원문 보기
Michael Burry has just positioned himself against one of the hottest sectors in the US stock market. This is the same investor who predicted the 2008 financial crisis, earning $100 million for himself and $725 million for investors by shorting the housing market. Over the past decade, his hedge fund Scion Asset Management has returned around 260%.
This equates to a 50% outperformance relative to the broader US stock market and over 200% compared to the average hedge fund. So, Burry seems to have a good track record of identifying market dislocations early. Today, Michael Burry has taken a short position against one of the most important sectors in the US market, semiconductors.
This is the same sector that's risen around 60% in just the past 2 months and over 225% in the past year, making it one of the strongest performing parts of the market. Burry has purchased put options on the semiconductor ETF with a strike price of $330 expiring in January 2027. With this ETF at $560, he's essentially positioning for at least a 41% decline in semiconductor stocks over the next 8 months.
His view is that the recent rally reflects speculative sentiment more than it does sustainable underlying fundamentals. And if he's proven right, there could be broader financial and economic consequences. Semiconductors now make up 14% of the entire US stock market. This is twice the market share the sector held at the peak of the 2000.com bubble.
Back then, what followed was an 85% decline in technology stocks coinciding with the broader economic recession. But a high concentration alone doesn't cause such an outcome. It typically takes a catalyst to set things off. If we replace this line with the price of oil, we can see that historically that catalyst has been oil.
The downturns in 1990, 2000, and 2008 were all preceded by a sharp spike in oil prices. After all, oil is the base input for almost everything in the economy. When it gets more expensive, the cost of producing and transporting goods rises across the board, pushing inflation higher. This typically leads the Fed to raise interest rates, making borrowing more expensive.
As a result, economic growth tends to slow, and sectors that have been priced for perfection tend to be the first to crack. In the past 2 months alone, oil prices have risen by over 60% and with the geopolitical tensions in the Middle East still elevated, oil has shown a limited signs of easing. Barrie's thesis seems to be that the current energy shock could be the catalyst that exposes the potential cracks in the semiconductor story.
To understand why, we need to look at what's actually been driving semiconductor demand. Data centers have become the single biggest source of semiconductor demand today. The direction of data center spending effectively shapes the foundation for the semiconductor sector. In 2022, when ChatGPT was released, data center spending stood at $1 billion per month.
Today, it's risen to $3.5 billion. This is now surpassed the spending on office buildings for the first time in history. Yet, despite that record level of spending, almost half of the data centers that were supposed to open this year have been delayed or canceled. One possible explanation can be seen through this chart showing the average price of electricity in the US.
This has risen from 6 cents per kilowatt hour in 1980 to nearly 20 cents today. But, the sharpest part of that rise has come in the past few years, and with the ongoing energy shock, the upward pressure may persist. Microsoft's CEO recently acknowledged that the company doesn't have enough electricity to install all its semiconductor inventory.
And even data centers in California, Nvidia's hometown, are facing a power shortage. The problem here is that data centers that struggle to get power are likely to consume fewer semiconductors. And the potential scale of this problem becomes clear when we look at how much electricity data centers consume.
Today, global data centers use approximately 450 terawatt hours of electricity per year, with half of it coming from the US alone. To put that in perspective, that's more than the electricity consumption of most countries, including Australia, Taiwan, and Italy. In fact, only 16 countries consume more electricity than all data centers combined.
And if we fast forward to 2030, these data centers are projected to more than double their electricity usage to 945 terawatt hours. This would be equivalent to powering all of Russia, the fourth largest electricity consumer in the world. Now, the reason this number keeps climbing is not just because more data centers are being built, but also because of a physics problem that a lot of investors are not aware of.
Semiconductors that are put inside chips use copper wires to carry data between each other. Copper comes with a fundamental physical limitation. The faster you push data through it, the more heat it generates that degrades the quality of the data itself. Jensen Huang, the CEO of Nvidia, stated that using copper on Nvidia systems wasted 20 kilowatts out of every 120 kilowatt system, primarily through heat.
All of this heat essentially represents wasted electricity, but more importantly, it needs to be managed. That's why most data centers rely on heavy cooling infrastructure running around the clock. This consumes additional electricity on top of the compute itself. To put this in a perspective, servers, networking, and storage only account for 57% of a data center's total operating costs.
The remaining 43% is attributed to cooling and power systems. So, at first glance, it might seem like Burry's position is primed to pay off. We're seeing record electricity costs that could negatively impact one of the most power intensive sectors in the market, but this is where the story gets interesting.
If the energy shock was hurting semiconductors, you would expect to see it show up in the earnings. Instead, the sector's earnings have just climbed to record levels, and we see a similar story through the profit margins, which have also reached a new high of 48%. So, neither earnings nor margins have rolled over despite oil being 60% higher.
It's been quite the opposite. This is very different to what we saw in 2022, for example, when Russia invaded triggering an energy shock. Back then, earnings and margins compressed for multiple months following the shock. What makes this even more interesting is that semiconductor valuations have actually gotten cheaper.
The sector's PE ratio, a measure of how expensive stocks are relative to their earnings, sits at some of the lowest levels since 2022. This is not because semiconductor stocks fell, but because earnings grew faster than their stock prices. So, Burry seems to be anticipating a turn in these fundamentals, likely triggered by the energy shock.
Now, to evaluate the probability of him being right, we need to understand what's been driving this earnings resilience. Believe it or not, a lot of this comes down to one thing, light. More specifically, a technology called photonics. This might sound complicated, but it's something you're likely already familiar with.
The reason all of our home internet got so fast is because copper lines were replaced with fiber optic cables that carry light instead of electrons. Light uses a fraction of the power, generates almost no heat, and carries more data. Research suggests that light consumes up to 80% less energy compared to traditional electronic equivalents.
This technology is now being applied at the semiconductor chip level, with Nvidia recently committing $4 billion in total to two photonic manufacturers. The pace of this transition will likely determine whether Burry's thesis plays out or not. For example, Lumentum, one of the two companies Nvidia invested in, has already reported that its existing manufacturing capacity is effectively sold out through the end of 2028.
Its backlog has surpassed $400 million dollars, with the majority of shipments not expected until the second half of 2026. So, if these companies are unable to scale fast enough and the energy shock persists, Burry's thesis could play out. In such a scenario, semiconductors could become a meaningful drag on the broader stock market and economy.
But, if these companies are able to continue expanding and delivering on their backlogs, a lot of the energy problem that Burry seems to be counting on could be addressed. In this type of a scenario, semiconductors could arguably become an even larger part of the market than they already are. At capital.com, we'll be closely tracking the semiconductor sector and its photonics story. Hit like and subscribe to follow our market updates. Thank you for watching.
Comments 0