Who are they paying? Or I guess a more accurate question is what are they spending money on?...like advertising campaigns or something?
OT: Nvidia/AI/Tech Exposure
- Thread starter horshack.sixpack
- Start date
Thats not how any of this works. (Though I think 2026 has double the AI costs as 2025). Its ongoing investment needs. Thats the problem. These data centers are obsolete in 5 years. Its ANOTHER $400b in debt for every, what, 2 or 3 years? The $35B a year isn't paying it off, just servicing it. So in a couple years, theres $70B gone to yearly profit. A couple more, $105B gone. See the problem? AI has to be paying for itself within 5 years. Probably 2 to avoid a major correction in markets.
With China in the picture, it seems a poor bet to bet on them not having competition, right?
Thats part of the problem here.
You're just assuming as fact that Google raises their profit by 5x???
True. I just dont think any of that is possible. The main problem, besides that their product doesnt work, that their customers hate it, and it doesnt turn a profit....is it doesnt SCALE. the product, by its nature, will be 10x as expensive to provide than the competition, but not be able to charge 10x more. Sure, you can get people to throw $10 a month for movies when they can get them free from the library.....but for this to work you'd have to get them to be spending $1000 a month. Its moronic. We're talking about people that wont even spend an extra dollar for Doritos.
I think "AI" becomes the new New Coke. Every little mistake from businesses that utilize it will be seen as another AI hallucination. And people will hate it.
But can they all maintain these valuations?
I think they merge into the "winners".
I've had plenty of Sr managers push the newest IT tool. They all sucked.
Ah, a retreat into "we dont really know anything". A classic.
Until it isnt.
Its just math. Let me make a baseball analogy. I am not contesting that the Phillies may win 140 games next year. I highly doubt it based on history, but its at least plausible. I am contesting that everyone in the NL east will win 140 games next year. Because its impossible, because they play each other.
(And yes, it started this as 120, checked it on chatgpt, and had to change to 140. Lol)
Companies generating this kind of profit causes effects that will have major consequences....that will eliminate those profits. They cant do it without government keeping China out. They cant do it without mass layoffs. They cant do it without absorbing all the profit from general business. Thats not just gonna happen and we all accept it!
Apparently, they are targeting Data Centers with misinformation. A lot of info out on the internet. I mean a lot. Where there's smoke there is fire. This is the type of stuff the ChiCom's do.
This just one.
China fueling U.S. data center resistance, AI groups claim
I believe it is happening. China will do anything to get ahead of the United States. They will probably crash their own system without using safe guards to win.
I’d like to know how many bubbles that SPS has predicted in the last 15 years…. Seems once a month someone else is predicting a bubble.
There is most definitely a boom right now that has to cool…. But I don’t think this market is about to crash.
There is most definitely a boom right now that has to cool…. But I don’t think this market is about to crash.
Yep but on the product side - at least for next 3-4 years. People don’t realize what damage has been done to the worlds refiners
At FYE25, Alphabet had an EBITDA of $181 billion and total debt (including lease liabilities) of about $59 billion. So essentially, they could retire every bit of their debt in about 4 months. Amazon had an ebitda of about $165 billion and total debt of $152 billion (including long term leases) they could retire obligations completely in slightly less than a year. This does not take either companies liquidity into account, and it’s massive with both companies.
These companies can handle so much more debt service they are in no real danger here.
You’re reaching a bit on this continuous debt cycle. The data centers aren’t getting totally mothballed after 2-3 years. That’d be a terrible strategy. Some of these things cover 600 acres. See above. They are already incorporating long term, modular strategies into the design framework to allow for refurbishment and restoration without affecting day-to-day operation. Yes, that costs money too, but its a drop in the bucket compared to new construction from scratch.
China is not building any data centers on U.S. soil. The Chinese equivalents of Amazon and Google will be more competitive when they’ve got their infrastructure up and running, but they’ve been trying to undercut Amazon and Google for years while making very little headway in other business (retail, etc.).
Personally, I don’t see any Chinese company being able to compete at all with AWS, GCP, or Azure on US soil, no matter how many data centers they build over there. The regulatory environment will eat them alive even if they are able to build a semi-competitive platform.
It really isn’t. Expecting any of the Mag 7 to “take a 50% hit to profit for decades” is not a reasonable position. It’s not any of their first rodeos in cutting edge tech developments.
