저자: OrionCipherAnalysis진행자 · 외국인
요약 보기
마이클 베리가 AI 관련 고평가·순환 금융 구조(엔비디아-오픈AI-오라클 등)를 이유로 약 11억 달러 규모의 숏 포지션을 유지·확대 중이라고 주장. 주요 숏 대상은 엔비디아(풋옵션), 마이크론, 팔란티어, 오라클, Nebius. 붕괴 촉발 요인으로 자본지출 둔화, OpenAI 자금조달 실패, 전력 인프라 병목을 제시. 반대 논거(실적 실재, 생산성 향상, 실수요)도 소개하며 투자자 자기점검을 권고. 특정 자산 가격 방향에 대한 명시적 1~3개월 예측은 없음.
다듬은 글
In 2008, one man made a billion dollars betting against the housing market while the rest of Wall Street called him insane. His name is Michael Bur. You already know him. He was the doctor with a glass eye and a t-shirt sitting in a small office in California, reading mortgage documents nobody else on the planet was willing to read.
He saw the collapse coming 18 months before it happened. He placed the trade. He got mocked. And then he got proven right in the most expensive way possible for everyone on the other side of that bet. Michael Bur is back and this time his target is not housing. This time his target is the artificial intelligence trade.
Six out of the seven short positions he currently holds are already profitable. He has been adding to those positions almost every month this year. and in a filing dated the 26th of August 2026, he made his largest single edition since he started the campaign. If you own an index fund, a retirement account, or any exposure to the biggest technology names in the world, the man who called the last great financial collapse is quietly telling you something is wrong.
And by the end of this video, I am going to walk you through exactly what he sees, exactly why the rest of Wall Street is pretending it does not exist and exactly what you can do about it before the story finishes writing itself. Because history rarely repeats, but it does rhyme. And right now, it is rhyming loudly.
Before we go anywhere, we have to clear something up. Almost every headline about Michael Bur frames the story as Bur versus artificial intelligence as if he is placing a bet against the technology itself. That framing is wrong and it is the reason most people who read those headlines are missing the actual argument.
Bur is not betting that artificial intelligence is fake. He is not betting that nobody will use it. He is not even betting that the biggest companies in the space will fail as businesses. He is betting on something much narrower and much harder to argue with. He is betting that the specific financial structure holding up the current valuations of a handful of artificial intelligence linked companies is going to break under its own weight, not the technology, the financing, the circular deals, the vendor loans, the revenue that quietly loops back to the vendor who provided the capital in the first place.
This distinction matters because it means the bulls and the bears are arguing about two different questions. The Bulls are asking, "Will artificial intelligence change the world?" The answer to that is almost certainly yes. The Bears, including Bur asking, are the current prices being paid for these companies justified by the current cash flows they are generating?
And the answer to that question on the numbers alone is a much harder yes to defend. So, when you hear someone say Bur is betting against artificial intelligence, correct them. He is betting against a very specific form of financial engineering that happens to be wearing an artificial intelligence costume this year.
Let us get concrete because vague warnings are cheap and Bur is being extraordinarily specific. According to his most recent filings, his short campaign centers on three groups of names. The first group is the chipmakers. This includes his largest single position, which is short exposure to Nvidia through put options.
He has also added positions in Micron and in Nebius, a smaller artificial intelligence infrastructure name that most retail investors have never heard of. The second group is the software and services layer. Palunteer sits at the center of this one. Palunteer has become the poster child of the retail artificial intelligence trade and Bur has taken a large position betting that its share price will fall.
Palanteer trades at a valuation that on any traditional measurement only makes sense if the company grows revenue at rates almost no software business in history has ever sustained for the required number of years. The third group is the cloud and data center providers. Oracle is the anchor of this bucket following its $300 billion cloud deal with OpenAI last year, which has left it with an enormous obligation and a rapidly ballooning debt load to deliver on the compute promised.
The total size of Bur's short campaign is estimated at roughly $1.1 billion in notional exposure. That is not a hedge. That is a conviction bet. And it is happening while Bur is publicly stating in his own words that this is quote an asset bubble plain and simple. The most striking part is not the size.
It is the concentration. Bur is not spreading his bet thin. He is picking specific names at specific price levels that he believes are the weakest links in the chain. If he is right and even one of them cracks, the story unravels for the rest. So what is the pattern Bur sees? Let me explain it the way he would if you were sitting across from him at a diner.
