Forgot your password?
Close
typodupeerror
AI Businesses Facebook Google Microsoft Oracle

Microsoft, Google, Amazon, Meta and Oracle Expect a Negative Cash Flow of $125 Billion Next Year (washingtonpost.com) 76

Posted by EditorDavid from the in-the-red dept.
The Washington Post shared surprising news this week about five top AI companies.

Microsoft, Google, Amazon, Meta and Oracle "are spending so much on developing AI and delivering it to customers that they're expected to bleed cash in the coming year, according to a Washington Post analysis of data compiled by S&P Global Market Intelligence." Free cash flow, which measures the cash left over after paying expenses and AI infrastructure costs, is now expected to shrink to almost nothing for the five companies combined in 2026, and decline again to negative $125 billion the following year... AI spending by Amazon and Google pushed the companies to an ignominious milestone: They lost more cash in the past three months than any other large U.S. companies, according to S&P Global data. Investment analysts expect Elon Musk's SpaceX to show even worse cash bleeding this week.
This discussion has been archived. No new comments can be posted.

Microsoft, Google, Amazon, Meta and Oracle Expect a Negative Cash Flow of $125 Billion Next Year More

Microsoft, Google, Amazon, Meta and Oracle Expect a Negative Cash Flow of $125 Billion Next Year

Comments Filter:

  • Woudn't be surprised if it all comes crashing down (Score:5, Insightful)

    by hwstar ( 35834 ) on Saturday August 08, 2026 @01:56PM (#66279166)

    Where is the prudent restraint?

    What we have here are business chasing other business over the same AI stuff. That's a positive feedback loop which at some point is going to become unstable and crash.

    Will these companies survive that? Maybe some, but not all of them.

    Will the investing public get hurt? Very likely.

    • I don't think this will kill Microsoft, Google, Meta, Amazon or Oracle. Their worwr case is having to dump their AI programs, take a stock hit, and issue some bonds. They all have enough revenue for the foreseeable future that they won't bankrupt.

      • Re: Woudn't be surprised if it all comes crashing (Score:5, Interesting)

        by gtall ( 79522 ) on Saturday August 08, 2026 @05:15PM (#66279426)

        Oracle may not, they are way over leveraged. Ellison caught FOMO disease. The Fed. Gov. cutting them a deal with the Department of DEFENSE was throwing them a fish to help them get by a disaster, but it's still peanuts compared to what Ellison sunk into the craze. And their reputed expected winnings are from AI companies, just the companies that will take it in the neck if China AI steals their markets.

        • What would happen if Oracle went down?

          I guess they would be asset stripped and their software treated like VMWare is treated by Broadcom - basically only companies that got incompetently trapped would stay on it and they would have to pay massively more just to get much worse service. Oracle DB itself is going down for sure. Amazon can definitely migrate companies off that without huge problems. Oracle's business software can be replaced by SAS consulting and so on?

          I guess that it's time for people that ear

    • Re:Woudn't be surprised if it all comes crashing d (Score:4, Insightful)

      by Ol Olsoc ( 1175323 ) on Saturday August 08, 2026 @03:32PM (#66279314)

      Where is the prudent restraint?

      What we have here are business chasing other business over the same AI stuff. That's a positive feedback loop which at some point is going to become unstable and crash.

      It's another bubble. The size of this bubble is pretty impressive. When it pops, much shit shall hit the fan.

      I have thought for some years now that AI is inevitable, but just like the dot-com bubble, and the subprime loan bubble, the present paradigm isn't sustainable, and companies are acing toward data centers that require restarting ancient nuc reactors, and making a hella mess out of groundwater.

      One of the telling tremors is Texas, which has so many data centers planned, they cannot provide enough power to them. So they are "temporarily halting them. Only 474 GW, https://hardware.slashdot.org/... [slashdot.org]

      Unsustainable concepts, but we do tend to march like lemmings to the cliffs.

      • Re: (Score:1)

        by Anonymous Coward

        It's another bubble.

        But previous bubbles had some form of value and income, and there was something of value left for others to purchase. This AI bubble is different, and when it pops it's going to be really painful.

        • AI definitely has some value and income.