No. First off, I was referring to AMZN. Secondly, they are already 75% of the way to $1T in revenue annually. They are 2~2.5 years away from that threshold. $1T in revenue could easily yield $250B in profits. They are already at ~$120B when you exclude the huge front end cut they took from their cash flow to start their AI / cloud computing build-out. $250B profit / $250B debt / $500B operating income is not unreasonable at all for them in 2-3 years time. They wouldn’t be raising profits by 5x, more like 2x ~ 2.5x.
I think you might be taking a somewhat narrow view here. “AI” is the buzzword and it brings about a visceral response from many, but these infrastructure investments are about so much more than AI alone. We’re headed to a world already where actual websites are becoming background noise and obsolete. More and more of the internet is running through AI queries that can pull info together so much faster than 100,000 people typing on 100,000 search engines at the same time. AI is the conduit, but what we’re really talking about is the next generation of the “information superhighway”.
People think of AI and they think of what can ChatGPT / Gemini / Alexa do, what can Anthropic do, what can Claude do, what can Tesla vehicles do, etc. It’s way bigger than all that. How good any of those are individually right now doesn’t really matter at all, they’re all going to be 1000x better 2,3,5 years from now. What Amazon, Google, and Microsoft are doing is not trying to have the fastest, smartest car on the road, they are quite literally taking on full ownership of the roads themselves. They are going to eventually divide ownership of the entire internet in the United States, and elsewhere. That’s how big this really is. THAT’S why these mega investments are a no-brainer for them.
Elon Musk is a giant pretentious douche bag, but he had an interesting perspective he shared on at least one thing, and that same principle applies here. A little while ago, he was quoted as saying that Tesla was “a charging company that also makes cars.” All the while, the carrot of Tesla vehicles had been what was dangled to the public as the profit center. But he literally gave away all the patents for everything. He wanted to be copied, and for that to start a serious conversation about EV’s, so he could start selling the “gas” to everyone. Here, the AI is the carrot, but the “gas” is actually the superhighway - the flow of the information itself, in any form that it takes.
“These valuations?” You tell me. By most educated accounts, the Mag 7 are currently trading at the cheapest valuations that they have in years, when taking account of all their fundamentals. Who do you view as having an outrageously high valuation? AMZN and MSFT specifically have been really subdued vs. their metrics for awhile.
Google merging with Microsoft? Meta folding into Amazon? Wowzers. That’s quite a take.
I don’t work for an IT company, and we rarely use “IT tools” of any sort.
I think that’s where you took it first, which was kind of my point.
It’s just going to be illegal all the sudden? How / why? You think these data centers being thrown up for hundreds of billions of dollars are being built with no foresight at all into the legal environment?
We’re saying the same thing. I’m not saying everyone shoots to some stupid parabolic earnings level. But take a step back and think - what happens if Google owns 30% of the US internet traffic, Amazon owns 25%, Microsoft owns 25%, Meta owns 10%, Tesla 3%, Apple 2%, and everyone else owns 5%. When I say they own it, they have complete control of ISP’s accessing on one end AND content coming in from the other end. They want to block out China, they can. They want to direct whatever AI driven ad targets toward individual users, they can. Unlimited access to user data? That’s possible, too. VIt’s all on the table. You’re telling me those individual companies wouldn’t all be doing 20%+ YoY profit increases for another long while?
That’s been said about all of them for 20+ years. Hasn’t slowed them down.
They can, actually. They can block China ISP’s and content quite easily. It’s actually the reverse problem, they only CAN’T do it if the US government lets China in, which sounds like a pretty toxic approach for either party to even attempt right about now.
We’ll see. Investors rarely care about that, though.
It’s already happening. The “general business” has changed for everyone. AWS and Azure are already the profit leader segments for Amazon and Microsoft. They’ve both moved wayyyyy past just having Windows / Office 365 and Prime Day. GCP for Google is a little behind both of them, but they are catching up quickly.
“We” don’t all have to accept it for it to happen.
Jet gas and diesel prices will send us into the next big downturn ….2008 style
Bring on the robot women! Imagine the sandwich skills of an AI woman, they can wash dishes and clothes timely, mix you a drink accurately, never have a headache, and do full bedroom Olympics, no constant spending and beeactching! Throw a new wig on them and it's some strange every day. That's the AI I want to invest in
I'm glad you're excited for it, think I'm gonna stick with the original for now.
Also, I think long-term the outlook for these stocks is good. But there could be some real ups and downs along the way.
misinformation or not, I still don’t want one in my backyard. And I want to have assurances that they will pay 100% of the cost to generate & supply the electricity they use, and I’m not convinced that’s the case right now.
