“Tao Dun Global” Last week, the bond market was turbulent and volatile. On Wednesday (14th), U.S. CPI exceeded expectations, causing government bond yields to surge across the board; on Thursday (15th), data center credit collapsed, corporate bonds were dumped, and CDS (credit default swaps) spiked sharply; on Friday (16th), Federal Reserve Chair Powell stated that economic conditions were “largely in line with expectations,” leading markets to believe the Fed was not in a hurry to raise interest rates, with futures pricing the probability of a rate hike in October dropping to 15%, while the probability for December remained at 90%. Amid concerns over AI company credit risks, the Nasdaq Index recorded its largest weekly decline since “Liberation Day,” while the Dow Jones Index, less affected by AI, rose.
The U.S. dollar index and 2-year Treasury yields retreated, while gold and silver faced selling pressure. A very large crude carrier hit a mine in the Strait of Hormuz, and Houthi militants attacked a Saudi airport, causing oil prices to surge. Trump lifted sanctions on some Russian products in exchange for Russian diesel exports to the U.S. Bond market volatility pushed VIX to the 20 threshold. MSCI may remove Indonesia’s stock market from its emerging markets index, causing the Jakarta Composite Index to plunge 15%. Amid AI-related turmoil, South Korea’s stock market experienced a circuit breaker, and the KOSPI also fell 15%.
U.S. data center CoreWeave (US.CRWV) executives publicly discussed raising their company’s credit rating to investment grade to access “an endless supply of funds” from the bond market. However, the market swiftly slapped them back: last week, CoreWeave’s stock price dropped 23%, its bond yields surged past 15%, its CDS reached 1128 basis points, and markets priced in over a 50% chance of debt default. The bond, with a face value of $1, now trades at around $0.78, already classified as “stressed debt.”
The AI revolution demands massive data infrastructure. Early-mover data center companies, known as NeoClouds (new cloud service providers), positioned themselves in the AI data center space before tech giants entered, securing numerous tenants and locking in long-term contracts. This provides predictable, stable cash flows, with demand expected to outstrip supply over the next five years. However, data centers are capital-intensive and must operate with leverage. During the construction phase, they incur significant capital outlays without corresponding cash flow returns, which is NeoCloud’s fatal flaw.
Tech consultancy Gartner recently published a report forecasting that AI data centers must generate $8.5 trillion in revenue by 2030 to achieve acceptable investment returns. However, businesses and consumers can only afford to pay up to $3.6 trillion. Gartner’s conclusion is that the current burn rate of data centers will create a deficit of at least $4 trillion, which may ultimately need to be absorbed by shareholders or creditors.
Will data centers like CoreWeave default in the short term? The author believes the likelihood is low. Indeed, current bond issuance for AI or data centers lacks sufficient free cash flow support, and these bonds must compete in the primary market with U.S. Treasuries. Unless backed by hyperscalers (ultra-large cloud providers) or giants like Nvidia (US.NVDA), only the most aggressive investors would consider them going forward.
In reality, data centers have circular financing, so there is no imminent default risk. However, if these large bonds require issuing new debt to repay old debt, issuance costs will naturally rise sharply, putting significant funding pressure on some issuers. The author’s view is that most bonds issued for AI capital expenditures have not yet reached renewal stage, and the market is primarily focused on interest payments. The author believes data centers have raised sufficient funds for the next 12 months, so a liquidity crisis is not imminent, though bond prices may remain volatile. U.S. stocks, which have so far been unresponsive to bond market turbulence, may also face shocks, potentially triggering deleveraging.
The author offers another observation: CoreWeave’s customer Anthropic achieved Q3 revenue of $16.8 billion, up 46% quarter-on-quarter. This performance is not bad, but for a company racing toward an IPO with a valuation inflated to $2 trillion, it appears lackluster. More embarrassingly, its operating profit margin is only 11%, far below the S&P 500’s average. Investors in utility stocks need not bear much risk and can achieve higher returns per dollar invested.
Although the author does not expect a wave of defaults in the near term, the point must be emphasized: today’s AI risk is no longer just individual corporate credit risk, but a financial risk. Cross-shareholding, circular ordering, and leveraged operations have tied the entire industry together. The AI sector needs a reshuffle, a Darwinian process of survival of the fittest. By clearing out financially weak players, the AI revolution may proceed more steadily and sustainably.
