As government debt becomes increasingly volatile, global capital is viewing the bonds of Microsoft and Alphabet as new safe havens – even though rising yields are threatening a broader range of AI transactions.
Major stock indices performed strongly at the close on Friday, recording their fourth consecutive week of gains, while tech stocks rebounded from the sharp declines seen in the middle of this week. The S&P 500 Index ( SPX ) rose 1.15% this week, while the Nasdaq 100 Index ( NDX ) and the Dow Jones Industrial Average ( DJIA ) rose 0.24% and 0.93% respectively.
Large tech stock benchmark indices set records on Monday and Tuesday, but then tumbled significantly over the following two days due to rising oil prices and concerns about a global bond sell-off; a media report sparked new revenue concerns, adding fuel to the fire. The Financial Times reported that OpenAI's annualized revenue run rate as of the end of September was close to $50 billion, about $20 billion lower than the approximately $70 billion previously circulated among investors. The market saw this as a sign that AI demand might slow down, triggering a sharp decline in tech and semiconductor stocks.
Fortunately, the next day the market realized that the real issue was merely a difference in accounting methods, and not a slowdown in the underlying demand or growth of OpenAI. The developer of ChatGPT disclosed the net amount after deducting commissions to cloud partners, while some comparisons with competitor Anthropic used the total amount. In addition, OpenAI indicated that it expects its annualized revenue run rate to reach at least $70 billion by the end of this year, driven particularly by strong growth in its enterprise business; the annualized run rate of its enterprise business in the third quarter is said to have exceeded 100%.
Despite this clarification, the rise of the Nasdaq 100 index on Friday was still somewhat hindered by the selling of telecommunications stocks, including T-Mobile ( TMUS ) and AT &T (T). After SpaceX ( SPCX ) announced a deal to acquire a nationwide low-frequency spectrum portfolio, these and related stocks tumbled significantly, as the market feared that Starlink Mobile might become a more formidable competitor to traditional wireless operators.
Wider macroeconomic and market trends have also supported a rebound in tech stocks. Investors interpreted the minutes of the Federal Reserve meeting as indicating that there are no immediate plans for interest rate hikes by the central bank, with the decision-making time being pushed back to the end of the year. In the current market sentiment, this is almost equivalent to never. Meanwhile, as concerns over the escalating situation in Iran eased, and Trump announced a diesel supply agreement with Russia, oil prices fell slightly. The yield on 10-year U.S. Treasury bonds stabilized after reaching a 24-year high on Wednesday. Despite pressure on global bond markets and generally significant increases in yields across major economies, the U.S. stock market is willing to ignore this brewing crisis, as the AI narrative remains intact.
Judging from the forecasts, the economy of AI seems to be still operating smoothly. The latest estimates from FactSet indicate that the earnings growth rate of companies in the S&P 500 is expected to exceed 35% in the third quarter. Among them, the energy sector is expected to lead the gains, but this is mainly due to a lower year-on-year base. At the same time, the sectors that led growth in the second-quarter reports are accelerating; the technology sector is expected to see earnings soar by more than 65%, and the communication services sector, led by Alphabet ( GOOGL ), is expected to see a growth rate of over 50%.
Overall, financial reports are key to continuing the AI trend, and unless the bond market continues to slide into crisis, they are likely to overshadow other intraday market fluctuations. Although a collapse similar to the global financial crisis (GFC) is not a high-probability scenario, a "normal" crisis—i.e., one where yields continue to rise due to weakened confidence in government bonds—poses a serious threat to the stock market, and AI trading is one of the more vulnerable areas within it.
For at least a certain period of time, the ultra-large-scale cloud service provider ( hyperscalers ) should be able to hold on, but other parts of the AI ecosystem will suffer impacts, which will eventually backfire on cloud computing suppliers. Unlike many similar situations in the past, where the threat came from a sudden deterioration in the confidence of corporate issuers, leading to a surge in relative government bond yield spreads, the current threat lies in the absolute cost of debt. Investment-grade corporate bonds currently have an average yield spread of 0.82% over U.S. Treasuries, whereas during the peak of market panic due to the COVID-19 pandemic in 2020, this figure was around 3.5%. Microsoft ( MSFT ), the U.S. non-financial company with the highest rating and the strongest fundamentals, pays a very small yield spread over U.S. Treasuries – yet even so, this increase is still significant given that U.S. Treasury yields are at their highest levels in over two decades.
Ultra-large cloud service providers and leading enterprises such as AI have issued a large amount of debt this year and plan to further finance. This is because even the substantial cash flows of Alphabet or Amazon (AMZN) are insufficient to support the astronomical costs associated with this world-changing technology. Estimates for 2027 indicate that AI capital expenditures will reach the trillion-dollar range, while a recent NBER report states that by 2032, the total cost of AI construction in the United States will amount to 9 trillion dollars.
At the same time, governments around the world also need to sell debt to finance their growing deficits, which has pushed up the ratio of debt to GDP. In the current environment, especially in regions outside of the United States where economic growth expectations have been lowered while inflation rates remain high, investors demand higher yields before they are willing to lend money to governments. This trust is further compromised when investors see a much healthier alternative option.
Indeed, why would investors buy bonds of countries that are in a long-term political crisis, have an economic growth rate close to zero, and whose debt levels are equivalent to or even greater than their economic size, with no visible path to repayment – such as France’s OAT or UK’s Bond Market Paper ( Gilts ) – just to obtain a yield of 4.7% to 5.6%? They could easily achieve similar returns on Microsoft’s or Alphabet’s bonds. These companies have excellent fundamentals, rapid growth, higher credit ratings than these governments, a lower debt-to-company value ratio, and a clear path to investment returns.
To some extent, helping American tech giants to raise funds for AI has become a safe haven for avoiding government debt risks. Global capital is rationally viewing the strongest US tech balance sheets as higher-quality value storage tools, as well as carriers for participating in the investment cycle of AI.
