저자: 미상 · 외국인
요약 보기
AI 설비투자는 금리 인상에도 둔감해 Ackman 주장대로 '브레이크 페달' 효과가 약화된 것으로 제시됨. AI 기업 회사채 스프레드 약 115bp(일반 투자등급 78bp 대비 높음), 하이퍼스케일러 회사채 발행 2027년 4,200억달러(+60%) 전망. 전력망 병목이 실질 제약이며, 금리 인상은 일반 경기만 억누르는 2속 경제 구조. 장기적으로 AI가 생산성을 높여 디스인플레이션을 가져올 수 있으나, 과도기 비용 부담자 문제를 제기.
다듬은 글
Hello and welcome to Investment Insight. Before we begin, make sure to subscribe and hit the like button. Now, let's break down what's happening in the markets today in a clear and simple way. So, if you've ever taken like an economics 101 class, or honestly, if you've just watched the financial news at any point over the last few decades, you learned this one very basic golden rule.
Right. The classic rule of the brake pedal. Exactly. The brake pedal. When the Federal Reserve raises interest rates, borrowing gets expensive. And because borrowing gets expensive, regular people and, you know, big companies, they start spending a lot less. Yeah, and because they spend less, the entire economy just naturally cools down.
Right. It is supposed to be this inescapable law of financial gravity. Well, yeah. I mean, the cost of money goes up, so the velocity of money slows down. It's the fundamental mechanism that central banks all over the world use to fight inflation. They rely entirely on this idea that high borrowing costs will eventually force everyone to uh to tighten their belts.
But today, looking at a massive stack of recent reports and data we've gathered, we have one huge question to answer. We are doing a deep dive into this fascinating theory proposed by billionaire investor Bill Ackman. We're going to explore whether this seemingly unstoppable wave of artificial intelligence investment is actually, well, breaking that fundamental rule of economics.
Yeah, and looking at the recent data, we're witnessing dynamics that, frankly, economists are having a really tough time modeling. I mean, they're using their traditional playbooks, and the historical models just aren't mapping onto the reality of corporate spending right now. It's wild. Okay, let's unpack this.
To understand Ackman's core argument here, I want you to imagine a very relatable everyday scenario. Imagine you want to renovate your kitchen. Oh, a a classic money pit. Exactly. You draw up the plans, and you go to the local bank for a home improvement loan. Now, if the bank tells you they're charging, say, a 3% interest rate, you might sign the paperwork on the spot, right?
You upgrade to the granite countertops, maybe get the fancy fridge. You're locking that in immediately. Right. But, if the bank says the interest rate is 7% or 8%, well, the math changes completely. Your monthly payment just skyrockets. You're probably going to cancel the project because it just got way too expensive to borrow that cash.
And that right there is the Fed's brake pedal working perfectly on the consumer level. The high rate successfully destroyed your demand for a new kitchen. Right. Which in turn means, you know, you aren't paying a contractor. You aren't buying lumber. You aren't buying appliances. That lack of spending is what actually cools inflation.
But, and this is the crucial part of Ackman's theory here, what if, right before you went to the bank, you found out with absolute certainty that there was a literal chest of gold buried under your kitchen floorboards? Okay, now we're talking. Right. Suddenly, you do not care if the bank is charging 7%, 10%, or even like 20% interest.
Not at all. You are going to borrow whatever money you need to go rent a jackhammer and rip up those floors because the expected payoff is just astronomically high. You will happily pay a really painful interest rate because the reward totally dwarfs the cost. That analogy perfectly illustrates the psychology in the tech sector right now.
Artificial intelligence, and specifically this frantic race to achieve what the industry calls super intelligence, is that chest of gold. The ultimate prize. Exactly. Ackman's thesis suggests that AI represents such a massive technological leap that the expected return on investment, the ROI, is viewed by these companies as practically infinite.
Infinite return. Wow. Yeah, and because of that perceived infinite return, they're suddenly just ignoring the Federal Reserve entirely. And we can see this in the hard data. The Fed actually raised short-term rates to a range of 3.75% back in mid-September. Which is pretty high by recent standards. Historically, a jump to those levels acts like a heavy wet blanket on corporate spending.
It makes funding new projects incredibly painful. But AI capital expenditure is just roaring ahead. Ackman isn't saying interest rates never work. He's saying the fastest growing sector right now is shockingly numb to them. See, I have to push back here though, because surely this expensive money has some effect.
I mean, they can't just ignore math, right? We saw this exact same psychology during the dot-com bubble. Oh, absolutely, the fiber optic boom. Exactly. Telecom companies ignored borrowing costs because the internet was going to change everything. They spent billions laying all this cable, and then they just ran out of cash before they became profitable.
Are AI companies really immune to this financial gravity, or are we just watching the inflation of a massive bubble that's going to pop? Well, what's fascinating here is that companies actually aren't ignoring the math. They are just willingly swallowing the higher costs. Okay, how so? According to recent bond market data, we can prove this.