Every great bubble in modern financial history has been built on one specific accounting trick. Not always the same trick, but always the same idea. The trick is finding a way to make $1 count as revenue in more than one place. In the late 1990s, telecom companies did it with vendor financing. Cisco, Nortell, and Lucent extended loans to fiber optic startups so those startups could turn around and buy equipment from Cisco, Nortell, and Lucent.
That equipment sale then showed up as booming revenue on the vendor's income statement, which pumped the vendor's stock, which allowed the vendor to raise more capital, which was then extended as more vendor financing. The music stopped in 2001. Cisco lost roughly 80% of its market value. Lucent went from $180 billion in market capitalization to essentially zero.
Nortell eventually filed for bankruptcy. In the mid200s, mortgage lenders and investment banks did the same trick with a different rapper. They packaged loans, sold them to a special purpose entity, bought them back at a slightly different price, and booked the whole cycle as fees, gains, and revenue.
Bur is the man who read the paperwork on that trick and figured out where the bomb was. The pattern he now sees in artificial intelligence is a modern version of the same idea. Nvidia has invested up to $100 billion into OpenAI. Open AAI has committed to spending $300 billion with Oracle. Oracle has committed to buying massive amounts of hardware from Nvidia.
The same dollar flows through the loop and each leg of the loop counts as revenue on someone's income statement. It is not illegal. It is not even hidden. Bloomberg published a full explainer on what it calls, quote, the circular deals underpinning the artificial intelligence boom. The Wall Street Journal ran the same story.
What is happening is that the market has decided the loop does not matter. Bur has decided it does. And history says when a man who has read every footnote of every filing has decided that something matters, it is worth at least a few minutes of your attention before you dismiss him. Now, before we go any further, we need to address the obvious counterargument because you are going to hear it a hundred times if this video gets shared.
Bur has been calling a bubble for years. Nvidia has kept going up. Palunteer has kept going up. The market has kept going up. Bur has been quote wrong quote for a long time. Except that is exactly what people said about him in 2006. When Bur placed his housing short in 2005, the trade lost money for almost 2 years.
His own investors began threatening to sue him. Some pulled their capital out at the worst possible moment. The financial press mocked him. He almost got fired from the fund he founded by the people whose money he was trying to protect. Then 2008 arrived and the same investors who had wanted him gone were suddenly writing him thank you letters.
The lesson from that period is not that Bur is always right. The lesson is that being early and being wrong look identical until the exact moment they do not in a bubble. The loudest evidence for the bulls is that the trade keeps going up. That evidence is also the reason the crash when it finally arrives has further to fall.
Bur himself has said in one of his rare public statements this year that in his view the market capitalization to gross domestic product ratio, sometimes called the Buffett indicator, is now approaching a level not seen in nearly 150 years of American financial history. He believes that indicator is on the edge of breaking a record it has held since before the Great Depression.
He might be wrong, but the setup is not a random hunch. It is a measurable extreme in a measurable market with a measurable financing structure that has cracked in almost every previous instance where it appeared. Let me do something most videos on this topic refuse to do. Let me actually argue the other side honestly and in its strongest form.
Because if you only hear the bare case, you are being sold, not informed. The bull argument for the artificial intelligence trade in 2026 goes something like this. First, the earnings are real. The four largest cloud providers in the United States have raised their combined capital expenditure guidance for this year to more than $700 billion.
That spending is showing up as revenue for a small number of vendors. But the customers behind that spending, Amazon, Microsoft, Google, and Meta are all profitable companies with hundreds of billions of dollars in existing cash flow. They are not fictional entities. They are the most profitable companies in the history of the world.
Second, the productivity gains from artificial intelligence are already measurable inside those same companies. Coding assistants are shipping working software. Customer service is being automated at scale. Advertising targeting has improved measurably. This is not a promised future. It is a booked present.
Third, unlike 1999, the physical infrastructure being built has a genuine end use. Fiber optic cable in the year 2000 was overbuilt because the end users had not yet arrived. The end users for the current data center buildout have already arrived. Every large enterprise in the western world is signing multi-year cloud contracts right now.
These arguments are not weak. They are the reason smart investors on the other side of Bur's trade are staying long and they may prove correct. The honest counter to all three points is this. Even if the technology is real and the earnings are real and the productivity is real, the price you pay for a share of that story still matters.
In 1999, the internet was also real. It still cost most of its early investors 80% of their money to hold the wrong companies through the correction. Being right about the technology and right about the trade are two completely different things. So, what actually causes this to unwind, if it unwinds at all?
