Apple, Alphabet, Microsoft, Oracle, Meta, and Nvidia spent $1.1 trillion on share buybacks in 5 years to pump up their shares. That’s at risk.
By Wolf Richter for WOLF STREET.
One of the big drivers of the surge of stock prices over the past many years has been the prodigious amount of corporate cash spent on share buybacks by Big Tech and other companies. Some of the share buybacks were funded from cash flow, some with borrowed money. Share buybacks are not like stock trading; they represent new money entering the stock market to remove shares from the stock market. Their purpose is to drive up the price of the remaining shares.
For example, from Q3 2020 through Q3 2025, six companies – Apple, Alphabet, Microsoft, Oracle, Meta, and Nvidia – spent $1.1 trillion on share buybacks (data via YCharts). These are amounts actually spent on share buybacks, not the announcements of future share buyback plans.
I’m leaving out Amazon for a reason: It already stopped share buybacks in 2022 to spend that cash on capital expenditures, such as AI infrastructure. And others may have to follow. More in a moment.
Apple topped the list with $437 billion in share buybacks over those five years, followed by Alphabet ($281 billion), Meta ($151 billion), Microsoft ($107 billion), and Nvidia ($87 billion). But Nvidia’s buyback program was ramped up in 2024; and over the past four quarters, Nvidia spent $43 billion on share buybacks.
Some of these share buybacks were funded with borrowed money by issuing bonds. And it shows on their balance sheets: Apple now has $112 billion in short-term and long-term debt, Microsoft $120 billion, Meta $50 billion, Alphabet $30 billion.
These companies are now engaged in an AI-spending war, on the premise that whoever spends the most wins?
The amounts of spending on capital expenditures, such as for data centers, are mindboggling. In the third quarter alone, just four companies – Microsoft, Amazon, Alphabet, and Meta – spent $114 billion; and they’re on track to exceed $400 billion in capital expenditures this year. And they all said they’d accelerate their spending further in 2026. And they’re not the only ones doing it.
Where does all this money come from?
Some of it comes from operating cash flow, some of it from short-term investments that they accumulated over the years, and a lot of it comes from massive amounts of new debt issuance.
Over the past three months, five companies alone – Alphabet, Amazon, Microsoft, Meta, and Oracle – have issued $88 billion in new investment-grade bonds, according to Dealogic: $18 billion in September, $30 billion in October, $40 billion in November.
The investment-grade bond market may have to absorb $1.5 trillion in AI data center bonds over the next five years, according to J.P. Morgan analysts, cited by Reuters.
Meta has come up with a way to keep a recent $27 billion AI bond sale off its own balance sheet in order to avoid having its ‘AA-‘ credit rating dented. This deal involves Meta and Blue Owl Capital. The issuing entity was a joint venture between Meta and Blue Owl Capital, which sold most of the bonds to Pimco. Meta is going to lease the computing power, but in four-year terms, to be renewed every four years, to avoid “finance lease” accounting, which would have put the liability on Meta’s balance sheet.
Ratings agencies are getting nervous about this type of off-balance-sheet debt. Fitch said today in a report on the circular hocus-pocus deals, as I’ve come to call them:
“Opacity is a rising risk factor, with limited public disclosures from AI model developers and increased use of off-balance-sheet debt from some public issuers [such as Meta]. Uneven disclosure on customers and contract terms obscures the extent of counterparty exposure.”
What’s left for the share buybacks…
But these are precisely the companies that have spent tens of billions of dollars every year to buy back their own shares – some of it with borrowed money.
Amazon has already spoken on this. Despite the super-hyped announcement in March 2022 that its board had authorized the company to buy back $10 billion in shares, the last share buybacks occurred in Q2 2022, and most of the authorization remains unused. Amazon has said that it has better things to do with the cash than blow it on share buybacks – it didn’t quite phrase it that way, it couched it in delicate corporate speak, but you get the idea.
Others will likely follow, maybe not this quarter or next quarter, but sooner or later, as the mindboggling spending amounts and the debt pile-up – even if it is off-balance-sheet – will make blowing huge amounts of cash on share buybacks an increasingly at-risk proposition.
Nvidia is now tangled up in massive circular deals with its own customers, such as OpenAI ($100 billion), and other companies. Those were just announcements at this point, but eventually the cash will start circulating. What will be left for share buybacks then?