          Not as much as it's being hyped as, but significant. Anthropic for example will get nearly 50 billion in revenue this year.

          And the value of people using AI smartly is definitely not zero. We're in the late 90s internet part of this. In 10 years there will be AI that we take for granted making some companies billions and a lot of failed companies. There's no reason to think otherwise.

          It's also not going to replace lawyers and doctors, it will be part of how they do the

          • Re: (Score:3)

            by NetCow ( 117556 )
            If you have 50 billion income on ten times as much expense, and expense is growing faster than income, then you have negative value.

            • Re: (Score:2)

              by zlives ( 2009072 )

              no worries, grok/spaceX is now in your 401k so you lose no matter what you do, thanks SEC.

          • Re: (Score:2)

            by zlives ( 2009072 )

            same 50 billion that all these AI companies are reporting, pass the buck baby, i have a thought
            give me that 1 tn investment from last year and i will give you that 50 billion, heck i'll give you 500 billion back. just keep giving me the trillion.

        • I remember a discussion about this when Facebook went to the stock market. Shares sold good, but no one knew how they were going to make money. I also think it is all inflated, but it is not all hot air. I see less and less people using Google and more and more people using chatgpt to search the web.
          I am one of them. Was repairing a water pump and needed an adapter. Local shop was clueless. Described what I needed to gpt. It searched around and found the perfect piece in a webshop. Would have taken me a

        • A lot of companies went bust during the dot com crash. Arenâ(TM)t a number of infrastructure companies handing free tokens to ai companies as investment, then booking use of those tokens as revenue, then going out spending cash or borrowing money to build hundreds of billions in capex?

      • Re:Woudn't be surprised if it all comes crashing d (Score:4, Interesting)

        by molarmass192 ( 608071 ) on Sunday August 09, 2026 @07:20PM (#66281052) Homepage Journal

        They're destined for West Texas, where gas companies will pay you to take the natural gas they have to burn off otherwise. They are installing private natural gas turbines to power these things. Amazon is building a huge one as we speak. What's halted is connections to the grid, the off-grid ones are marching along unabated, and there are a lot of them.

        • They're destined for West Texas, where gas companies will pay you to take the natural gas they have to burn off otherwise. They are installing private natural gas turbines to power these things. Amazon is building a huge one as we speak. What's halted is connections to the grid, the off-grid ones are marching along unabated, and there are a lot of them.

          Fortunately, NatGas and petroleum are in infinite supply, and no need at all to connect to any grid.

    • They're paying each other. Google is renting out processing while also renting processing elsewhere. Selling hardware and then renting the same hardware back for their own use. It may make sense on some level (more than it sounds at first glance), but it's very circular. What does this mean? It means we can't look at a single company's finances. We have to look at the industry as a whole, because a lot of those balance sheet items cancel each other out.

    • âoe Will these companies survive that?â

      Ask your AI!

    • Re: (Score:2)

      by taustin ( 171655 )

      Will the investing public get hurt? Very likely.

      Followed by taxpayers taking in the shorts to pay for the bailouts. And there will be bailouts.

    • Re: (Score:3)

      by gweihir ( 88907 )

      They have done massive circular financing and massive burning of stupid ("investor") money. I would be very surprised if it does not come crashing down at this point. No matter what happens now, they cannot get profitable in the next few years. That is just completely outside of anything realistically possible. And at the same time, the problems with "LLM for everything" become more and more obvious.

    • Re: (Score:2)

      by allo ( 1728082 )

      Which of these four is likely to die? Tell us, so we can laugh.

  • Is this fine, really? (Score:3, Interesting)

    by GeekWithAKnife ( 2717871 ) on Saturday August 08, 2026 @02:09PM (#66279182)
    When most of the jobs they removed due to using AI instead need to be rehired for because unsubsidised AI usages is more expensive I'm sure it'll all work out.

    They will ask later, "but were there any signs of a bubble?" - for those that are busy with stocks that are "going to the moon" and riding the new paradigm I'm sure this is perfectly sustainable.

    I, personally, use AI with great results that deliver actual commercial benefit for me...however there are 1000 people for every one person like me that manage to just spend and achieve mediocre or worse results and I say this to ask one question: If all this AI is really so great, so effective and so needed why then must it be subsidised?

