A key concern among investors is that revenue growth may not keep pace with the scale of recent investments. Tech analyst Gartner predicts that 80% of neocloud providers will “cease independent operations” by 2030, mainly because of a massive gap between capital expenditures and the likely return on those expenditures. Moreover, Gartner said that AI data center owners will need to generate US$8.5 trillion in revenue by 2030 to cover the cost of capex as well as generate a good return on their investments. It predicts, however, that such owners will generate revenue of “no more than US$3.6 trillion” by 2030—in other words, there is a nearly US$5 trillion gap. Gartner concluded that “a continued AI infrastructure investment trajectory could leave more than US$4 trillion in asset value exposed to write-down risk through 2032.”
Investors appear anxious enough about holding AI-related debt that the yield on US Treasury securities appears to have fallen as investors dumped AI bonds. The yield on the 10-year US government bond fell sharply to its lowest level in three weeks, although it remains relatively elevated. The same was true for bonds issued by the German government, which were often seen as a very safe asset. It could be that bond investors are concerned that a weakness of tech-company bonds might be the first sign of broader economic risks.
Investor caution is not limited to bonds. It is also affecting the equity prices of companies that produce AI-related products, especially chips. For example, one of the indices of semiconductor equities fell 4.3% yesterday while the overall S&P 500 index was down by a more modest 1.1%.
And the issue is not only insufficient demand for AI services. The issue is also whether companies that are building AI infrastructure will have sufficient electricity to supply services. There are reports of infrastructure sitting idle due to lack of electricity. Plus, Morgan Stanley estimated that, between 2026 and 2028, more than half of new AI servers might not have powered infrastructure into which they can be placed. They estimated that new infrastructure will require 68 gigawatts of new electricity, but that only 30 gigawatts were accounted for by existing projects. The rest remains uncertain.
Some infrastructure sites are developing their own power off the grid using natural gas. However, it is reported that there is a shortage of turbines to provide the necessary power. China has increased its electricity-generation capacity, which may have contributed to the competitiveness of Chinese AI services.
Their principal insight was that economic policy decisions alone are not the only factor driving the behavior of millions of households and businesses. Rather, households are influenced by expectations about policy. For example, if participants in the market economy expect a tightening of monetary policy and, consequently, higher long-term borrowing costs, it will influence their behavior. This means that policy will be most effective when future shifts in policy are publicly signaled.
Indeed, the insights they offered played a role in leading major central banks to engage in forward guidance during the past quarter-century. That is, central banks have become accustomed to signaling their intentions and the likely path of monetary policy. Moreover, Professor Woodford’s 2003 book, Interest and Prices, became something of a bible for central bankers.
The prize comes as the Federal Reserve, under new leadership, has changed its approach to communicating policy decisions: Indeed, when the Fed raised the benchmark interest rate last month, it did not provide an explanation of its thinking until the publication of the minutes of the meeting three weeks later. This is unprecedented. Moreover, it left market participants uncertain about the Fed’s future intentions. According to the framework developed by Hall and Woodford, reduced policy signaling could hamper the effective transmission of monetary policy.
In any event, the US government reported that consumer-price inflation accelerated in September. Consumer prices were up 3.7% in September versus a year earlier—the highest inflation rate since May. Prices were up 0.6% from August to September—the biggest monthly gain since April. Energy played a key role: In September, energy prices were up 20.4% from a year earlier and up 5.2% from the previous month. The price of gasoline was up 34.2% from a year earlier and up 9.1% from the previous month.
When volatile energy and food prices are excluded, core prices were up 2.6% from a year earlier—the highest increase since June. Core prices were up 0.3% from the previous month. The monthly increase has not been higher since April. As such, underlying inflation is starting to accelerate but remains relatively tame. Still, some nonenergy prices are certainly influenced by energy prices. For example, airline fares were up 22.4% from a year earlier.
Meanwhile, although energy has played the primary role in driving accelerated inflation, the massive investment in AI has also contributed to inflation. The rollout of data centers has tightened the availability of information technology inputs. Consequently, the latest inflation report shows that the prices of computers and peripherals were up 9.7% in September versus a year earlier. Plus, the prices of software and accessories were up 28.4% from a year earlier. On the other hand, the prices of smartphones continue to recede, down 9.9% in September from a year earlier.
Going forward, there are reasons to expect a further acceleration in US inflation, regardless of whether oil prices rise further. This is because the elevated price of diesel fuel, now at a record high, may take time to work its way into the prices of goods being delivered by trucks (lorries). This will be especially true of food. In September, food prices were up 2.5% from a year earlier. This is now expected to likely accelerate in the months to come.