There still will be the much-hyped announcements of share buyback “authorizations,” but actual share buybacks might get quietly reduced or put on ice entirely, as every cent that isn’t nailed down gets plowed into the AI-spending war.
Share buybacks incinerate corporate cash for the purpose of driving up share prices. Plowing that cash instead into the AI-spending war, in the worst-case scenario, would be no worse than burning it on share buybacks, but it would actually accomplish something real: Building AI data centers. Not everyone might be happy though: Stock prices might suffer when that prop gets pulled out from under them.
Enjoy reading WOLF STREET and want to support it? You can donate. I appreciate it immensely. Click on the mug to find out how:
The Verge recently had an interesting article on circular hocus-pocus, “I looked into CoreWeave and the abyss gazed back”
That was quite an eye opening article. Thanks.
“Some of it comes from operating cash flow, some of it from short-term investments that they accumulated over the years, and a lot of it comes from massive amounts of new debt issuance.”
And likley also from Uncle Sammy.
What’s the article title? I’d love to check it out
“I looked into CoreWeave and the abyss gazed back” is the title
I quoted the title. Comes up as result no. 4 when I google. Wolf will probably be adding this company to his pantheon of imploded stocks within a not too distant future.
Wolf, for the large tech companies, where a significant amount of their compensation is via equity grants, isn’t ‘equity dilution prevention’ one of the main reasons they do buybacks? Specifically, buybacks counteract the continuous issuance of new shares to employees.
“… isn’t equity dilution prevention’ one of the main reasons they do buybacks?”
No. What does equity dilution lead to? Relatively lower share prices (because each share represents a smaller slice of the company). To pump up their share prices DESPITE equity dilution, companies buy back their own shares, usually in larger number than new issuance from stock compensation plans.
Of the tech companies mentioned here with share buybacks, Oracle was the only one whose share count declined over the past few years (less share buybacks than stock compensation). The share counts of the others dropped as share buybacks exceeded share issuance due to due to stock compensation.
“Meta is going to lease the computing power, but in four-year terms”
In a recent interview the OpenAI CFO said that a new data center cost $45-50 billion. 15 Billion for the plant and 30 billion for “compute”. Compute being rows and rows of servers packed with NVIDIA Hyperscaler processors.
Meanwhile NVIDIA tells us that the current generation of processor package has a life of about 3-4 years – after that it is either obsolete or broken due to the temps the SC are necessarily exposed to (I’ve never been clear on what the cause of End-of-life is)
If this is remotely true, it explains why Meta would want to limit it’s lease to 4 years (besides the accounting benefit). Question is what happens after EOL ? Has the DC made enough money from it’s lease to invest in refitting it’s compute with a new generation of processors ? Has the price of compute declined really fast so that the next generation only costs 10 billion rather than 30 billion ?
Maybe the CFO gal was just being a bit undisciplined with her interview. It is after all, in that same interview that she suggested that the Federal government should backstop Open AI’s trillion dollars in buildout commitments….
Can it possibly be true that Open AI expects 2025 sales to be $13 billion while the company expects to lose $9 billion ???? Who is buying these bonds – PIMCO you say ???
“If this is remotely true, it explains why Meta would want to limit it’s lease to 4 years (besides the accounting benefit).”
The renewal is not optional. It’s required. For 20 years. Meta cannot walk away. That was a requirement to sell the bonds. Which is why this deal is so iffy from an accounting perspective.
So they’re considering a 4-year lease that has a mandatory renewal for 4 additional terms to be different under GAAP than a 20-year term? LOL!
I wonder who their auditors are who signed off on that.
TSonder305,
A June 28, 2022 headline in the Wall Street Journal said, “Ernst & Young Fined $100 million in ethics exam cheating probe.”
The article also noted that KPMG was fined $50 million in 2019 for ethical violations that included claims that some auditors cheated on training exams.
So to answer your question, these auditors. No doubt PWC and Deloitte have done things equally slimy, in case it was one of them.
I saw an interview recently where some billionaire asked OpenAI CEO something like, ‘How can a company with $13 billion in sales have $1.4 trillion in spending commitments?’ To which Sam Altman-Fried’s non-answer was, ‘If you don’t want your shares, I will find a buyer for you.”
That $1.4 trilly is where a lot of this hocus-pocus citcular money is.. err.. isn’t
re: lifespan for AI-chips.
It depends. The current AI hype was started due to breakthroughs in the theoretical models that are behind it and following that, once the bandwagon got rolling, the development of chips which are optimized for those calculations.