There are three plausible triggers and they are worth naming out loud because if you know what you are watching for, you will not be surprised when you see it. The first break point is capital expenditure fatigue. The four largest hyperscalers have raised their spending guidance almost every quarter for the last 2 years.
There is a psychological ceiling somewhere around $1.5 trillion in cumulative artificial intelligence infrastructure spending where analysts will begin demanding measurable revenue attached to the spend. When one of the big four finally guides its capital expenditure down instead of up, the other three will follow within a quarter because none of them wants to be the last one still spending into a slowdown.
The second break point is customer failure. OpenAI is on pace to lose roughly $14 billion in the current year on an annual revenue run rate of approximately $25 billion. Its own internal projections show cumulative cash burn of more than $200 billion between now and the year 2029. If any part of that funding path breaks, a delayed funding round, a change in valuation, a competitor taking market share, the ability to honor existing multi-year compute contracts comes into question.
That question would then land on the earnings call of Oracle, which would land on the next earnings call of Nvidia in that order. The third break point, and the one I personally think is most likely, is electricity. All of this artificial intelligence infrastructure has to run on physical power. The electrical grid in the United States was not designed for a doubling of industrial demand in 3 years.
Interconnection cues in some regions now stretch out 5 to 7 years for new large-scale power delivery. A concrete data center can be built in 18 months. A gawatt of new power cannot be summoned from the ground on the same schedule. This mismatch is real and it is getting worse and it may quietly slow the entire buildout without a single dramatic headline.
Any one of these three on its own is enough to break the loop. And when a loop breaks, it does not slow down gracefully. It unwinds. Here is the part that almost no video on this topic ever talks about. When a bubble ends, the money does not disappear. it moves and a small group of patient, disciplined investors end up on the other side of it every single time.
The first group is cashrich acquirers. When artificial intelligence valuations reset, the companies that stay disciplined during the boom will be able to buy real assets, data centers, chip inventory, infrastructure businesses at prices that would have been unthinkable at the peak. This is exactly what happened in the year 2002 with fiber.
Companies that bought those distressed assets became the giants of the following decade. The second group is the physical infrastructure owners, independent power producers, grid equipment manufacturers, copper, transformers, cooling systems, water. Every one of those industries has quietly repriced upward to reflect artificial intelligence era demand without carrying the same speculative overhang as the technology names themselves.
If the artificial intelligence trade cools, the physical inputs it depended on stay expensive because the demand for reliable electricity does not vanish. The third group is patient capital. Warren Buffett, who has now built the largest cash pile in the history of his company, is not sitting on that money by accident.
Neither are the pension funds that quietly rebalanced out of technology names in the first half of this year. The people who win the next cycle will not be the people who timed the peak of this one. They will be the people who had dry powder when the good businesses went on sale. Now, let us bring this back down to your life.
If you have a retirement account in the United States, a workplace pension in the United Kingdom, or a registered retirement savings plan in Canada, you own some slice of the trade Bur is betting against. You did not choose to. That is simply how modern passive investing works. The largest handful of technology names now make up more than a third of the total value of the standard American stock index.
Which means when this loop wobbles, so does your retirement. This is not a call to panic sell. I do not do that on this channel. It is a call to know what you actually own. There are three practical questions worth asking yourself tonight. And I mean literally saying them out loud. The first question, if Nvidia dropped 40% tomorrow morning, what would happen to your portfolio?
If you do not know the answer, that itself is the answer. The second question, how much of your investment returns in the last 2 years came from actual earnings growth in the companies you own and how much came from the same companies simply being valued at a higher multiple of those earnings. Multiple expansion works in both directions.
The third question, if the artificial intelligence buildout takes 10 years to fully monetize instead of three, does the current price you are paying for these stocks still make sense? Because history is very clear on this point. Every major technology wave has taken longer to reach real profitability than the peak of its stock market cycle assumed.
None of these questions require you to be right about the exact timing. They only require you to know what you are holding and why. Here is the entire video in one sentence. Michael Bur is not betting against artificial intelligence. He is betting against the price the market has decided to pay for a very specific piece of financial engineering wearing an artificial intelligence costume.
The rest of Wall Street has decided this bet does not matter. History suggests it does. And when the loop finally breaks, the people who understood the structure will be the ones buying pieces from the people who did not. You do not have to agree with Bur to protect yourself. You just have to know that a man who read the footnotes on the housing market before anyone else is now reading the footnotes on this one, and he is placing his own money accordingly. The smart money is not shouting. It is quietly moving. Now you know where.
Comments 0