Gonna pick and choose some replies here, to cut down on the tediousness. If you want a reply to something in lsrticular i skipped over, just ask, im not avoiding anything.
If they cut the cord on this stuff today, I agree they wont be affected by much. Im assuming they actually continue this path for 2 to 3 years, then have nothing to show for the expenses.
You're picturing this going like this:
But the reality is this:
Thats why theyre building more and more data centers. It takes more and more computing to see minimal gains. And thats not going to change. It doesnt scale.
i didnt say they were. I said theyd be mostly obsolete in 5 years. Which is true.
you dont say
Given that the overwhelming expense is the chips, the things theyre putting in modular racks to repeatedly replace, its hardly a drop in the bucket. Hard to take you seriously if you're going to make this argument.
False,on all 3 counts.
They already are!
You're right, they've wasted tons on useless products plenty of times before.
If they cut the cord on this stuff today, I agree they wont be affected by much. Im assuming they actually continue this path for 2 to 3 years, then have nothing to show for the expenses.
I think you are mismashing revenue and profits.
Or just a few billion in profits. You dont appear to have accepted that "AI" isnt a traditional tech product. It doesnt scale to increasing profitability. It doesnt scale.
Again, you seem to think theyll do a quick (but large) build out and then be done with the massive expenses.
Fair point. Presumably the data centers will have utility for their other products. But I doubt theres much ROI on that.
Thats the thing. Theyre NOT going to be 1000x times betterm. For the same reasons that self-driving cars arent 1000x better than 5 years ago.
You're picturing this going like this:
But the reality is this:
Thats why theyre building more and more data centers. It takes more and more computing to see minimal gains. And thats not going to change. It doesnt scale.
Tech valuations are historically low? Thats quite a take.
Absent a sea change in anti-trust enforcement, its inevitable, if what you say is true.
I dont work for IT either. I've had multiple engineering roles across large US companies. The ones that your view relies on implementing AI. They are REALLY ****** at IT.
The things you are saying will happen, will not happen in a vacuum. Every action has an equal and opposite reaction.
Aren't you?
Im Not following how you got here.
Fair, but youre not saying we'll see a repeat of the last 20 years.
So, you think they are going to get the govt to block US companies from using cheaper AI tools from competitors, so that a handful of US companies can continue making redonk profits while driving Depression level unemployment? Do i have that right?
Its not about whether they care. Its about the math.
True, but not my point. Democracy as we know it would have to cease to be. Which...may be the plan.
The majority of the infrastructure will not be obsolete. The cooling, power supply, the building itself of course, the racks, etc. are all being designed for longevity.
The chips / boards will be updated periodically and continuously, based on need.
To go down your rabbit hole, and just say 100% of processing hardware has to be replaced every 5 years, that’s not going to be the same cost as a whole new data center. Not even close.
This is an overwhelming expense now, because there is a global shortage. It’s not going to remain that way for the next 5 years as companies start emerging to fill the gap. Supply and demand and what not. If the HBM chip prices today were guaranteed to be locked in forever, a stock like Micron would be worth about $10,000 per share, as opposed to the current ~$950.
Please provide a link of a Chinese company building a data center on US soil.
I’m not. I’m clearly talking about revenue and simply scaling the profits accordingly.
It doesn’t have to scale, it just has to become the predominant conduit for online activity in the US. That’s my point. As of the end of net neutrality and advancements in AI happening simultaneously in 2024-2025, the internet essentially went up for sale in the US to the highest bidders. They are now bidding to get the biggest pieces of it.
I don’t think that expenses go away completely. Obviously there will be regular maintenance costs. I think we disagree on how big those expenses are going to be after the front end investment. That’s fine.
I should have perhaps rephrased to say “more capable”. As in more broadly focused on a larger variety of tasks. More seamlessly integrated into everything we do online. I’m not talking about an image generator getting better at not misspelling words. That’s missing the forest for the trees.
The 'Magnificent 7' stocks are trading at their cheapest valuation in more than a decade
It's a rarity to see the “Magnificent Seven” tech stocks this cheap.
finance.yahoo.com
I’m not following this one.
These political reactions are unknowable. Far too many variables. There’s always a lot that COULD go sideways with unforseen government interventions, foreign conflicts, global pandemics, etc. Or, conversely, things that could go really right in other areas. Can’t assign a valuation with these things in mind. You can only look at the numbers and near term projections.
But in general, you can expect the US government to heavily protect huge US businesses that are investing trillions of dollars on US soil, and creating US jobs.
