Fearless of AI's huge spending concerns! Meta's bond issuance received a record oversubscription of $125 billion

According to Zhitong Finance APP, Meta Platforms (META.US) received record strong demand in its bond issuance on Thursday, despite a significant drop in its stock price, indicating that bond investors were not swayed by its artificial intelligence (AI) spending plans.

It is reported that Meta issued a total of $30 billion in bonds, marking the largest sale of high-rated bonds in the U.S. since 2023, attracting subscription demand as high as $125 billion, setting a record for public company bond issuance history. This is also Meta's second issuance since it first entered the bond market in 2022, when the company issued $10 billion in bonds. The scale of this financing has significantly expanded, reflecting the ongoing capital investment needs in the AI infrastructure sector.

Meta CEO Mark Zuckerberg previously stated that the company plans to spend hundreds of billions of dollars over the next decade to build data centers and other AI infrastructure, aiming to achieve near-human-level AI capabilities and integrate them into its products, such as Facebook and Instagram. Meta stated in its earnings report on Wednesday that capital expenditures this year will reach up to $72 billion, and capital expenditures in 2026 will be "significantly higher" than in 2025.

For Meta, at least part of its capital expenditures will be driven by borrowing. Currently, U.S. corporate bond investors are eagerly looking to lend to Meta. This demand partly stems from investors pouring funds into short- and medium-term high-rated bond funds for 25 consecutive weeks to lock in yields and prevent further declines. According to LSEG Lipper data, this is the longest inflow cycle in four years. Meanwhile, most corporate bond issuances this year have been for refinancing existing debt rather than new net borrowing.

Robert Cohen, head of global developed market credit at DoubleLine Capital, stated that this situation has made investors particularly eager for new bonds. Cohen noted that the timing of this round of AI-related borrowing is "just right," and "as long as the transaction structure is reasonable, the capital markets will be happy to finance these projects."

The differing reactions of the stock and bond markets can be attributed in part to their focus on different aspects of the company's earnings report. Meta's stock price fell on Thursday partly because the company disclosed in its earnings report that it recorded a one-time non-cash income tax expense of $15.93 billion in its third-quarter income tax expenses due to the implementation of the "Big and Beautiful Act," leading to a year-on-year net profit drop of 83% to $2.709 billion in the third quarter. In response, Interactive Brokers chief strategist Steve Sosnick stated that this impact is not significant enough to affect the company's credit status.

On the other hand, stock investors remain skeptical about whether Meta can significantly enhance the profitability of its advertising business through AI—especially given the company's plans to further increase spending on data centers and chips next year. However, the earnings reports from competitors Alphabet (GOOGL.US) and Microsoft (MSFT.US) indicate that overall demand for data centers is still far from being met Microsoft's order backlog for commercial customers has reached $392 billion (including some non-cloud expenditures), while Alphabet's order backlog stands at $155 billion—almost double what it was 18 months ago.

Meanwhile, Meta's cash flow from operating activities in the third quarter reached $30 billion. For bond investors, the primary concern is whether they can recover their principal and interest in full and on time, making Meta's bond issuance "extremely attractive." Steve Sosnick stated, "Meta continues to demonstrate that it has sufficient profitability to generate cash flow to meet the primary goal of bond investors—recovering their investments on time."

Winners and Losers

Tech companies have been borrowing heavily to meet AI-related demands. Data shows that as of the end of September, tech companies in the U.S. public bond market have borrowed approximately $157 billion. More debt is on the way. According to Morgan Stanley, by the end of 2028, large tech companies are expected to invest about $3 trillion in infrastructure, including data centers.

Thomas Murphy, a portfolio manager at Threadneedle Investments, stated, "The scale of these numbers and their nearly exponential growth rate indeed warrant a rethinking of the logic behind the entire issue."

Oracle (ORCL.US) issued $18 billion in investment-grade bonds last month, while this week the banking sector also launched a total of $38 billion in bond issuance to fund data center projects related to the company. However, Oracle's cash flow turned negative for the first time since 1992 this year, and analysts expect this metric to continue to deteriorate in the coming years, only returning to positive territory by 2029. Bond investors have begun purchasing protection contracts against the company's default risk, and Morgan Stanley believes this trend will continue in the short term.

Robert Cohen stated that he has been buying new AI-related bonds, but on a smaller scale, preferring trades backed by high-rated issuers. He tends to choose projects with some form of guarantee, such as lease payment guarantees, and is cautious about high-risk "junk-rated AI bonds," as the scale of demand for AI computing power remains uncertain in the coming years. Robert Cohen added, "The devil is in the details. There will be good projects and bad projects. We are very cautious about this because we know that not all AI projects will succeed."