothing can possibly go wrong.
(And also, one suspects they are counting on us paying a good deal more than €6 /month for their invaluable service.)
You go ahead and sign deals to buy valuable real estate in Ney York, Paris and Dubai. Then go to your bank and explain that you only need $250 billion to pull it off.
See how well that goes (hint : it won't go well).
But because this is a oublicly traded company (that is continually losing money), somehow these deals generate enthusiasm on the stock market.
I thought those "golden boys" were supposed to be finance geniuses. To me, they're looking more and more like grifters who are just after feeding like vampires while something is going up and pulling out as soon as the deals start to sour, without taking any account of the total lack of backbone the company had to realize the deal in the first place.
Wall Street creates disasters, not money. In the best case scenario, when the economy is working, Wall Street is just leeching profits. It is most certainly not helping money flow.
Just a small correction, OpenAI isn't a publicly traded company (yet). If it was, it would have to follow those annoying rules about accounting, transparency, risk, etc. Far better to be a tech darling, burning through other people's money on wild promises about making every job redundant.
"Wall Street creates disasters, not money."
Perhaps because like most stock exchanges, NYSE long ceased to provide capital for startups, and is simply now secondary trading of existing businesses. IPOs are usually of very mature businesses and equity investors are totally unaccustomed to looking critically at business plans for new ventures and new markets. They think that stocks go up as if by magic, and talk of equity risk without actually taking much on. NYSE (and NASDAQ, LSE, Euronext etc) are just parking lots for fund-money (pensions, insurers etc) and playgrounds for those who can gamble with other people's money.
"HSBC predicts that OpenAI's ChatGPT consumer products will attract 3 billion regular users by 2030, up from 800 million last month, and equivalent to 44 percent of the world's population over 15 years old."
I hope there is some good research to back up these rather bold claims?
I wonder how many of them regular "users" will be AI bots?
It continually amazes me that these huge start-ups can continue to obtain funding based on unrealistic market penetration and earnings projections1, and then to pass a measurable amount of this to the founders and other high-up execs as salary and/or bonuses without these people ever causing the company to make a dime (most of them are American) in profit.
Having taken a cushy lifestyle, with a high personal profile in the start-up years, they will, in all probability, just walk away before the financial realities hit home, leaving with personal money and reputation, while the people who follow them will take the blame for the companies financial woes when the chickens come home to roost.
With smaller venture start-ups, there is a catch net. If it produces something of value, there is a chance that there will be an even larger company prepared to by it, but the supposed value of these AI companies is just so huge there is no guarantee that there is anyone who could afford to do this!
1The global penetration of these projections just beggers belief. If you believe the global inequity reports, over 50% of the worlds income goes to just 10% of the worlds population. This means that to achieve the figures, AI has to become an essential part of the lives of some of the poorest people on earth. Good luck on extracting significant revenue from them. They're more worried about whether they can put food on the table rather than paying a subscription service for something that AI can deliver!
I found this table which pretty well describes your point. https://www.worlddata.info/average-income.php If we look at some of the big populations, India comes in at around 3K annual income on average. So cross out India as a big subscriber group. As someone else said, I don't see China allowing subscriptions. So what cross off 2 billion possible consumers. And that table had many countries where I just don't see the population going, Let's pay for chatgpt and not eat this month.
They have convinced politicians that AI is the next moon landing/nuclear deterrent, so it is a national requirement to stay in the game. When the bubble bursts, their govt. chums will then bail them out with taxpayer's money. The AI stuff will then be reconfigured as a system of universal surveillance. You clicked the terms and conditions, so it can access everything on your system, it can be all be stored in those datacentres, and govt. spooks can access it.
The VC's tend to point to the last 20 odd years of "successes" and ignore external factors, especially the cost of credit, which was near zero for most of the time. There are two aims: get a big enough valuation to draw in real money from institutional investors and then, at some point, go public. Once you're listed, you can sort of rely on exchange-traded funds (ETFs) to keep people buying your stock. Without this, companies like Uber would have gone bankrupt years ago.