No, I’m not. I’m simply saying these companies will likely continue their current earnings and profit trajectory of the past 20-30 years. I don’t think that’s a wild take. But it’s obviously very different from yours - that some of these companies may not even survive. That’s quite the hill to die on.
Again, that’s almost exactly what I’m saying. Stock price / profit / EPS growth in the neighborhood of 12~25% per year, which is more or less what the Mag 7 has done over the long haul…..better than almost anyone.
The thing is, they’ve all gotten so damn big that the profit growth trajectory demands that they make these gigantic investments and transcendent business shifts to keep it going. That’s the simplest math there is.
Again, it’s so much bigger than the AI tools. Those will be a dime a dozen before too long. You already have things happening like AI tools writing the software for other AI tools. It’s the fuel running them that matters. Whoever controls that infrastructure will have the ability to influence end consumers to an extent that we’ve never seen before.
To your example, China has an AI tool that is cheap for the license, they will still potentially have to lease space from US data centers controlled by US private enterprise in order to run it. That is IF the Mag 7 choose that particular route to monetize it. They may also just restrict it entirely.
AI will increasingly be a good tool, largely via purpose built agents. So, effectively, it will usher in the next round of automation and efficiency for processes that humans determine need such efficiencies. Like all markets, competition will drive the margins and price down to a "normal" business model that considers cost, price and profitability. There will be winners and losers and I expect a lot of consolidation. I expect, given a liberal tech lifecycle of 5-7 years, the winning data centers will be on a continuous tech refresh cycle that keeps chip makers in business and flatten the growth curve substantially. A number of them will just close because the business justification for refreshing the tech and keeping the doors open just won't be there.
I used Co-Pilot AI to demonstrate why AI isn't I and won't be...
| Category | Human Brain | Current AI Models (GPT-5-era systems) |
|---|---|---|
| Energy Use | ~20 watts | Thousands to millions of watts for training |
| Learning Efficiency | Learns from few examples | Often requires massive datasets |
| Memory | Associative, lifelong, contextual | Large but limited context and stored weights |
| Reasoning | General-purpose, adaptable | Strong but specialized statistical reasoning |
| Creativity | Original, experience-based | Recombinational and pattern-based |
| Sensory Integration | Natural multimodal perception | Artificially integrated modalities |
| Self-Awareness | Subjective consciousness (as far as known) | No evidence of consciousness |
| Speed | Slow neurons (~1-200 Hz) | Extremely fast digital operations |
| Reliability | Can be emotional and biased | Can hallucinate and lack common sense |
| General Intelligence | Broad, flexible | Narrow-to-broad, but not truly general |
I had to trim your quotes to stay under the char limit
(Caveat, I have seen some job recs that included some details that raised my eyebrow, as it was much more longevity focused than I ever see, just on the material expertise being sought. But then the recs got cancelled, so who knows.)
Example, you cant say a homeowner will find a loophole and never owe property taxes, and then will tell others and they wont owe either, for forever. By the nature of what's happening, the law will change. The bigger the change, the more likely the reaction that mitigates it.
I think the big pre-existing players will always be around in some form or another. IBM is still kicking around. Its the OpenAIs and Anthropics, the startups with nothing to them but debt and AI hopes and dreams, that will either be absorbed or go bankrupt.
Are they? You're getting towards my area of expertise. I dont know how these data centers are being built, but in general such things are not built for longevity. Even when they try, they try to do so cheaply, and might as well not have bothered. You have to design such things in from the beginning, and US companies....just dont.
(Caveat, I have seen some job recs that included some details that raised my eyebrow, as it was much more longevity focused than I ever see, just on the material expertise being sought. But then the recs got cancelled, so who knows.)
I disagree, based on the chip costs. I rack of chips literally costs like $5M. Ok, so the rack is now modular and they save $10k on the future switch out. Whoopidy doo.