When a massive corporation needs billions to build out infrastructure, they don't go to a local bank. They issue corporate bonds, which is basically them asking the open market of investors for a giant loan. Just to clarify for the listener, when recent analysis mentions these companies, we hear the term hyperscalers a lot.
We're talking about the massive cloud providers, right? Your Amazons, Google's, Microsoft's. Yes, the big players building data centers the size of small cities. Right. And those bond investors demand to be paid an interest rate based on how risky they think the loan is. Like how a credit score changes a mortgage rate.
Exactly how it works. And investors have been demanding a noticeable premium to lend these AI companies. Recent financial analysis shows the bond spreads, which is just the extra risk penalty these companies pay over safe government bonds, are sitting at about 115 basis points for AI corporate bonds.
Okay, for those of us who don't speak Wall Street, 115 basis points is just 1.15%, right? Right, exactly. But the mechanism of why they're charging that is what matters. Investors see these hyperscalers borrowing billions to buy microchips, but they also see that the revenue from everyday consumers paying for AI hasn't fully materialized yet.
So, they demand that extra penalty. And that penalty is real. To put it in context, the average for standard investment grade corporate bonds in other industries is only about 78 basis points. Oh, wow. So, it's a huge jump. It is. AI companies are paying a significantly higher penalty to borrow money compared to standard blue-chip corporations.
And yet, they're totally happy to pay it. Expected corporate bond issuance from these hyperscalers is projected to hit $420 billion in 2027 alone. That's a 60% jump from 2026. Here's where it gets really interesting, though. Because you can throw all the money in the world at a problem. You can pay whatever risk penalty the bond market demands.
But eventually, financial engineering hits a physical wall. It turns out it isn't the money that's stopping the AI boom. It's electricity. Yes. This is perhaps the most critical bottleneck right now. The speed limit has completely shifted away from interest rates to the physical capacity of the power grid.
Right, because AI doesn't just run on magic invisible code in the cloud. It runs on physical servers. Massive data centers. And these are fundamentally different from the data centers we were building even 5 years ago. Because of the chips, right? These advanced GPUs run insanely hot. You can't just slap a big fan in the room anymore.
You need these advanced liquid cooling systems, and they just draw a staggering amount of power. The power density is historically unprecedented. When we look at recent numbers from the US Energy Information Administration, the EIA, the projections are daunting. They're forecasting US power consumption to hit a record-breaking 4.35 trillion kilowatt hours by 2027.
4.35 trillion, that number is just it's so large, it's almost abstract. But we can look at a real-world example. According to recent reports, Oracle and Blue Owl are backing this massive multi-billion-dollar data center in New Mexico called Project Jupiter. Right, a state-of-the-art facility. Exactly.
But it faced serious potential delays. And it wasn't because Wall Street said no to the funding, the delay was because they simply couldn't secure enough electricity from the local grid to run the thing. It highlights this fundamental truth that tech companies are currently relearning. You can print money, you know, you could write new code overnight, but you cannot print a power plant.
Right. The AI cycle isn't just about buying chips anymore. It's a frantic race for transformers, heavy-duty cooling systems, natural gas facilities. They're even talking about restarting dormant nuclear plants. And even with the current Trump administration focusing heavily on deregulation, trying to cut red tape, and push for more domestic energy production, the physical reality of construction is still there.
Oh, absolutely. You still have to survey the land, pour the concrete, build the substations. Deregulation speeds up the paperwork, but it doesn't bend the laws of physics or construction timelines. It still takes years. The physical grid is the ultimate arbiter here. But I want to circle back to the economic theory for a second.
Because the way this AI spending interacts with the rest of the economy is where things get truly complicated for the average person. This is exactly what I wanted to ask about. So, if AI companies just keep spending, and they keep buying up all this land and power regardless of interest rates. Does that mean Ackman is right and raising rates actually causes inflation now because they're just driving up the prices of all those physical materials?
That is a very common misinterpretation. We have to be careful with that logic. Recent analysis from the Minneapolis Fed provide some vital clarity here. They point out that while the AI sector might currently be immune to high interest rates because of that chest of gold mentality, the rest of the economy absolutely is not.
The brake pedal still works perfectly on regular demand. Okay, I picture it like this massive two-lane highway. You have one lane reserved for everyday Americans and local small businesses. And right now, the Fed has covered that lane in massive brutal speed bumps. The high interest rates. Right. If you try to drive down that lane to buy a house or like take out a loan for your local bakery, you're going to wreck your financial suspension.
It actively stops people from spending. Yep, it hurts. But right next to it, there's this luxury toll road built specifically for Big Tech. The tolls are insanely high, but these companies have so much cash in the trunk that they just toss it out the window and keep speeding at 100 miles an hour. That two-lane highway is a really accurate way to visualize a two-speed economy.
Everyday sectors like commercial real estate or local retail are incredibly sensitive to borrowing costs and they are hurting. And if we connect this to the bigger picture, you start to see the nightmare scenario this creates for the Federal Reserve. Because the Fed only has one primary tool, the overall interest rate.
