Where's the money for AI coming from?
Discussion
I see AI firms getting trillion dollar valuations.
I understand business usage but the valuations are based on mass appeal.
Yet outside technology not many people actually pay to use AI. So how would you convert Joe bloggs to pay for AI , and make the valuations not appear like a massive technology bubble it feels like?
I understand business usage but the valuations are based on mass appeal.
Yet outside technology not many people actually pay to use AI. So how would you convert Joe bloggs to pay for AI , and make the valuations not appear like a massive technology bubble it feels like?
There’s more than one way to monetise, Facebook, Google etc don’t charge end users either.
But I suspect that as we allow AI to actually start managing parts of our digital lives (email, data, calendar, financial etc) rather than just treating like a slightly smarter version of Google we will also become more willing to see value and pay for the services it offers.
But I suspect that as we allow AI to actually start managing parts of our digital lives (email, data, calendar, financial etc) rather than just treating like a slightly smarter version of Google we will also become more willing to see value and pay for the services it offers.
Ironically I asked AI this a while back 
The maths worked out that every single office worker and paying consumer in the developed world would need to maintain 2 to 3 active premium AI subscriptions (e.g., ChatGPT Plus + Microsoft Copilot + Claude Pro) nonstop to make the current valuations remotely tenable.
A bigger problem is circular financing between the chipmakers, hyperscalers and AI developer, it's all a house of cards
The maths worked out that every single office worker and paying consumer in the developed world would need to maintain 2 to 3 active premium AI subscriptions (e.g., ChatGPT Plus + Microsoft Copilot + Claude Pro) nonstop to make the current valuations remotely tenable.A bigger problem is circular financing between the chipmakers, hyperscalers and AI developer, it's all a house of cards
In the long term, I think it will make an awful lot of money - it's a game changer, and will become central to how we use computers / phones / the internet. Businesses will certainly be paying for it, and as already mentioned - even without that, there are ways to make money without subscription models.
I am currently paying for Claude Code at home for private use, although admittedly I have a tech / software background. But it's a relatively small amount of money - basically similar to Netflix - and I currently get real use out of it. I think it's possible that an increasing number of people would pay in the future - perhaps initially growing fairly slowly.
After all, I'm old enough to remember the vast majority of the population being quite convinced that they didn't need to carry a telephone around with them all day!
I am currently paying for Claude Code at home for private use, although admittedly I have a tech / software background. But it's a relatively small amount of money - basically similar to Netflix - and I currently get real use out of it. I think it's possible that an increasing number of people would pay in the future - perhaps initially growing fairly slowly.
After all, I'm old enough to remember the vast majority of the population being quite convinced that they didn't need to carry a telephone around with them all day!
davek_964 said:
After all, I'm old enough to remember the vast majority of the population being quite convinced that they didn't need to carry a telephone around with them all day!
This is a very good point. When I bought my first mobile in 1990 texting didn’t even exist. Every contract was £25/month + 25p/minute - a lot of money in 1990!
The sales guy was raving on about how one day everyone would have a phone, landlines would be dead and we would all keep in touch through personal communication networks. It sounded deluded at the time but in fact he was barely scratching the surface of where we are today.
If you described some of the tasks no-subs AI can do now to someone even just five years ago they would think it sounds like Science Fiction.
I know the AI companies are creating profiles on users.
Oracle has already created 100 million plus profiles, that are used like black mirror cookies to model voting etc and model reactions to polices.
Which is probably where the most value is I guess. Pretty messed up considering people never actually consented to it but it was developed from pooling data.
Oracle has already created 100 million plus profiles, that are used like black mirror cookies to model voting etc and model reactions to polices.
Which is probably where the most value is I guess. Pretty messed up considering people never actually consented to it but it was developed from pooling data.
Edited by Flyingakite on Friday 30th October 10:42
The whole AI category seems to cover quite a broad range of technology companies. Samsung, Microsoft, Nvidia, Google etc all fall under the AI category, Nvidia alone is worth over 1 trillion dollars which raises valuations and expectations on the general AI sector.
Then you have companies like Micron and SK Hynix that make the semiconductor's, they are earning 2-300 billion dollar's each so far this year, and those earnings have risen by 2-300% this year. They are unable to keep up with demand and are sold out for the next 2-3 years, that has driven the cost of their products to all time highs which also raises the over valuation and expectation on the value of AI.
The money being invested into AI is staggering, and some of that is borrowed ( in some cases $60 billion at a time) while these companies are beating earnings expectations the overall value will just keep climbing to staggering levels, once demand slows down and earnings fall short on expectations then we can expect some sort of AI bubble burst. If its severe enough, then expect a recession in some countries such as Korea and Taiwan, America would also look rather shaky for a while.