Fair point. But I dont see the buildouts slowing enough to significantly change prices. Plus, the drag on chipmakers and silicon and such is the risk that AI collapses and takes chip demand with it. In such a scenario, Micron and such stocks crash. This has to be priced in.
chinese data centers in the US Chinese data centers in the US fall into two main categories: cloud infrastructure hubs (Alibaba, Tencent) and telecom PoPs or Points of Presence. These facilities are operated by Chinese tech conglomerates and state-owned telecom carriers.Key Chinese provider data centers in the US include:Alibaba Cloud: Operates two US regions, each with multiple availability zones:US East (Virginia)US West (Silicon Valley, California)Tencent Cloud: Operates multiple US cloud infrastructure locations:Silicon Valley, CA (hosts two availability zones)Ashburn, VAChina Telecom / China Mobile / China Unicom: Maintain extensive network Points of Presence (PoPs) and data center infrastructure in hubs like New York (60 Hudson Street), Los Angeles (818 W 7th St), San Jose (1735 Lundy Ave), and Ashburn, VA.Regulatory and Security Status:These Chinese-owned data center operations face significant regulatory shifts in the US. Citing national security concerns, the Federal Communications Commission (FCC) voted unanimously in spring 2026 to advance a rule that would bar state-owned Chinese telecoms (including China Telecom and China Mobile) from operating data centers and network gateways in the United States.
So you are assuming no diminishing returns?
So basically the new Google, and it wont matter that options like DuckDuckGo exist. Cause people are dumb. I get it, and agree....to a point. Like I said earlier, its one thing for people to shell out $10 a month to a streamer when movies are free at the library, its another when the cost is $10k. I believe these valuations assume people will pay the $10k.
Agreed.
I'll believe it when spell check works.
Relative to the S&P500. This is why I have money in XMAG. But its not on point.The 'Magnificent 7' stocks are trading at their cheapest valuation in more than a decade
It's a rarity to see the “Magnificent Seven” tech stocks this cheap.
Anti trust enforcement SHOULD be used to prevent such mergers, and theres been movement that way, but it likely wont happen. As such, theres too much profit in being one company that "owns" 70% of the internet than two companies that own around 35% each. Mergers will happen. Its just economics.
Thats like saying I dont know when the market goes up or down, therefore it will never go up or down. No. If what you are predicting happens, it demands a reaction. We cant predict when or how that reaction will occur, but we know from history and basic human nature that it will happen.
Example, you cant say a homeowner will find a loophole and never owe property taxes, and then will tell others and they wont owe either, for forever. By the nature of what's happening, the law will change. The bigger the change, the more likely the reaction that mitigates it.
True. But if those businesses are driving depression level unemployment, all bets are off. *Ma Bell enters the chat.*. There's already a large push that Amazon is anti-competitive and should be broken up. In past ages, that would have already happened. If what you are predicting comes to pass, most likely we see anti-trust actions that would make FDR blush.
Its a pretty wild take, as those 20-30 years are unprecedented, and would result in them being like 80% of GDP.
I think the big pre-existing players will always be around in some form or another. IBM is still kicking around. Its the OpenAIs and Anthropics, the startups with nothing to them but debt and AI hopes and dreams, that will either be absorbed or go bankrupt.
Until they are what % of GDP? What is the rest of the economy doing during this? Growing the same, or giving up all their profits to AI? Whats employment doing? Pay?
Thats why its quite a take to think their growth will go on at the same rate after the expansion period they've already gone through.
Part of my argument is how do they monetize a product that is expensive to provide but has easy cheap competition? Branding is part of the answer, and i think a main part of why they are building so fast. First adopter advantage. People still pay for coke when water is right there. People google when its no longer the best search engine. But I think there's a limit to where branding can take you.
This is good, but part of what it misses is that AI is LLMs, meaning its based on word choice prediction and probabilistics. Its NOT math based, and is actually pretty horrible at doing math. Math is the language of logic. All past software tech was based on programming logic. It was deterministic. A program would have the same result every time. You could accurately predict the outcome based on the inputs and the program. But LLMs dont operate that way. Thats why they can never be accurate enough for most things. Big tech hopes if they just dump more processing power and more training into it, that it will take that next step. But it never will. It has its uses, but will always be too limited to be transformational.
Hence my belief that task specific agents trained on closed data sets, and bound by tested prompts will be the next meaningful thing that we get out of it. That means that it is far from being a point and shoot tool. Example: I have a large national franchisee who has meetings with franchises often to discuss financial performance. Obviously the franchise has many financial reports but often other questions arise that they do not have a report ready to support. They don't want to build and deliver canned reports for all franchises, but they do want an agent that trains on the most common questions so that they can answer them in real-time, during meetings. They will continue to build out the agent capabilities. In most cases, these questions can be gotten from the data, but the accounting system doesn't directly support the report that is needed.
I think there’s way too much at stake here for them to miscalculate the budget and planned useful life by any appreciable amount. These data centers all have redundancies on top of redundancies. Backup power, backup cooling, back up processing, etc.