    • Re:Is this fine, really? (Score:5, Funny)

      by reiterate ( 1965732 ) on Saturday August 08, 2026 @02:18PM (#66279196)
      Phew thanks for reassuring us that you're a very special, one in a thousand person, and that you're objectively using this stuff correctly, We needed that reinforcement.

      • That was the best/only joke Slashdot can come up with these years? Sadness.

        I was hoping to see something about AI-powered FinTech counter-wagers to protect each other. Sort of a corporate cancer circle jerk.

        Oh how about a grand scheme to get the money back based on wagers by human suckers as to which AI will win the race? They could let Microsoft hold the pot since they apparently got left out of the race. My money would be on Oracle for most secretive and ruthless and evil and therefore the most likely win

        • Re: (Score:2)

          by shanen ( 462549 )

          My bad. I forgot that Microsoft had been mentioned. More evidence of senility? Or I just can't take Microsoft seriously based on my experiments and experiences with Copilot? So let IBM hold the pot for old times sake?

      • I consider it a public service. You don't have to thank me. I'm special like that. ðY(TM)f

      • I believe their entire point was that it is possible to use AI effectively, not to humble-brag. Why waste time being a hater?

    • Oracle borrowed something like $600 billion to fund AI projects.
      Their biggest customer here is OpenAI to the tune of around $300 billion, a customer that is expected to make a loss of around €14 billion this year.
      Probably fine?

    • Re: (Score:2)

      by gweihir ( 88907 )

      While LLMs have some specific areas where they are useful, there is nothing positive or useful about the insane hype they are used in.

      • You are going to go a long way in your quest to understand and have a clue about AI, because you've got a long way to go.

        • Re: (Score:2)

          by gweihir ( 88907 )

          You continue to be without insight. Nothing new.

          • Don't sell yourself short. You've come from there is no such thing as AI to and it can't do possibly do what it definitely does as the way to "it has some uses" in just a few years. While you are only a few years slow on the uptake I'm thoroughly convinced that I'm just a couple more years you can get yourself to the point where your understanding is only a decade behind the curve.

            • AI gooners are always telling us what it *will* do if we just wait a few more years - AGI is just around the corner! -The plain fact is that LLM inference systems are plateauing, with exponentially more power needed for smaller and smaller gains. The limit they are hitting is real and far below AGI.

              • That is completely untrue. Where are you getting your misinformation?

                • Re: (Score:2)

                  by gweihir ( 88907 )

                  And more lack of insight. Because that "misinformation" is what actually happens at the moment.

                  • I'm still waiting for you to cite an actual source of information. I will no doubt wait in perpetuity.

                    • Re: (Score:2)

                      by gweihir ( 88907 )

                      That is because you do not understand that the one making an extraordinary claim has the burden of proof. And that one is YOU here.

                      But since you have so much faith in LLMs, here is a quote ChatGPT gave me 5 minutes ago for whether LLMs can do AGI:
                      “We don't yet know whether scaling and augmenting LLMs is sufficient for AGI.”

                      Taking into account that OpenAI has a massive, one could say existential, interest in presenting things too positively for this question, I

                    • Remember' Chat-Gpt is just a word predictor! It doesn't think, and it will never be of any use! Only an idiot would be filled into thinking it does anything other than generate words! Thank you so much for trying to make a post to make me look foolish, and in doing so, unwittingly proving that you now admit that you have been clueless in every exchange we have had regarding AI. ROTFLMAO! Off you go now ...
                    • I forgot to mention in my most recent post ... The extraordinary claim is the one made by you, to wit that the exponential improvement of AI models that has been going on for years and still continues has plateud and that you are the only one who knows about it. I'm always amazed at how truly stupid you actually are.

                    • Re: (Score:2)

                      by gweihir ( 88907 )

                      Well, you are established as an idiot and have been so for a long time. Why do I even try?

                    • If there is one thing I love about posts from you, it is that whenever I see a reply from you in my inbox I know I'm about to be entertained by a clown. LOL
                    • I don't think they're the only one who knows. I think pretty much everybody who uses frontier models daily and extensively (which includes me) notices it.