Each newer generation was able to do those “AI-calculations” more efficient, faster, with less power required. And power is a major cost.
It might be that in 4 years newer chips are so competitive that a server farm running older chips can’t rent them out for a price that covers their operating costs, mainly electricity. That is already today a major challenge for some of those operators. They are sitting on stranded assets.
Normal servers do not depreciate that quickly and it could be that the rate of advancement slows down and they are good for longer than 4 years.
But the bigger problem is that there are all these hughe announcements for hughe server farms, there might be (probably will be) a glut of AI compute available and those hyperscalers will have trouble to sell all those computing capacity.
Deepseek showed that it is possible to develop such a model much cheaper. Most of the advancement is beeing done by Physicits and mathematicians in the theoretical field. But all the money is beeing spend in training those models (which costs enourmous amounts of money) and giving access to it to the public or companies below cost. But they are hoping to make money on future breakthroughs but those aren’t made from training those models.
The training is the last step, after someone advanced the unterlying theoretical work. Retraining on ever larger databases are “just” tweaking the model. It might improve its usability but they basically know already the limitations.
The traditional business model for new technology, whether it’s chips, chemicals. plastics, TV sets, etc, usually has a lot of new entries and at some point, there is too much available capacity. Smaller players will fall out and prices will become very competitive. Happens all the time and this won’t be an exception.
True. I wonder what smaller player falls out and stops spending: Google, MSFT, Meta, Oracle, Tesla, Amazon, OpenAI.
Thay are all pretty big. LOL
Now it seems that some of these big companies own their own datacenters and are signing big lease deals with small datacenter companies that will probably fail. I guess you can get out of a lease easily after the datacenter company fails and the leasing instead of building removes some risks?
I am reminded of animals with huge exotic antlers evolved for dueling. I also think of a “winner’s curse” where one overspends, to best the other player. Competition sometimes does interesting things with resource allocation. As equity holders, we are somewhere in the back of the line of capital structure, after bonuses and (now-growing) debts are paid. Suddenly today my GOOG stock has me thinking I’m an investing genius, but tomorrow, who knows?
S&P 500 is so concentrated, there are tens of trillions of dollars standing on a needle. Google is fighting the last war and is in precarious exponential ascend. Wealth divide is ocean wide. 2026 is going to be for the history books.
I think Google is wasting their money like the rest of the LLM bubble, but I bet they’ll be one of the few companies still serving chatbots at the end of this. Anthropic and OpenAI have no path to profitability. At least Google has the rest of its business to keep it alive.
Google has to find a new business model. AI is already killing their traditional advertiser based Search and they don’t have an AWS side business to fall back on.
Though currently, at least acording to google, they make money with their AI.
They basically steal the content google search links to. And serving you a AI-slop version of it. When users are not going to the linked website but stay on google, all the revenue is potentially googles as they don’t have to split your attention (and potential ad revenue) with a third party.
The major danger to that is that it threatens the business model of a lot of websites which might just go away for ever. And google might just kill itself doing that.
They do, it is GCP, and although half the size of AWS (by revenue), it is growing much faster. That being said, I agree about the advertising bit … I am reluctant to see how they start baking ads into whatever LLM outputs they serve
That’s wrong on both accounts.
Search is up YoY and google cloud is absolutely a thing.
I’m expecting the day soon when during a Google AI question I ask it that a talking person pops out and tries to sell me a new set of tires.
Andy, maybe check out an equal weight SP500 fund.
Yes, I will look into this, thanks. Google AI now saves me countless hours of sifting through Google-sponsored links, clicking on sites, and seeing annoying Google ads. It’s the best, really.
The Gemini logo should be a snake swallowing its tail.
andy,
True but it also is content stealing and putting smaller sites out of business. Even Amazon is worried the impact AI clients will have on their business model. As content disappears the LLMs will train on fewer perspectives and some will be completely lost.
Equal weight S&P is actually more volatile than cap weight. If you want to diversify away from tech megacaps, consider a total market index or international like EAFE. I know it’s appealing, but look at the track record – equal weight didn’t do any better than S&P in the housing bust.
Or you might look for a fund tracking the MSCI high dividend index. By definition the Magnificent 7 are not included. I take some solace in the fact that blue chip dividends fell by only 11% during the 1929 Depression and by only 6% during the 2008 GFC.
https://www.msci.com/documents/10199/fee511ac-bb6e-48df-bd21-fa399b9bd1eb
Thanks. I diversified into various Mag 7 puts. I’m so short the gentlemen in Chinatown look down on me.