The fact that the companies building these things are typically not construction / engineering experts by trade is, in my opinion, a good thing because they are outsourcing that expertise to those who are, and not doing the typical corner cutting that you mention.
You’re not wrong though about what most US companies typically do in greenfield construction. I’ve seen that myself, too. But the Mag 7 are not most US companies.
$5M is not the cost in perpetuity. I already addressed that.
May not happen for another year or two….but they will come down.
They have no actual data centers here. They lease the cloud server space from existing data centers that are owned and operated by US enterprise. That faucet can be cut off at any time, and the price can go up at any time.
It was a thought experiment, not necessarily a prediction. But $1T revenue converted to $250B profit / $500B operating / $250B debt….that’s 25% ratio of profit to revenue.
Currently, AMZN is $717B revenue, $120B net profit (if you exclude the heavy cash dump into AI front end infrastructure….$90B otherwise), $380B gross profit. Assuming they just grow the revenues only without growing business efficiency at all, they’d be at around $170B in profit at $1T revenues, in early 2029. If efficiency improves, debt scaled back / paid down, getting to $250B isn’t out of the question. I’d consider it somewhere in the middle of a 50% percentile outcome, and best case scenario. Personally, I’d lean more towards $200B in profit by early 2029, and I think any AMZN investors would take that right now.
Well, think of it more like AMZN throttling Google ISP traffic through their data centers, and vice versa, and/or charging the other companies a premium to not throttle, AMZN entering the “search engine space” and packaging everything together in a bundle that competes with Google / YouTube / Gemini. Maybe they both even get into the mobile carrier domain and leverage the data center computing for that.
What other reference point could you possibly use? The Russell 2000? A crypto ETF?
I don’t know that they will….not between multiple Mag 7’s, anyway.
This requires too many specific underlying assumptions to really dive into any further. I’m just gonna say, “OK”.
How are they going to drive depression levels of unemployment? Why would they? That’d kill the golden goose.
AMZN would likely just sell off the retail or shipping segments to somebody else, and focus more on everything else, and their return %’s would probably get even better. But I don’t think the argument that they are anti-competitive really holds water because they have numerous competitors in every segment of their operation.
I do think there is an argument to be made that maybe they have gotten too big for their own good, they are trying to do too much, and that is hurting their ability to efficiently turn profits. And I think their current return rate and valuation possibly reflects that.
Well, they are currently at 60% of GDP right now, after representing only 10% of GDP as recently 2014. Is another 20% over the next 4-5 years really an unlikely situation?
I do agree with most all of this.
Who knows? Again, they are 60% right now.
Probably still growing, albeit more slowly than tech (as has been the case for awhile).
I don’t think the rest of the economy is going to be footing nearly the “AI bill” as the Mag 7 are. But again, this is bigger than AI. They are buying up the internet bandwidth, with the assumption that the internet will be 95% AI driven within 5-10 years.
I mean, something has to give. All of their stock prices assume at least 60-70% of their recent growth rates will remain intact. We’ll see.
Again, the product they’re monetizing isn’t their own unique AI platform. It’s the engine that will power ALL AI platforms. That’s what you keep missing.
Maybe. Its the architectural equivalent of re-inventing the wheel. I highly doubt it.
Even if costs drop 80%, thats still $1M a rack, and the modularity savings cant be more than around 1% of that. Its basically a warehouse that holds gold bars. The shelves ain't the cost driver my dude.
In what scenario where AI is transformational? You think compute needs will massively go down?
A lease is a contract. So sure, if the law doesnt matter. Often it doesnt anymore. But its the same if they owned it. Still doesnt account for China offering goods and services cheaper than US providers. Protectionism is hard these days. And if you're using it to prop up massive Mag 7 profits which drive unemployment? Thats pitchforks territory. Good luck with that.
This sure sounds like assuming no diminishing returns. Generally, when a good costs more per unit to make, and the next buyer is less willing to pay than the last.....you dont keep making the same profit the more you sell.
So a monopoly on the internet, in other words? Where were you on the net neutrality debate btw?
The dollar
Monopoly power tends to work better when you have an actual monopoly.
Almost getting it!
The profits you envision are impossible without mass layoffs. Its the savings from layoffs that pay for the tokens that pay for the AI.
Otherwise, for profits from monopolizing the internet, these are profits without creating value. Rents. Meaning it decreases profit and wealth elsewhere.
Thats missing the point that Amazon drives profits via uncompetitive behavior.
Mag 7 profits are current 1.5% of GDP.
So, rents.
Yep.
It doesnt make much sense.