The US 30-year Treasury yield rose 6 basis points to 5.31%, reaching its highest level since 2007, while the 10-year yield is around 4.72%. What's particularly noteworthy is that the US isn't alone. France's 30-year yield is at an 18-year high of 4.86%, and Japan's is at 4.135%. Canada is testing levels not seen since 2010, and Germany since 2011.

This simultaneity is no coincidence. Long-term bond supply is increasing as government deficits grow, with the annual deficit in the US hovering around $2 trillion. Inflation has been above the Fed's target for five years, and investors are demanding higher returns each year they commit to long-term investments. This is compounded by company issuances financing AI investments. In June, foreign holdings of US Treasury bonds decreased, primarily in Japan and China.

The second line of focus is energy. Brent is at its highest level since July 30th at $91.14, while WTI is at $85.04. Iran's hardening tone and the lifting of the ceasefire agreement have reignited supply uncertainty. The perception of energy costs as a permanent rather than a temporary fluctuation directly fuels the term premium in long-term interest rates. The two stories are essentially the same.

There is no consensus. Barclays believes the sell-off in long-term bonds is not temporary. Yardeni Research, however, sees no reason for panic yet and argues that the 10-year yield can remain in the 4-5% range without harming the economy. However, the yield is currently at its upper limit. The $16 billion 20-year auction is the closest test of this problem. Last week's $25 billion 30-year auction resulted in a yield of 5.216%, the highest level since 2001. If the global discount rate is rising, this affects both equity valuations and the flow of capital into emerging market assets.

Lately, 10- and 30-year bond interest rates are a hot topic among stock investors. You might say, "I'm not a bond investor, so what does this have to do with me?", but that's not entirely true. Especially if you're investing in artificial intelligence, the bond side is very relevant. There are several reasons for this, but I'll only focus on the part that concerns AI infrastructure companies. Look closely at the image below. The image tells us this:

It shows the expenses of companies with high capex spending, such as $META, $GOOGL, $AMZN, $MSFT, which don't immediately appear on their balance sheets. AI infrastructure companies are investing excessively, and they don't have enough cash flow (FCF) to cover these investments. Therefore, these companies choose to borrow, and this choice isn't fully reflected on their balance sheets.

There's high demand for Google, Meta, and Amazon's borrowing needs. There are several reasons for this. Firstly, these bonds offer interest rates 1-1.5% higher than current US Treasury bonds. Therefore, instead of lending to the US at 4.7%, bond investors prefer to lend to what we call hyperscalers, which offer yields between 5.5% and 6%. As a result, US companies are essentially competing directly with US bonds when it comes to spending. Be aware of this.

Doesn't this bother the government? Actually, it does, but there's also the reality that a large portion of US growth comes from AI spending. Therefore, the government prioritizes growth more than the rise in bond yields. Is this bad? Actually, no, what makes the US government strong is its strong technology companies in the bond market. Many might think of the petrodollar system, but the US knows it's no longer sufficient. It now believes that AI will be more powerful than this petrodollar system, which is very true. If you sell technology to the world, your currency will remain strong, and your country will continue to attract investment and capital. Notice that recently, countries with AI companies have stronger growth and attract more capital. Therefore, money doesn't flow to developing countries as easily as before; it's becoming much more selective.

According to Morgan Stanley, US companies will invest $1.4 trillion by 2027. While they have very strong backlog figures, these figures don't match the $1.4 trillion target. This is because companies first establish their chip and energy infrastructure. Once they've established this, they invest first, providing the money, but they receive a return on that investment over time. This is the simplest accounting rule. In fact, companies, especially hyperscalers, prefer operational expenses (opex) over capital expenditures (capex) on their balance sheets. Companies like $NVDA, $AVGO, $AMD, aware of their capital needs, are trying to support companies that need capex through both borrowing and investment, which you may have read about last week. Is this bad? In my opinion, no, but I know many people dislike it.

Also, I think one of the biggest reasons the Fed doesn't want to raise interest rates is this: if the Fed raises interest rates while these companies have capital needs, the present value of their future cash flows will decrease. Therefore, the valuations of technology stocks with high growth expectations are suppressed. The US administration would be shooting itself in the foot. China is doing the exact opposite, offering more flexible loans or incentives to companies that invest.

Another point is that the more returns the US markets provide, the more investors will come. The simplest example of this is individual or institutional investors who invest in the funding of companies that sell ATM shares, buying their shares. Many growth companies, such as $CRWV, $ORCL, have high CDS. This is the easiest way for them to access capital. I would like to elaborate further, but I don't want to confuse things too much. Artificial intelligence will definitely reach great heights in the long term; but until the infrastructure is established, companies will use every legal means available. I think we will continue on our way, with ups and downs.