We seen it before with the video streaming companies. No commercials. But eventually commercials or you pay to remove commercials. Newpapers used to be free online and relied on ads but that went away and now most are subscriptions.
I am guessing to be profitable, these AI search agents will eventually require a monthly subscription fee or they instead will have pop commercials. I am already seeing polls asking what you would pay per month to get a feel of what they can charge.
AI = Atrocious Idiocy
As my Boss told me a few weeks ago, “Like it or not, this is the way things are headed, so it would be a good idea to start looking into how to use the tools”.
As a scientist I just used ChatGPT to automate data processing for a new research project. The data processing method was my idea, and I would use some non-AI software to carry it out, but there was a lot of cumbersome interaction with a GUI. The software can be automated by scripting, but I never learned this particular scripting language, and learning it is not trivial. So I gave the chatbot a plain language explanation for how I did the processing, and asked it to script it. It got it wrong a bunch of times, and I had to keep feeding it the errors, and iteratively it fixed them and eventually gave the correct output. How do I know it’s not just giving me garbage? Because I have an output file generated with the non-scripted method (which I never showed to the chatbot), and the scripted output matches line for line.
I have also used it for literature research, and it has improved greatly in just the past 6 months. Until recently it would give often true info (as verified elsewhere) but fake references. Now it gives links to actual publications backing up its assertions.
Right now, I would never just hand AI some data and say “process this”. No way I would trust it, and doing so could result in Atrocious Idiocy as you say. But in the last year, it’s proven more useful to me than I could have imagined, as detailed above. Maybe reconsider your hard-line opinion.
We all know what this is, we’ve all seen it before. The only question is are we at the top? Or are we just half way there? Every time I’ve taken gains the last 2 years it’s been the wrong choice….
The thing I was talking about with a friend this week was so what if I sell some tech stocks to reduce my concentration, if NVDA goes down they’re taking the whole market down with them, nowhere is safe besides bonds if this implodes. So then the question is do I take 3.8% returns against what’s been an annualized return close to 20%? Missing those gains is also devestating.
Watching this gives me all sorts of déjà vu – too big to fail, Enron, .com bubble all in one.
For me wealth protection is important and no one has gone bankrupt taking profits
But everyone is different and I am fine with not taking 20 percent big gains
For me most valuable is peace of mind
Me, too. I’ve closed out nearly all my investments (holding onto the bonds) and I’m happy to miss whatever gains are left on the table. My 401k almost doubled, I’ll take my winnings and go home, smiling the whole way.
Now, we just need to sell the house before everything tanks.
Both long-term and short-term money flows are falling. So R-gDp and inflation will fall. So the FED will cut rates in Dec. That usually props up stocks. But I still think stocks have already topped.
And to assume that the Federal Reserve can solve our unemployment problem is to assume the problem is so simple that its solution requires only that the manager of the Open Market Account buy a sufficient quantity of U.S. obligations. That is utter naiveté.
What unemployment problem? I don’t see one yet….
Expert who predicted the dotcom crash says Americans could face a much bigger crisis soon
An economic crisis isn’t something that happens all of a sudden, but a lot of factors weaken the economy before a major blow triggers a meltdown. The famous bearish strategist, Albert Edwards, of Societe Generale, who had predicted the dot com crash, has sounded the alarm about a looming financial crisis, bigger than the 2008 market crash. The analyst, who refers to himself as a “perma bear,” spoke to Bloomberg and Fortune, sharing his opinions on the current ‘AI Bubble’ and the possibility of a market correction.
Edwards, who admits that he is a very bearish market strategist, has made some high-profile and dramatic predictions in the past, including the dot-com bubble burst. However, not all of his warnings have panned out, Fortune noted. “I think there’s a bubble, but there again I always think there’s a bubble,” Edwards told Bloomberg’s Merryn Somerset Webb, in a podcast. He was also firm in his opinion that “it will end in tears,” saying, “that much I’m sure of.” He further told Fortune in an interview that previous theories of a bubble before the 1999 and early 2000 dot com crash, and the 2008 financial crisis were also “very convincing.”