At this year’s IMF (International Monetary Fund) and World Bank annual meetings in Bangkok, the most discussed topic among attendees was industrialized nations’ budgetary chaos and exploding fiscal deficits. Developed countries’ total bond issuance this year will account for 5.2% of GDP, the highest ratio since the fiscal stimulus triggered by the 2021 pandemic, and extremely rare in modern economic history—and all nations are simultaneously succumbing. The IMF has called on developed nations to take “immediate action” to maintain debt sustainability and financial market stability.
Market concerns over deficit spending are growing, yet governments appear indifferent, causing investors to demand higher premiums to participate in transactions, making government bond auctions more difficult and issuance costs clearly rising. As lower-interest legacy bonds mature, newly issued bonds carry higher rates, further deteriorating fiscal conditions and creating a vicious cycle. Monetary authorities in the U.S., Europe, Japan, and the U.K. have successively restarted their rate-hiking cycles, adding fuel to already rising government bond yields.
Interestingly, despite continuously rising policy rates and issuance costs, and despite Iran-related conflicts triggering soaring energy prices, global GDP growth has shown remarkable resilience, with most countries’ growth figures exceeding expectations. The IMF raised its global growth forecast for this year to 3.1%, thanks to strong AI sector capital expenditures and government fiscal stimulus policies.
Since 2008, consecutive quarterly GDP contractions have been rare, and economic recessions in the textbook sense have disappeared. Business cycles now consist only of upswings, with no downturns. The reason is that governments, driven by electoral interests, resort to deficit spending at every downturn, sacrificing long-term fiscal sustainability for short-term political gains. One side effect is the normalization of quantitative easing, with central banks sacrificing their policy independence.
It’s not entirely fair to say central banks have no independence at all. Central banks, especially the Fed and the Bank of Japan, have restarted rate-hiking cycles despite political pressure from executive branches. However, how long this independence can last remains to be seen. Selling pressure in the long bond market not only pushes up mortgage rates and hits the real economy but has begun threatening financial stability and could even create systemic risks. The author estimates rate hikes will stop by next year, as liquidity will become a policy consideration central banks cannot ignore.
Key events this week: China’s GDP, U.S. consumer confidence, and UK CPI.
Watch now: https://youtu.be/bmSAlFOhQ8E
“Tao Dun”
*This week’s article expresses the author’s understanding and views on the economy, policy, and markets. It represents personal opinion and is not investment advice or solicitation.
*“Economic Information” publishes signed and/or unsigned articles reflecting authors’ personal opinions, which do not represent the position of “Economic Information.” “Economic Information” serves as a platform for free expression.
The U.S. dollar index and 2-year Treasury yields retreated, while gold and silver faced selling pressure. A very large crude carrier hit a mine in the Strait of Hormuz, and Houthi militants attacked a Saudi airport, causing oil prices to surge. Trump lifted sanctions on some Russian products in exchange for Russian diesel exports to the U.S. Bond market volatility pushed VIX to the 20 threshold. MSCI may remove Indonesia’s stock market from its emerging markets index, causing the Jakarta Composite Index to plunge 15%. Amid AI-related turmoil, South Korea’s stock market experienced a circuit breaker, and the KOSPI also fell 15%.
U.S. data center CoreWeave (US.CRWV) executives publicly discussed raising their company’s credit rating to investment grade to access “an endless supply of funds” from the bond market. However, the market swiftly slapped them back: last week, CoreWeave’s stock price dropped 23%, its bond yields surged past 15%, its CDS reached 1128 basis points, and markets priced in over a 50% chance of debt default. The bond, with a face value of $1, now trades at around $0.78, already classified as “stressed debt.”
The AI revolution demands massive data infrastructure. Early-mover data center companies, known as NeoClouds (new cloud service providers), positioned themselves in the AI data center space before tech giants entered, securing numerous tenants and locking in long-term contracts. This provides predictable, stable cash flows, with demand expected to outstrip supply over the next five years. However, data centers are capital-intensive and must operate with leverage. During the construction phase, they incur significant capital outlays without corresponding cash flow returns, which is NeoCloud’s fatal flaw.
Tech consultancy Gartner recently published a report forecasting that AI data centers must generate $8.5 trillion in revenue by 2030 to achieve acceptable investment returns. However, businesses and consumers can only afford to pay up to $3.6 trillion. Gartner’s conclusion is that the current burn rate of data centers will create a deficit of at least $4 trillion, which may ultimately need to be absorbed by shareholders or creditors.