                      The hype was real a couple of years ago. Now, the reactions to every new, supposedly groundbreaking model is split about 50:50 between "maybe slightly better" and "worse than the last one". Sounds like plateauing to me.

                      Most actual recent progress was in harnesses, connecting tools, and stuffing things in loops, not model performance.
                    • So the "hype" was real (news flash ... if it was real it wasn't hype) but now ... Just a few months ago ... this year ... much more capable models didn't get released? That's an absurd claim. It sounds like you simply didn't know what the word capable means, and think it is about improvement in a single domain. Capability is about models used across domains. If a model doesn't improve at all in a single application domain (say code generation) the capability still improves dramatically if it improves in mul

                • Re: Is this fine, really? (Score:4, Interesting)

                  by memory_register ( 6248354 ) on Sunday August 09, 2026 @03:23PM (#66280658)

                  Two great sources to get you started:

                  Michael Burry, the guy from The Big Short who foresaw the housing crisis of 2008. This guy know his numbers: https://michaeljburry.substack... [substack.com]

                  Ed Zitron, investigative journalist. He's been cataloging the AI bubble for a few years now, demonstrating that AI is not profitable and not AGI: https://www.wheresyoured.at/ [wheresyoured.at]

                  These guys are in-depth, detailed, and thorough. You'll have to put a few hours into reading them to really understand all the details, but you can get a pretty good AI summary quickly, ironically enough. They both say essentially the same thing: LLMs might be a useful $1000B SaaS vertical, but they are not a $5 Trillion replacement for human labor - and we're investing capex like they are the latter. This is a bad thing.

                  • So you suggest that I should go to a financial expert for my information to determine if AI is still improving exponentially or it has stopped improving? I'm sure it is a good source of information, for financial analysis, but I think I'll trust AI experts to speak on the veracity of this claim: "The plain fact is that LLM inference systems are plateauing, with exponentially more power needed for smaller and smaller gains. The limit they are hitting is real and far below AGI."The plain fact is that LLM inf

                    • Sorry, I should have claified: Ed Zitron and Michael Burry have both referenced studies by AI researchers showing that we have discovered certain scaling laws. Here's a primer video to get you started: https://www.youtube.com/watch?... [youtube.com]

                      So short version: computer science experts say it, and it fits with the financial problems we are seeing. OpenAI and Anthropic know this is an existential problem yet are silent about it.

                    • More references if you need direct sources. These are on the youtube video but it's good to have them here as well:

                      REFERENCES
                      A Neural Scaling Law from the Dimension of the Data Manifold: https://arxiv.org/pdf/2004.108... [arxiv.org]
                      First 2020 OpenAI Scaling Paper: https://arxiv.org/pdf/2001.083... [arxiv.org]
                      GPT-3 Paper: https://arxiv.org/pdf/2005.141... [arxiv.org]
                      Second 202 OpenAI Scaling Paper: https://arxiv.org/pdf/2010.147... [arxiv.org]
                      Google Deepmind “Chinchilla Scaling” Paper: https://arxiv.org/abs/2203.155... [arxiv.org]
                      Nice summary of Chinchill

                    • I'm going to look more deeply at some of the links you provided, however I can also already tell you that you are confusing scaling (throwing more hardware at a given model) with the exponential improvement of model capabilities with each new model that has been occurring exponentially since before 2025, and that capability increase has accelerated in 2026, not plateud. You are also confusing LLMs with AI. LLMs are just one component of a complete AI stack. Given the same exact hardware the capabilities are

                    • Let's start by having you read about the scaling laws, then get back to me.