Edwards pointed to two key elements that would play a major role in the bursting of the bubble. Drawing parallels to the markets before the dot-com crash, he noted that some things were different today, which could make the crash much worse. He explained that previous bursts were triggered by the monetary policies of the Federal Reserve and the hikes in rates that exposed the market froth. However, this time, Edwards anticipates that the Fed will move away from “quantitative tightening to quantitative easing” with rate cuts, which won’t trigger a burst. He told Bloomberg that this policy could lead to a “further meltup,” making the eventual burst more devastating.
“What’s more worrying about the AI bubble is how much more dependent the economy is on this theme, not just for the business investments, which is driving growth, but also the fact that consumption growth is being dominated far more than normal by the top quintile,” Edwards told Fortune.
Little discussed issue but may end up making all this investment in power a sunk cost. Quantum computers are expected to lower AI’s energy demand by enabling more efficient calculations, which could reduce energy consumption by up to 90% compared to classical systems for certain tasks. They will achieve this by using quantum bits (\(qubits\)) to process vast amounts of information and perform calculations in a far more parallel and efficient way than classical processors.
This is a good point. If Quantum computing happens in the near term that changes everything.
I don’t know if Quantum computing is like fusion energy, always 50 years in the future, but it sure resembles it. I would not bet on Quantum computing efficiency to be a viable thing for quite some time.
This is what AI tells us, for what its worth,
Current Status (Now)
Niche Commercial Use: Specialized quantum annealers for optimization problems, such as those from D-Wave, are already in commercial use for tasks like supply chain logistics and manufacturing process streamlining.
Cloud Access: Access to quantum processors is widely available via cloud platforms like IBM Quantum, Amazon Braket, and Microsoft Azure Quantum, enabling researchers and businesses to run experiments without significant infrastructure investment.
Early Advantage: Some early cases of “quantum advantage” (outperforming classical computers for a specific, real-world task) have been demonstrated in limited applications like specialized material science simulations and financial modeling.
Near-Term Timeline (2025-2030)
The focus in this period is on developing “utility-scale” machines and hybrid quantum-classical systems for specific, high-value problem classes.
Key Milestones: Companies are targeting machines with hundreds of high-quality logical qubits.
2026: QuEra aims to have a 100-logical-qubit machine capable of outperforming supercomputers for specific tasks.
2027: Quantinuum projects hundreds of logical qubits.
2029: IBM plans to deliver a fault-tolerant system with 200 logical qubits.
Applications: Initial applications are expected in materials science, drug discovery, finance (risk analysis, option pricing), and logistics optimization.
Cybersecurity Shift: This timeframe is critical for migrating to post-quantum cryptography (PQC) standards in anticipation of future quantum threats to current encryption methods.
Long-Term Timeline (2030-2045+)
This era will feature fully fault-tolerant, large-scale quantum computers capable of broad, transformative commercial applications.
Fault Tolerance: Experts project that millions of stable qubits, necessary for breaking modern cryptography or performing complex simulations reliably, may be available by the mid-2030s or later.
Broad Utility: Quantum computing is expected to become a foundational technology for select industries, with the total market potentially reaching nearly $200 billion by 2040.
Consumer Access: These powerful machines are not expected to be personal devices but rather specialized co-processors accessed through cloud services, much like supercomputers today.
Quantum Computing Timelines 2025 – by Brian Lenahan
Apr 28, 2025 — * Optimistic Corporate Projections: Companies like D-Wave and PsiQuantum predict near-term commercial applications. D-
Substack
The timelines: when can we expect useful quantum computers?
Jun 23, 2025 — Assuming an exponential growth similar to Moore’s law, we predict that the first applications could be within reach ar…
Introduction to Quantum Computing for Business
Quantum Computing Industry Trends 2025: A Year of … – SpinQ
Oct 30, 2025 — Quantum Computing Industry Trends 2025: A Year of Breakthrough Milestones and Commercial Transition * Market Expansion…
I think this is highly improbable. Yes, there are technically applications in which quantum methods can provide speedups in network training (ex. optimization problems such as quantum annealing), but current technology lacks the qubit scale and connectivity, and nothing has been proven in practice yet. We are still very far away from this becoming a reality. We may see multiple bubbles before this point.
I’ve worked in the area and have a couple friends doing research work in quantum computing right now. Personally, I am very pessimistic on quantum anything being relevant to AI advances, as the most useful aspects of quantum seem to be in security and encryption (at least for now). Our tech for semiconductors is so advanced that quantum would need decades of advancement to even come close – not to mention the lack of any new and useful quantum algorithms in the past decade.
Even in optimistic scenarios for quantum computing, we wouldn’t see results in time to impact the current AI race, at least from my vantage point.