What are the implications of rising inflation? First, we know that, for the past several months, wage increases have not kept pace with price increases. This means that households’ true purchasing power is declining, which, in turn, puts pressure on financially stressed households. They reduce the share of income they save, dip into savings, take on more debt, shift to lower-priced products or services, and/or postpone some purchases. This also has an impact on consumer sentiment and may likely influence the upcoming mid-term elections.
Second, accelerated inflation changes expectations for monetary policy: The futures market shows that, although investors still expect the Federal Reserve to not change the benchmark interest rate this month, the probability has shifted considerably. For example, a week before writing, there was an implied probability of 82.3% that the Fed will leave the rate unchanged later this month. Yesterday, the probability was 77.9%, which was not a big change. But after the release of the inflation report, the probability was 66.8%. Plus, the implied probability of two rate hikes before the end of the year increased from 18.7% yesterday to 28.7% today. Meanwhile, following the inflation report, bond yields rose while equity prices declined.
In any event, China’s exports, denominated in US dollars, were up 36.3% in September versus a year earlier. This was the fastest growth rate since February and the second-fastest rate of increase since February 2021 when the Chinese economy was starting to emerge from the COVID-19 pandemic.
As indicated above, this growth was fueled, in part, by AI. For example, exports of semiconductors and automatic data-processing equipment accounted for 17.9% of total exports in September. Yet, at the same time, these two categories accounted for 36% of the growth of exports in September versus a year earlier. In addition, it is reported that exports of semiconductors were up 136% from a year earlier. However, when adjusted for prices, such exports were up only 1%. This means that almost all of the increase in export revenue from semiconductors was due to rising prices.
Strong export growth was not entirely due to AI. There was also a significant increase in exports of automobiles, which were up 62% from a year earlier. Automotive trade has been a point of political tension with other countries, especially the European Union. As such, the European Union and China just signed a deal to impose quantitative restrictions on the movement of hybrid vehicles from China to the European Union. Within Europe, but especially within Germany, there have been concerns that China’s growing strength in the automotive sector could pose a significant challenge to Europe’s automotive industry—one that directly and indirectly employs millions. Perhaps it is too early to make such an assessment given Chinese branded vehicles still only account for 13% of EU sales.
China’s exports to the United States were up 39.6% in September versus a year earlier. This news did not appear to set off new trade tensions because the leaders of the two countries met recently and agreed to roll back tariffs. Exports to the European Union were up 25.3% in September versus a year earlier. Also, exports to Southeast Asia were up 52.3%.
Also, Chinese imports (denominated in US dollars) were up 31% in September versus a year earlier, hitting the highest level on record. The strength of imports was due to strong demand for industrial raw materials, minerals (including oil), and electronic inputs. This demand was largely driven by the needs of exporters rather than being related to domestic demand, which remains comparatively subdued.
Finally, it should be noted that the sharp rise in both exports and imports is not simply due to surging volume demand. It is partly due to rising prices of commodities and electronic inputs. This surge, in turn, is partly fueled by strong AI-related demand. It is also fueled by the conflict in the Middle East, which has led to shortages of key mineral commodities.
In the United States, August was a good month for consumer spending, despite no growth in real income. Specifically, the US government reported that, in August, real (inflation-adjusted) disposable personal income (income after taxes) was unchanged from the previous month.
Yet, real personal consumption expenditures were up 0.6% from the previous month. This took place because the personal saving rate dropped from 4.6% in July to 4.1% in August; as recently as January, it was 5.6%. Households are keen on maintaining their standard of living in the face of declining real wages. Yet, reduced savings can only go so far. If real incomes continue to stagnate, eventually it may have a negative impact on consumer spending.
Meanwhile, the US government also reported on the personal consumption expenditure deflator (PCE-deflator)—the Fed’s principal measure of inflation—which was up 3.4% in August versus a year earlier. This was the same level as July and lower than the three preceding months. Plus, the PCE-deflator was up 0.3% from July to August—the biggest monthly gain since May. When volatile food and energy prices are excluded, the core PCE-deflator was up 3% in August versus a year earlier—the same as in June and July and lower than in the previous three months. Core prices were up 0.3% from July to August, which was the biggest gain since April.
This report suggests that inflation is not accelerating, despite a recent jump in oil prices, new tariffs, and strong demand from the technology sector. This news pleased investors, who then revised their expectations for monetary policy. The futures market’s implied probability that the Fed will raise its benchmark interest rate next month fell. The probability of two rate hikes before the end of this year fell as well.
