![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
It's not really that absurd. I work for a very large company that has nothing to do with tech at all, it's a hospitality company. Even then one department might request something where the department that it was requested to will reject the request depending on market demands at the time of the request. It's more common than people realize for very large companies.
I literally just paid off my phone. But the back glass is cracked and not waterproof, the lightning port stopped working a few months ago, and the battery is at 87%
I noticed right away when I went from a phone with 12 GB to one with 8 GB.
What gets me is the slower storage spec's that even some premium models will saddle you with going the smaller route.
No kidding... They're basically full fledged computers now more than phones.
Even as a theory about profit maximization, that only works if the immediate gains from selling RAM to AI suppliers more than fully offsets the losses from selling fewer phones, the losses from Samsung accessories and support sales, losses from phone related software purchases, losses from in-app purchases for which Samsung would get a revenue share, and harder to quantify values such as the perceived and actual decline in Samsung's phone market share along with consumer reputational brand damage.
Yes if it would be profitable to them to outbid the other buyers... they should (all things being equal we can imagine they know more what the price that make sense than internet people).
Yes and no. It doesn't always push the resource towards what we value the most. It pushes resources towards who's paying the most. Right now AI companies are paying a fortune but the general public for the most part certainly isn't lining up to use AI except for memes and silly images. And, if you asked them to pay for it they'd abandon ship like a savage. So, right now, in the case of lots of chips, it is going towards what investors currently value a lot and not what "people" value the most. If asked, people would value their phones more than larger AI data centres.
that why the toward and not perfectly aligned, paying the most is the best proxy we have for value the most (and people discover it themselve only in the actual act of paying, they do not know before)
OpenAI chatgpt chatbot has 800 millions active users (has any product got the general public lined up faster than that in human history ?), waymo rides per weeks are more than doubling yearly pace despite charging more than humans drivers. It has been unlike most new tech one that was particularly directly driven per regular people end users demands (not rich enthusiast, not enterprise, not government, not military), GPT 3 reaction was not particularly driven by some government/big enterprise/investor top-down plan, but a worldwide viral affair of regular end-users.
Not in waymo, tesla FSD case, or claude and other copliots that do charge good money or many case, Google processed its last quarter 1.3 quadrillions tokens per months, that was a 20x increase year over year, Gemini went over 600 millions active users, microsoft/google revenues exploded in part because their internal ai got better and meant better ads revenues but also because people pay them for tokens.
Phone at a good price yes and they will make hundreds of millions of them, new flagship Samsung phone, vast majority would not, people will use chatgpt and not buy those.
That's the thing with anecdotal evidence. I know one person that uses AI and they only use it for making funny pictures of their dog doing weird stuff. They'll get bored of that in short order and move on to something else. I personally don't engage with it at all.
yes but exploding giant token generation and 600m/800m user base is not, is just that anecdotal evidence very much support the statistics.
? revenues they for sure are flying in, that not the question right now (enough versus spending and for how many of the players it will be), take an OpenAI few company in world history (if any) grew revenues that high that fast and they are charging enough money to pay for the current model (would it be used long enough), the making of the next one is where the big lost come from, it is not operating at a lost the current one.
| Microsoft Annual Operating Income (Millions of US $) | |
|---|---|
| 2025 | $128,528 |
| 2024 | $109,433 |
| 2023 | $88,523 |
| 2022 | $83,383 |
| Meta Platforms Annual Net Income (Millions of US $) | |
|---|---|
| 2024 | $62,360 |
| 2023 | $39,098 |
| 2022 | $23,200 |
| 2021 | $39,370 |
| Alphabet Annual Net Income (Millions of US $) | |
|---|---|
| 2024 | $100,118 |
| 2023 | $73,795 |
| 2022 | $59,972 |
| 2021 | $76,033 |
| 2020 | $40,269 |
As of now, microsoft/google are charging significantly more per token to make giant profits and they internal use for it (meta/google) exploded ads revenues enough to pay for it as well.