I mean the one thing I’ll say is never underestimate the govts ability to f*** up the free market. So yeah it’s all sketchy, but will they actually let it implode? My understanding from politicians is we’re done with bubbles and recessions, things only go up.
OpenAI’s CFO has been hinting that they’re expecting a government backstop because it’s so important that we win the AI arms race. Apparently the logic goes that since we’re all allegedly on our way to AGI, and getting there first is of national importance, then there’s a de facto government backstop and thus banks should lend to us and our debt should be cheaper. The result is that stonks go up.
Unpopular opinion but I think we get stuck at LLMs for quite awhile. AGI is a huge step away from that
I am still awaiting fully autonomous vehicles. Having hyper accurate maps of certain cities and motion sensors is not fully autonomous.
I don’t seriously think we’re going to get to AGI, at least not with this technology. My bet has been that we’re headed towards a lot of AI slop and model collapse.
Bloomberg just had a great Odd Lots podcast on this where one of these analysts was saying that you have to make a distinction between whether they’re building business productivity tools or whether they’re trying for AGI. He said that everyone in the industry will tell you they’re doing business tools until you confront them about how the financials of that are going to work, in which case they’ll then pivot to it being about AGI, and all of their claims will depend on that being the case, like you see with this government bailout idea.
Matt B,
Definitely an arms race! Don’t even know the fascination with AGI although really depends on what definition you used. I see the danger of turning over military analysis and decision making fully over to humans. Not to say they won’t just kill people either way but now they will blame the mistakes on computers rather than any accountability whatsoever. The military needs to do this as only way to win is to fight computers with computers but greatly raises the potential of a black swan type of event without human decision making as hard for AI to really comprehend geopolitical and other implications. Seems all but inevitable at this point though. UN could of course ban it but who listens to them anyway!
Weren’t share buybacks effectively “illegal” until the 1980s, and for very good reasons? Hasn’t the subsequent relaxation in regulations led to the dangerous situation now set out in the article?
Seems even more important to own gold ( and silver).
Until the ‘70’s it was believed that government had a responsibility and the authority to regulate bad behavior in society. Democratic mob rule and their chocolate donut laws was to be kept in check with representative government for all in a republic where legislators were respected for their ability to transcend the pressures to cave. Since then we have allowed for the creation of the genetic freak e-Kahn and his blind followers, the shapeshifting Lord Garth has finally taken charge of the white asylum, and everybody is convinced that investing in Nomad is a good deal at 2% back. Maybe it might be time to own aluminum foil too.
Yes, John, stock buybacks were illegal until 1982, recognized as they were as a form of stock manipulation.
But we’re just so much smarter than we were in the 1930s, amirite?
AI – the new Dot Com frenzy. During the dot com bubble, many companies were not making a profit, but their stock prices were sky high. AI is the new insanity that will come crashing back down to earth.
Note – about 2 months ago, there was a story saying that Nvidia was now “worth” $5 Trillion !!! I did a quick search and saw that the entire US economy in 1987 was $5 Trillion. Now, they are saying that ONE COMPANY is worth the entire U.S. economy in 1987, It’s beyond insanity.
Market cap valuations of a company’s “worth” are a bad joke utilizing Grade 3 math (take the most recent sale price of a stock and multiply it by number of shares outstanding) and the idiotic premise that were all owners of shares to simultaneously SELL it would generate zero downward pressure on the price — against all experience of supply and demand ever in recoded history.
They just might have some utility in making comparisons between companies, but beyond that you’re off with the faeries …
“Market cap valuations of a company’s “worth” are a bad joke utilizing Grade 3 math”
🤣 yes because if you want to buy the entire company, it will cost you a LOT more than market cap because you will have to pay a premium. Which is why market cap spikes when a merger or buyout is discussed/announced. So, the value agreed on in a merger is the true market value of the entire company, and it’s a lot higher than market cap. In a way, market cap is the LOWEST price that you can buy the company at this moment.
There are very few “takeunders” where a buyout price is below market cap.
No, the dot.com bubble was generated by Alan Greenspan injecting, then draining legal reserves period. He was afraid of Y2K.
Share buybacks rock, look at what it did for names like Boeing and GE. Now for those of you that think this is sarcasm… it kind of is but the shares of those stocks have not suffered that much…. All things considered.
Split adjusted GE is doing great and BA, well off its 2022 lows. Heheh. So, one could reasonably say that share values hasn’t been impacted that much.