Then you have companies like Micron and SK Hynix that make the semiconductor's, they are earning 2-300 billion dollar's each so far this year, and those earnings have risen by 2-300% this year. They are unable to keep up with demand and are sold out for the next 2-3 years, that has driven the cost of their products to all time highs which also raises the over valuation and expectation on the value of AI.
The money being invested into AI is staggering, and some of that is borrowed ( in some cases $60 billion at a time) while these companies are beating earnings expectations the overall value will just keep climbing to staggering levels, once demand slows down and earnings fall short on expectations then we can expect some sort of AI bubble burst. If its severe enough, then expect a recession in some countries such as Korea and Taiwan, America would also look rather shaky for a while.
I would imagine the hope is that the cost of AI in the long term will come down. Models will become more efficient, new tech will lower power consumption etc. AI companies lose money initially but gain customers, then over the long term as costs come down they'll start to make profit. Or they'll simply wait until such a point where the world depends on AI and start to increase cost.
Steve H said:
But I suspect that as we allow AI to actually start managing parts of our digital lives (email, data, calendar, financial etc) rather than just treating like a slightly smarter version of Google we will also become more willing to see value and pay for the services it offers.
But how many average people need our lives - let alone digital lives - managing beyond the very basics? Until AI is integrated into humanoid robots and can genuinely do things like cooking dinner, cleaning the house and doing the school run, I personally don’t have much call for it. Maybe if you could tell your AI to devise a meal plan for the week within a set budget, automatically order the ingredients from Ocado and then generate step-by-step daily instructions…….maybe then I might be interested, but still not interested enough to pay much for it. Not until it cooks it and cleans up.
And then even if it frees up spare time for people most of them will just waste it scrolling on Facebook or PH.
RichFN2 said:
The whole AI category seems to cover quite a broad range of technology companies. Samsung, Microsoft, Nvidia, Google etc all fall under the AI category, Nvidia alone is worth over 1 trillion dollars which raises valuations and expectations on the general AI sector.
Then you have companies like Micron and SK Hynix that make the semiconductor's, they are earning 2-300 billion dollar's each so far this year, and those earnings have risen by 2-300% this year. They are unable to keep up with demand and are sold out for the next 2-3 years, that has driven the cost of their products to all time highs which also raises the over valuation and expectation on the value of AI.
The money being invested into AI is staggering, and some of that is borrowed ( in some cases $60 billion at a time) while these companies are beating earnings expectations the overall value will just keep climbing to staggering levels, once demand slows down and earnings fall short on expectations then we can expect some sort of AI bubble burst. If its severe enough, then expect a recession in some countries such as Korea and Taiwan, America would also look rather shaky for a while.
small correction - micron and sk hynix aren't 'earning 200-300 billion dollars so far this year' even if their earnings have risen an absolute heap. a quick check suggests micron is in line for about 85b of income for the year. they are forecasting approx 2x revenue growth into the Q1 next year so are probably comfortably expecting to double their income. I believe SK hynix have made more than micron but not in the 200-300b USD order. Then you have companies like Micron and SK Hynix that make the semiconductor's, they are earning 2-300 billion dollar's each so far this year, and those earnings have risen by 2-300% this year. They are unable to keep up with demand and are sold out for the next 2-3 years, that has driven the cost of their products to all time highs which also raises the over valuation and expectation on the value of AI.
The money being invested into AI is staggering, and some of that is borrowed ( in some cases $60 billion at a time) while these companies are beating earnings expectations the overall value will just keep climbing to staggering levels, once demand slows down and earnings fall short on expectations then we can expect some sort of AI bubble burst. If its severe enough, then expect a recession in some countries such as Korea and Taiwan, America would also look rather shaky for a while.
the issue for the 'main' people driving the boom (ie openai, anthropic google etc) i suppose is that over time, cheaper open source models, while last capable than leading edge stuff will prove 'more than good enough' so unless token cost continues to fall (and it already has relatively a lot), whether the frontier models can justify the expense is.... unclear i think. All a moot point obviously when skynet/our AI overlords come into being when AGI takes over the world I suppose......
Crumpet said:
But how many average people need our lives - let alone digital lives - managing beyond the very basics?
.
Digital lives are getting more involved and more integrated..
Most people have their entire financial position online and on apps.
House automation is already big with heating, doorbells, cameras, gates.