Seems like the same sort of circular argument I had with my mum 50 odd years ago about my need for a bike to cure my perenial lateness for,well, everything.
It didn't wash then, and it shouldn't now.
I think we need the solution she found
which was an alarm clock
Just ask the A.I. what's so special about the number 42
Oh and we need a priesthood for the A.I. with appropriate rituals etc
...... No, we have that already
Aha! now to wait for the US grid to implode, when they turn on all those data centers they are building.
I suppose it's the Venezuelan oil they plan to steal which will power the things.
TTFN
Maybe I see this differently.
We have gone through decades of wealth being pushed upwards by whatever means government can get away with. Tax breaks, privatisations, forced policies, taxation and so on.
What if we get to a point where government says we/it can't function without AI and/or AI is too big to fail as the worlds economy would collapse? Where does the money come from then? Maybe this is what they mean by 3bn people paying for it. You aren't going to subscribe but your government is going to subscribe for you and you will pay for it. That's what the bank bail outs were.
It remains to be seen but after 2008 the governments and regulators (Trumps America excepted now) have been running regular stress tests/regulations on banks to ensure that they can't fail in such a spectacular fashion.
Of course the problem is that they (the reckless ones) are always coming up with new and better ways to spin the financial wheel plus the governments are now so deep in debt after Covid that they *can't afford* to bail anyone out.
It is going to crash and it won't be pretty when it goes but the sooner it happens the less painful it will be. And see which government blinks first.
Setting aside industry and commerce, for which the potential benefits of 'AI' are highly variable, how might the 'consumer on the street' improve their quality of life by communing with ChatGPT in their homes or via mobile devices when out and about?
That question is left for other people to discuss. An ancillary matter to consider is how, what mostly will be simple or trivial demands on 'AI', would require people to buy-into commercial offerings.
Almost incomprehensible sums of money are poured into data centres stuffed with high-end 'chips', these generating sufficient heat to grow tons of tomatoes, and meanwhile draining freshwater courses. It must not be assumed that such concentrations of computational power guarantee deep or helpful insights to people taking out basic subscriptions. In essence, most people will be consuming a tiny slice of a collective utility; no specific, and protected share, will be allocated to each ordinary user; that works out as renting the equivalent of a basic present day Nvidia card or its like.
People will be connected to one of many multitasking behemoths which, when under routine demand, can offer only a tiny slice of the much-hyped abundance of 'knowledge' and 'intelligence'. A currently only moderately high-end personal device can privately run one of many 'AI' variants freely available from the Internet. Additionally, the variants shall be available from across the globe, China being a major player. Moreover, personal computing devices continue to become more powerful and at lessening price.
Another possibility is for personal computing devices to ship with a local 'AI', just as many now offer no choice about having MS Windows or Android. The local 'AI' could be designed to contact a larger, remote one when 'it' judges a need to consult a bigger database. Even so, unless OpenAI, and its partners, succeed in locking down devices similarly to preinstalled MS Windows and Google's proprietary version of Android, there will be immense competition: don't forget China and BRICS.
"HSBC predicts that OpenAI's ChatGPT consumer products will attract 3 billion regular users by 2030, up from 800 million last month, and equivalent to 44 percent of the world's population over 15 years old."
I wonder which AI system they asked to come up with that prediction? What is even better is that just 300 million of those regular users are expected to be paying for the service.........
Getting the amount of power and then radiating the large amount of resultant heat would be a huge barrier for current AI infrastructure in space.
And if they could reduce the power and heat requirements, then doing it on Earth would become easier in the first place.
Was I missing something? Was there supposed to be a joke icon?
As OpenAi are widely speculating a market share of 3 billion - and a quick population by continent puts only 2.5 billion OUTSIDE Asia.
That's assuming Russia and Europe at large don't eschew them.
China certainly won't let them get a market drop.
India - I'm on the fence about.
Point being their figure of 3 billion people is not aggressive, it's a moonshot figure.
Why not simply put 30 Billion as we're making up potential, I mean sounds better right?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