And the mag 7, sure one can argue they are burning cash on buyback was a waste, but until recently, they didn’t have anything better to blow their cash on. So, may as well enrich the insiders with all of that corporate buybacks. I think Zuck might need to build another bunker, may be not on Kauai, how about Niihau. And this insane spending they are doing is helping to float some portion of the economy. And in 20 years, is anyone really going to miss Facebook or Nvidia, for sure no one misses Intel or Cisco right now.
May we all dump their stock before the inevitable crash. LOL
Wolf I was reading from a martlet commentator. Do you agree?
Mega cap technology companies are using their high credit ratings (AAA/AA) to issue large amounts of long-dated bonds, which are quickly purchased by life insurers, sovereign wealth funds, and other investors seeking stable, long-term returns in a low-inflation environment.
• These companies, such as Amazon, are deploying capital raised through the bond markets to invest in or lend to leading AI firms like OpenAI and Anthropic, often combining funding with cross equity ownership to participate directly in AI growth and infrastructure.
• The process involves circular financing: mega caps invest in AI firms, who in turn purchase cloud and hardware services from their investors, creating a feedback loop that amplifies both technical dependency and perceived growth across the ecosystem.
• This strategy enables mega caps to vertically integrate within the AI industry while capturing upside in both debt and equity, positioning them to benefit whether AI delivers extraordinary growth or simply provides reliable yield—making them winners in volatile markets.
I may hold a different view.
There will be money to be made on AI, but not everyone will be certain of success. The technology is still immature and mostly performs specific, well-defined tasks like information amalgamation and analysis. It can also create well-defined code, photos, music, etc.
Don’t forget ML for automation, and maybe we will see AI that learns from all of its current tasks to become significantly intelligent.
The Industrial Revolution took over 100 years to become mainstream, so I would not completely discount AI as of yet.
I agree completely and in 15 to 20 years we’ll see some of today’s promises will come to fruition. But there will likely be a crash in the meantime. Google, Microsoft, and Apple all crashed during the dotcom bust. And look at them now.
When companies DECIDE to borrow rather than issue new stock.
When companies DECIDE to borrow to buy back their own stock, they must have decided that rates are attractive.
When rates in the repo market continually bubble up despite the SRF….
When the national debt surges $330 Billion in just a couple of weeks….
this all suggests rates are way too low…….manipulated to unrealistic levels, and the Fed suggests more cuts as if to reverse layoffs.
Add in embedded inflation spike of the past 4 years, records stock prices, gold, insurance, medical care, etc….
Reckless Fed, IMO.
The U S is all about consumption. So when can we expect AI products on the shelves?
It seems right now the only AI in the consumer pocket is higher electricity bills.
1. Most of what consumers consume are services, not goods. And services are where most inflation is. Whether services get more expensive or cheaper depends on market conditions, not AI. A lot of services that you’re using are free to consumers. Others cost a bundle.
2. In terms of goods: Automation, which has been improving for decades in leaps and bounds, has made most manufactured goods at lot better and relatively cheaper. Look at an $800 laptop today. A computer of similar power would have cost a fortune 30 years ago… Cars have gotten more expensive, but have gotten vastly better than 30 years ago, in every aspect, from power to safety. I know there is always someone here that thinks their IBM PC with two floppy drives and 640k of memory running on Microsoft DOS ($4,000 at the time) was the best computer ever built, and their 1985 Buick the best car ever built. But that stuff is just funny.
Wolf, don’t knock the IBM PC in 1985. The two floppy drives might be a bit obsolete today, but when the 20MB hard drive came out, I was in computing nirvana, never to need more!
I agree the 1985 Buick was not the best car ever built. A friend’s grandparents had a 1979 AMC Pacer. That was one fine machine!
I have a Windows XP, purchased in 2004 that works fine. I use it to scan documents and as an archive machine for storing important data going back years, including tax returns, old appraisal PDF’s and invoices. It’s off-line so no one can hack into it.
There are no Aristocrats in the Mag 7.
Not sure if it matters who wins the AI race for who is best! If number two is almost as good at less than half the price.
For a few the FOMO factor will matter, but for most if is all about cost.
Seems we would have already learned something with are dealings with China. We lost that battle and must now hide behind tariffs to protect us from free trade.
I have an honest question:
What does someone mean when they say : Win the AI race .
I see a world where there would be thousands of LLMs of varying capability.