They can't just raise rates for Google and lower them for a home buyer in Ohio. Exactly. They have to apply a blunt instrument to a divided economy. If the Fed looks at the data and says, "AI spending is out of control. We need to raise rates to cool down raw material costs." The tech giants barely feel it.
To squeeze the AI sector hard enough to actually make them stop spending, the Fed would have to raise rates so astronomically high that it would completely crush the average consumer in the other lane. And there's a secondary effect here that is equally concerning. We've established the sheer scale of money required is staggering.
Like Morgan Stanley estimates AI will need 1.5 trillion dollars in external funding by 2028. And Moody's projects CapEx for just the top six tech companies will hit 1 trillion dollars by 2027. It's a massive vacuum effect. Right. The capital markets are huge, but they aren't infinite. When one single sector demands trillions from the financial system, it just sucks up all the available capital.
And that means AI might actually be keeping market interest rates high for everyone else. Money becomes scarce for regular loans because it's all flowing to data centers. So, the everyday person paying a 7% mortgage isn't just suffering because of the Fed. They might indirectly be paying the price for the tech industry's insatiable demand for capital.
Wow. So, what does this all mean for the listener? If we're living in this imbalanced two-speed economy, the traditional ways we measure market health are going to be warped. I mean, if you checked your retirement account in late September, everything looked incredibly green. The S&P 500 was climbing, the NASDAQ was up, Microsoft was hitting highs.
It all looked totally healthy. So, where is the danger actually hiding? The stock market is blinding us right now. Equity markets just reflect the unbridled optimism of the AI gold rush. If you want to see the real warning lights flashing, you have to look away from stocks and focus on the bond market.
Well, because bonds don't lie about risk, right? They reflect the cold, unglamorous reality of cash flow and debt obligations. Exactly. Moving forward, a company's financial structure, its balance sheet, is going to matter just as much as its tech. If you're a massive tech giant building a data center using cash from your already profitable software business, you're fine.
You're insulated from the high cost of debt. You're driving on the toll road, but you own the car outright. Right, but if you are a smaller data center operator or a secondary AI startup lying heavy on project financing or massive amounts of high interest debt, you are in extreme danger. If the AI revenue doesn't materialize exactly as fast as you planned, the interest payments on those bonds will eat your company alive.
It's a question of endurance. Who has the balance sheet to survive the waiting period? Precisely. But I want to pivot to a long-term perspective here, because based on the research we've been looking at, this economic impact of AI has a massive twist coming down the line. Yes, and this is crucial. We are currently living in stage one of the AI timeline, and as we've discussed, stage one is highly inflationary, gobbling up land, power, materials, capital.
The sheer demand drives prices up. But the Minneapolis Fed's analysis suggest that if we look further down the road to stage two, AI could actually trigger massive disinflation. Okay, let's define that mechanism for a second. Disinflation means the rate of inflation slows down, and things actually become vastly cheaper to produce.
How does building insanely expensive data centers lead to things getting cheaper? Think about the fundamental purpose of artificial intelligence. Once these data centers are built and the models are trained, the ultimate goal is to make human labor wildly more productive. If an AI system lets a company accomplish with 10 employees what used to take 100, the cost of operating that business plummets.
Oh, it's like the transition from hand-weaving textiles to the industrial loom. That's a perfect analogy. The initial phase is messy and expensive, right? You have to build the factory, buy the raw steel, secure the coal for the steam engines, which drives up local prices. But once the factory's actually running, it just floods the market with cheap shirts, driving the cost of clothing down for everyone.
That is the exact mechanism. If AI eventually makes complex tasks like legal research, medical diagnostics, routine coding, vastly cheaper and faster, it expands the supply capacity of the entire economy. It drives operating costs down across the board. So, the inflation we are enduring today from building the physical infrastructure might actually be the very thing that cures inflation a decade from now.
It's painful right now for the promise of massive abundance later, but that bridge between the painful construction phase and the abundant future phase is exactly where the friction lies. Uh this raises an important question, though, and it's perhaps the most critical takeaway. During this chaotic, expensive transition period between stage one and stage two, who ultimately pays the bill for this multi-trillion-dollar revolution?
And that is exactly the thought I want you, the listener, to chew on as we wrap up today. Look at your own life, your local community, the economy around you. As this unbelievable multi-trillion-dollar infrastructure is frantically being built across the country, who is really bearing the true cost?
Right. Is it the tech shareholders? Right, the ones getting rich on the stock market. Or is it the bond lenders demanding higher premiums? Or maybe it's the everyday electricity consumer in a place like New Mexico whose monthly power bills might go up simply because a massive AI data center moved in next door and strained the local grid.
Or the final customer who eventually has to buy these AI subscriptions just to stay competitive. Exactly. The next time you see a flashy headline about the newest artificial intelligence model, don't just ask what the technology is capable of doing. Ask yourself, who is financing the concrete and the electricity behind it?
And who is really paying the toll? The financial plumbing beneath the technology is where the real story of the next decade is going to be written. Absolutely. Well, that wraps up our deep dive for today. Don't forget to subscribe and hit that like button.
Comments 0