Will data centers like CoreWeave default in the short term? The author believes the likelihood is low. Indeed, current bond issuance for AI or data centers lacks sufficient free cash flow support, and these bonds must compete in the primary market with U.S. Treasuries. Unless backed by hyperscalers (ultra-large cloud providers) or giants like Nvidia (US.NVDA), only the most aggressive investors would consider them going forward.
In reality, data centers have circular financing, so there is no imminent default risk. However, if these large bonds require issuing new debt to repay old debt, issuance costs will naturally rise sharply, putting significant funding pressure on some issuers. The author’s view is that most bonds issued for AI capital expenditures have not yet reached renewal stage, and the market is primarily focused on interest payments. The author believes data centers have raised sufficient funds for the next 12 months, so a liquidity crisis is not imminent, though bond prices may remain volatile. U.S. stocks, which have so far been unresponsive to bond market turbulence, may also face shocks, potentially triggering deleveraging.
The author offers another observation: CoreWeave’s customer Anthropic achieved Q3 revenue of $16.8 billion, up 46% quarter-on-quarter. This performance is not bad, but for a company racing toward an IPO with a valuation inflated to $2 trillion, it appears lackluster. More embarrassingly, its operating profit margin is only 11%, far below the S&P 500’s average. Investors in utility stocks need not bear much risk and can achieve higher returns per dollar invested.
Although the author does not expect a wave of defaults in the near term, the point must be emphasized: today’s AI risk is no longer just individual corporate credit risk, but a financial risk. Cross-shareholding, circular ordering, and leveraged operations have tied the entire industry together. The AI sector needs a reshuffle, a Darwinian process of survival of the fittest. By clearing out financially weak players, the AI revolution may proceed more steadily and sustainably.
At this year’s IMF (International Monetary Fund) and World Bank annual meetings in Bangkok, the most discussed topic among attendees was industrialized nations’ budgetary chaos and exploding fiscal deficits. Developed countries’ total bond issuance this year will account for 5.2% of GDP, the highest ratio since the fiscal stimulus triggered by the 2021 pandemic, and extremely rare in modern economic history—and all nations are simultaneously succumbing. The IMF has called on developed nations to take “immediate action” to maintain debt sustainability and financial market stability.
Market concerns over deficit spending are growing, yet governments appear indifferent, causing investors to demand higher premiums to participate in transactions, making government bond auctions more difficult and issuance costs clearly rising. As lower-interest legacy bonds mature, newly issued bonds carry higher rates, further deteriorating fiscal conditions and creating a vicious cycle. Monetary authorities in the U.S., Europe, Japan, and the U.K. have successively restarted their rate-hiking cycles, adding fuel to already rising government bond yields.
Interestingly, despite continuously rising policy rates and issuance costs, and despite Iran-related conflicts triggering soaring energy prices, global GDP growth has shown remarkable resilience, with most countries’ growth figures exceeding expectations. The IMF raised its global growth forecast for this year to 3.1%, thanks to strong AI sector capital expenditures and government fiscal stimulus policies.
Since 2008, consecutive quarterly GDP contractions have been rare, and economic recessions in the textbook sense have disappeared. Business cycles now consist only of upswings, with no downturns. The reason is that governments, driven by electoral interests, resort to deficit spending at every downturn, sacrificing long-term fiscal sustainability for short-term political gains. One side effect is the normalization of quantitative easing, with central banks sacrificing their policy independence.
It’s not entirely fair to say central banks have no independence at all. Central banks, especially the Fed and the Bank of Japan, have restarted rate-hiking cycles despite political pressure from executive branches. However, how long this independence can last remains to be seen. Selling pressure in the long bond market not only pushes up mortgage rates and hits the real economy but has begun threatening financial stability and could even create systemic risks. The author estimates rate hikes will stop by next year, as liquidity will become a policy consideration central banks cannot ignore.
Key events this week: China’s GDP, U.S. consumer confidence, and UK CPI.
Watch now: https://youtu.be/bmSAlFOhQ8E
“Tao Dun”
*This week’s article expresses the author’s understanding and views on the economy, policy, and markets. It represents personal opinion and is not investment advice or solicitation.
*“Economic Information” publishes signed and/or unsigned articles reflecting authors’ personal opinions, which do not represent the position of “Economic Information.” “Economic Information” serves as a platform for free expression.