Chinese and other countries would come with LLM model which provides 90% of the features at 10% of the cost ( e.g. deepseek ) and would make the weights opensource.
How would these hyperscaler then make money of their trillions of dollars of investment on AI.
AMD’s stock is having its worst month in three years
AMD -8.25%
Kohl’s stock enjoys meme-like rally as blowout earnings confirm turnaround has taken hold
KSS +31.88%
Despite the rally, the share price is where it had first been in 1998.
If the long running “the Aristocrats” joke among comics is any indicator, there are certainly plenty of them in the Mag 7 C-suites …
Oops — that was intended as a reply to Bob B above
…..And giving themselves big “dividends”!
Pretty funny that you mentioned Kohl’s. I just ordered two pairs of compression socks from them online 10 minutes ago. I can’t believe I’m now at the age where I’m ordering compression socks!
“I can’t believe I’m now at the age where I’m ordering compression socks!”
Look at it from bright side: “At least, I’m still ordering…” 🤣❤️
In a downturn where does the money flow? My take is it sits on the sidelines but in the end it must make its way back into the equity market, and of course each time being more concentrated. Those who get out at the right time and predict the best time to get back in simply get more wealth. I even did this will COVID collapse on a small scale although I didn’t get out since it wasn’t sold as something massive but I did take sideline money and put it into index funds. Where else who the money go? Bitcoin, gold, etc, but not in the numbers that come out of market.
I always wonder about this kind of statement: Money on the sidelines.
How can money be on the sidelines.
I have sold my stocks to keep my money on the sideline but at the same time, someone has spent his/her money to buy my stocks. So money on the side line is a myth . Isn’t it ? Unless FED creates money out of thin air or do something.
Bitcoin’s down because total reserves are down:
Oct 2025: 2,944.7
Sep 2025: 3,068.1
Aug 2025: 3,281.9
Jul 2025: 3,340.3
Jun 2025: 3,356.0
Powell’s kicking ass.
Where the hype or more politely the great expectations meet cold reality is when the projected data center orders a supply of electricity. ‘Fine’ says the provider, but before we gear up just for you, and commission some gas turbine generators, we need a fifteen year contract, with guarantees.
Apparently the projected demand for data center power in Texas, is a considerable percentage of the entire state’s existing supply. This is the grid that crashed in 2022.
Great article Wolf. Sell program should kick in by 2pm. $vix should bounce now!!!
Maybe I’m off base here but a lot of AI investment is by Amazon for data centers for AWS, which is pretty foundational to a lot of business operations. So in a way it makes sense as helping strengthen the backbone of commerce. However, it’s still a bubble the valuations and expected returns on investment are seemingly insane.
Wolf,
I’m getting e-mails from some people who read your site and harvesting names and e-mails from your comments section. They and are sending me junk e-mails addressed to “Swamp”. I wonder if you have ever heard of this happening before.
No. But everyone’s inbox sooner or later ends up getting junk email, and eventually lots of it, which is why there are junk folders and junk warnings in your system. If they got your email from this site, they could see that your name isn’t swamp but the name you use in your email.
Fake emails work fine for posting comments. I encourage that. Just make sure you remember how to spell it.
Here’s an example the junk e-mails I’m receiving :
From: Joe, NRSC HQ
Sent: Nov 25, 2025 6:04 PM
To:
Subject: I have shocking news, Swamp…
If/when the AI bubble explodes and we have a dotcom crash event, I’ll always remember that it was here at WolfStreet.com where I first heard it called. :(
This was about share-buybacks being at risk.
Wolf, thank you for the bit about Meta and Blue Owl, very interesting.
Q on shares buyback: if we replace shares buyback with paying an equivalent qualified dividend, do you see any difference? If yes, please explain. If not, are you opposing dividends too?
I’m opposed to share buybacks if they’re funded by debt. Dividends shouldn’t be funded by debt either.
Sam Altman raids NVDA for $10B for an IOU. NVDA $100B credit to
Sam Altman is down to $90B. Sam Altman blocks AMD from competing
with NVDA. AMD issued credit to Sam Altman to counter NVDA. Other chip mfg and software co issued Sam Altman fake credit to be in the money,
Someone please enlighten me:
1. “Meta is going to lease the computing power, but in four-year terms”
How they manage to keep that off book?
I happen to own a commercial on four year lease with renewal option and my tenants are reporting it q/year as a spending…
2. How all these AI gorillas companies are going to make money at “consumer level”…