I am not sure people updated their discourse to microsoft/google/meta last 2 quarter cloud growth, few expected their capex return to be that big that fast and a lot of the pundit talks as not adjusted to those results yet, people could obviously stop buying AI services from those infrastructure player of course at those price and they could get in trouble, but that not the current trend.
Guess how google is making money
Revenue is growing, yes, but they're still spending way more than they are making. The whole AI thing is still purely being built on investor income. You're spoiled for choice in terms of articles asking how much longer investors are going to keep funding AI builds without seeing a return on their investment. So they're operating deep into the red with no short term plan to start operating in the black and starting to earn profit. You quote growth numbers for companies as a whole which is a combination of many different products, not just AI. And in some cases, that AI growth is one AI company buying data centre useage from another. The money just changing hands no different than the hardware circle-jerk we saw happening recently with nVidia and a bunch of companies. Just money changing hands and having it look like everything is ok but some companies are leveraged beyond safe levels trying to make it look like things are fine.
Many of the big one are using cashflow/reserve, google is still returning money to investor not asking them to dilude themselves (11.5 billions in cashback + 2.5B in dividents in Q3, just a bit less than usual to finance that extra capex), same for Microsoft, they are not deluding itself for this, it depends on which investor/company we are talking about here and a lot of articles do not necessarily engage with Google/meta revenue success and talk about something else (say what going on at OpenAI or other players), google tend to be a bit drama free, does not involve an Nvidia to get gamers click and like meta a lot of their revenues explosion from better audience capture and ads targeting is not something they will themselve proudly talk about with the publics.
When it come of revenues gains for a meta or google it become hard to distinguish what non AI revenues are, all ads revenues and the time capture of people to be exposed to ads are AI driven and tend to be a lot of their revenues, for an amazon it get harder and harder as AWS get used for it, their ads business, shipping and more and more putting stuff in box part of the business will be AI driven and the driving part.
That's the point. The big tech companies are growing. Some of that growth is AI, yes. Some of that growth is selling AI server space to companies who're purely fueled on investor money. Every article I read says the same thing ... that AI can't continue at its current growth pace for much longer and that investors are soon going to slow down the money drain. When that happens things are going to get very interesting unless there is a technological break through before then that allows them to process on less hardware for less energy cost. We've got dozens of posts on these forums talking about that very thing.
for the global field (i.e. every articles think there will certainly be loosers, which of course with the large amount of players), does not mean that will happen to Google or Amazon (and it is not investor financing them, but their own cashflow), yes it is possible that many of google/microsoft/amazon clients are not profitable but that can make the conversation strange.
For sure (this is trivial), every 6 months we have those, at least from the observable space of private model (price, token usage, speed) and open model, that in part what made the general public coverage of deepseek a bit strange, it was talked as if it was not something that happen 2-3 time a year.
we had lot of post about how deepseek launch changed everything, how after training hardware will not be needed to just do inference or tons of message about how after 3 years H100 are done being a productive hardware that can generate revenues... we (and comments section of internet in general) have been wrong a lot at every steps, giant amount of post/article are not from people that followed that space that much, you get answer like what AlphaFold ? from people that speak with "authority" on the matter. DLSS/FSR cannot ever be better imagine quality than native level of understanding of what is going on.
But for someone like say a Google, what percentage would not be ? (as ads revenues is AI, search is AI, as capturing people with content recommendation on youtube is AI), that what drive about all their revenues and even more so revenue growth.
Not only that, how many google searches with nobody clicking on or reading the AI summary is counting against their "AI engagement". Easy enough to tweak the numbers when you're baking AI into existing products and then counting it all as a big win even if only some people use it.
Google search result are also heavily relying on AI (and has been for a long time), not just the AI summary, it so central to all their products that trying to isolate AI revenues would be really hard, if possible at all.
![Zarathustra[H]](/data/avatars/m/9/9298.jpg?1723407683){kind=avatar}
You are right, the world would be infinitely better without it, but not nearly as profitable.