Increasingly people get tens or hundreds of emails every day, have thousands of digital pictures, multiple social media accounts, health apps, calendars, alarms, apps for their car, gym, childcare and just about everything else.
The big change will come as instead of doing a specific task or answering a question within it’s own app, we allow AI to manage other parts of our digital lives, following our directions but taking away a lot of the repetition, improving the efficiency, watching out for the security, keeping on top of everything and making sure we don’t forget to do something.
Do we want to be giving AI access to our email accounts, our bank accounts, or our health records? Instinctively many would say absolutely not, just like they wouldn’t have trusted an app for their bank account or to order a repeat prescription on a decade ago.
But things move on fast, it’s not hard to imagine hitting a point where people will see the value in a product that helps them manage their digital complexities.
RichFN2 said:
The money being invested into AI is staggering, and some of that is borrowed ( in some cases $60 billion at a time) while these companies are beating earnings expectations the overall value will just keep climbing to staggering levels, once demand slows down and earnings fall short on expectations then we can expect some sort of AI bubble burst. If its severe enough, then expect a recession in some countries such as Korea and Taiwan, America would also look rather shaky for a while.
Anthropic has 800 billion debt maturing. The worst thing is they rent processing power from their own investors. But not a bubble. lolwhich then asks it they rent processing power where the actual value is coming from?
Flyingakite said:
Anthropic has 800 billion debt maturing. The worst thing is they rent processing power from their own investors. But not a bubble. lol
which then asks it they rent processing power where the actual value is coming from?
Nope, Anthropic has 800billion of projected long term financial commitments in compute/infrastructure over the next decade. they do not have '800b debt maturing' as you are claiming....which then asks it they rent processing power where the actual value is coming from?
isaldiri said:
Nope, Anthropic has 800billion of projected long term financial commitments in compute/infrastructure over the next decade. they do not have '800b debt maturing' as you are claiming....
what do you call it then when its all on the tick?''(non-cancellable)''
The biggest other issue is the power limitations to run all these future data centers everyone needs based on AI being used by everyone on the planet and paying a premium for it.
I find it weird that no one seems to be talking about the negatives on AI.
This business model reminds me of Wework.
Mabbs9 said:
Like a lot of things. Make you reliant on it then charge a lot. Then dilute service and charge more for the full service and repeat. See Amazon, Sky etc.
This is what I think will happen too, like the crack dealers business model, free samples get them hooked then charge for it.Google owner Alphabet suffers its worst-ever day on stock market as value plunges $500bn
https://news.sky.com/story/google-parent-alphabet-...
Flyingakite said:
isaldiri said:
Nope, Anthropic has 800billion of projected long term financial commitments in compute/infrastructure over the next decade. they do not have '800b debt maturing' as you are claiming....
what do you call it then when its all on the tick?''(non-cancellable)''
The biggest other issue is the power limitations to run all these future data centers everyone needs based on AI being used by everyone on the planet and paying a premium for it.
I find it weird that no one seems to be talking about the negatives on AI.
This business model reminds me of Wework.
davek_964 said:
In the long term, I think it will make an awful lot of money - it's a game changer, and will become central to how we use computers / phones / the internet. Businesses will certainly be paying for it, and as already mentioned - even without that, there are ways to make money without subscription models.
I am currently paying for Claude Code at home for private use, although admittedly I have a tech / software background. But it's a relatively small amount of money - basically similar to Netflix - and I currently get real use out of it. I think it's possible that an increasing number of people would pay in the future - perhaps initially growing fairly slowly.
After all, I'm old enough to remember the vast majority of the population being quite convinced that they didn't need to carry a telephone around with them all day!
Question for the thread, not davek_964: Who will provide businesses with revenue for sales when everyone is replaced by AI & automation?I am currently paying for Claude Code at home for private use, although admittedly I have a tech / software background. But it's a relatively small amount of money - basically similar to Netflix - and I currently get real use out of it. I think it's possible that an increasing number of people would pay in the future - perhaps initially growing fairly slowly.
After all, I'm old enough to remember the vast majority of the population being quite convinced that they didn't need to carry a telephone around with them all day!
This isn't me being a luddite, just that I see this argument being rolled out all the time with no recognition that companies will only utilise AI if it is more efficient to do so than to employ humans for the same task. To achieve AI earning an awful lot of money, an awful lot of people in employment will become not in employment and if you think the AI Companies will pay enough tax to pay a universal income at the level of lifestyle we are used to, that will fund the revenue streams the AI user companies & Wall Street are used to then I have a bridge to sell you.
Gassing Station | Finance | Top of Page | What's New | My Stuff