                    • Again ... Scaling is adding more hardware resources to a single running instance of a model to gain improvement. It has never been true that you get an exponential improvement that way, and nobody ever claimed it did. The purpose of building data centers isn't scaling. It is to allow more model instances to run concurrently. Why would I start by focusing on the wrong studies?
                    • I just watched the video. It doesn't at all say what you claim it says. It doesn't even remotely claim that AI model capabilities have stopped improving exponentially. I'm not sure where you got that idea. Advancements aren't limited to making bigger LLMS. Imagine if you will that you have a brain that has never been trained ... a baby. Over time it first learns to recognize objects and gains object permanence, then it learns the alphabet, then words, then sentence structure. It eventually gets to the poin

                    • I think you misunderstood the video, which is common since there is a lot of complex math involved. What the video actually says is that we're getting diminishing returns and those returns are diminishing exponentially / returns are improving at lower rates. In addition, we are bound by the math behind LLMs to always have a certain percentage of hallucinations. These aren't abstract problems that you can solve simply by throwing more hardware at it. These are baked-in existential problems related to the nat

                    • I understood the video just fine. You don't understand the difference between performance and capability. You are also confusing training time with quality of training. AI models (models,not LLMs) continue to improve in *capability*, and nobody has suggested they have stopped improving or found any reason to believe that this exponential upward trend in capability has suddenly plateud.

  • "surprising" (Score:3)

    by ardmhacha ( 192482 ) on Saturday August 08, 2026 @05:08PM (#66279422)

    Only to people who have no knowledge of the dot com boom.

    • History teaches us that people do not learn from history and keep making the same mistakes...

    • Re: (Score:3)

      by Torodung ( 31985 )

      There were winners in the dot-com boom/bust. The problem is the wins always rush to the top, and nothing trickles down.

      Some pigs got slaughtered. That's all. Bulls make money. Bears make money. Pigs get slaughtered.

      But more importantly, the new market's signature dish is the distribution of most of the wealth to the few, who then rig up the next boom/bust to "distribute" even more wealth to themselves. This is why this activity is being engaged in so fearlessly. And it is, as real economic value crawls to a

      • Re: (Score:2, Interesting)

        by swillden ( 191260 )

        There were winners in the dot-com boom/bust. The problem is the wins always rush to the top, and nothing trickles down.

        Nah. It's just like every previous innovation-driven boom (railroads, oil, telecoms...) the growth and buildout phase generates incredible new wealth, highly concentrated (Vanderbilt, Stanford, Rockefeller, Carnegie, etc.), . Then there are several decades of adaptation and adoption, during which the benefits accrue to the population broadly. This doesn't decrease the wealth of those who got crazy rich during the boom, but their wealth stops growing rapidly and they just become the new "old money". Then

        • Nah. It's just like every previous innovation-driven boom (railroads, oil, telecoms...) the growth and buildout phase generates incredible new wealth

          Just because there's excitement and (the illusion of) growth and a tremendous build-out phase does NOT mean it will end like the railroads or the oil boom. The railroad boom, for example, was much different. It opened expansion and transportation of goods and people across the country, and, even if the tickets and fares didn't produce what was expected, they still owned all this valuable land. When AI busts, what will they do with all of these worthless data centers? Likely dump and run, leaving the local c

    • Exactly. That's how we know AI is just hype and will not last, the same way that happened with online commerce!

  • Transfer of cash to NVIDIA (Score:3)

    by david.emery ( 127135 ) on Saturday August 08, 2026 @06:20PM (#66279512)

    Horace Dideu characterized this as an unprecedented transfer of cash from Microsoft, Google, Amazon, Meta and Oracle shareholders to NVIDIA shareholders. Those are companies with real, substantial revenues and until now, significant positive cash flow (yielding shareholder value.) Draining the tub this way is something that should worry those shareholders. But of course, it's a great year to be an NVIDIA shareholder, with all those other companies (plus OpenAI, Anthropic, and other smaller AI companies) all pouring their free cash flow AND their growing debt, into NVIDIA, and into other companies that supply AI data centers.

  • That has been clear for a while now. And, unless the AI hype collapses, the next year will not be any better.

  • most of those companies have serious issues zoning data centers and getting power sorted out.

    Right now, we're living in a world where for operations, all the companies need new modern data centers with the power grid and additional infrastructure to support them. By spending on AI, they get the land, the leases, the power, the infrastructure, etc... all because in the name of national security, the geriatrics... I mean congress and the president will just throw money at making it happen. They don't really n

Slashdot Top Deals

Have you ever noticed that the people who are always trying to tell you `there's a time for work and a time for play' never find the time for play?

Close