Despite the favorable news on inflation and a reduced expectation for monetary tightening, the yield on the US Treasury’s 10-year bond continued to rise. The yield passed 5.3% today for the first time since 2002. It is likely that the rise in the bond yield meant that investors were more focused on the state of the real economy rather than inflation or Fed policy. Indeed, there was positive news regarding the strength of the economy. The US government released a revised estimate of second-quarter GDP. Previously, the government estimated that real GDP grew at an annualized rate of 1.5% in the second quarter. Now, it says real GDP grew at a rate of 2.2%.
Moreover, growth in consumer spending was revised from 3.4% to 3.8%. And if imports and exports, inventories, and government spending are excluded, the economy grew at a rate of 4.6%—revised upward from 4.2%. This is often considered a good measure of underlying domestic private sector demand. As such, investors saw an economy that appears to be exceptionally strong in the second quarter.
Plus, based on recent purchasing managers’ indices, the economy might have been strong in the third quarter. This may help to explain rising bond yields. On the other hand, if bond yields continue rising, it could have a negative impact on credit-market activity, especially in the housing market, which, in turn, could ultimately cause an economic slowdown.
On the other hand, the US government reported that employment growth was surprisingly weak in September. This, in turn, led investors to further downgrade their expectations for monetary tightening. Of course, one month does not make a trend. Yet, investors saw the September report as potentially signaling a weakening of the economic picture and, consequently, weaker inflation. Let’s look at the details:
The US government publishes a monthly employment report that includes the results of two surveys: One is a survey of households; the other is a survey of establishments. The latter determined that, in September, only 29,000 new jobs were created, down from 133,000 in August. Moreover, employment fell in July and grew only modestly in June. Thus, employment grew strongly in only one of the last four months, which could signal weakness.
Moreover, in September, employment fell in financial services, professional and business services, information, mining, and all levels of government (federal, state, and local). Employment grew by 23,000 in healthcare and social assistance. Plus, employment grew by 10,000 in leisure and hospitality. Thus, excluding these two categories, employment declined modestly in September.
The establishment survey also includes data on average hourly earnings. This indicator was up 3% in September versus a year earlier—the smallest increase since May 2021. Moreover, with inflation running well above 3%, real (inflation-adjusted) earnings are declining. This could weigh on consumer spending. In recent months, spending has grown faster than income because households have been cutting back on saving. This cannot go on indefinitely.
The separate survey of households, which includes self-employment, found that, in September, the labor force (those either working or seeking work) grew significantly faster than the working-age population, which means labor-force participation increased. On the other hand, employment grew more slowly than the labor force, leading the unemployment rate to rise from 4.1% in August to 4.2% in September. Still, the unemployment rate remains very low.
This jobs report, which suggests the possibility of weakness, led investors to revise their expectations for Fed policy. Specifically, the futures market’s implied probability that the Fed will keep the benchmark interest rate unchanged later this month, increased. Indeed, this view was echoed in comments by two Fed leaders: Philip Jefferson, member of the Federal Reserve Board, and John Williams, president of the New York Fed.
Jefferson said that “since our September meeting, yields across the term structure have increased further, a sign that investors are reassessing the evolving macroeconomic landscape. My colleagues and I will need to come to our own judgment, which may take more time.” Williams said that there is “no need for urgency” about deciding on rate increases. These comments appeared to influence futures market probabilities.
Lately, inflation expectations have been relatively tame and stable, despite sharp rises in bond yields. The rise in yields appears to be largely due to an increase in the yield after inflation (the real yield). This is not simply my opinion. Rather, we can estimate investor expectations of inflation and, consequently, the real yield on bonds. And, as of now, the real yield is at a record high.
Thus, contrary to what some observers say, this is not simply a return to the normalcy seen before 2007. Rather, it likely reflects at least a partial shift in supply and demand conditions in the bond market. The massive issuance of bonds by technology companies has likely increased demand for credit, thereby contributing to the rise in real yields. It may also reflect an increase in the risk premium that investors require to hold bonds. That likely has to do with increasing concerns about fiscal sustainability.
Should we be worried? Perhaps. A high real yield means that the true cost (after inflation) of borrowing money is very high. That, in turn, will likely have a negative impact on credit-market activity. It could dampen activity in the already weakened US housing market. After all, mortgage interest rates follow Treasury bond yields. And it could stifle transactions such as mergers and acquisitions. On the other hand, high real yields could simply reflect expectations of a strong economy and, consequently, strong demand for credit.
What about tech investment in artificial intelligence? Technology companies have issued a massive volume of debt to fund the rollout of data centers. Plus, they plan to continue issuing more debt in the near future. High real yields on bonds will mean a higher cost of servicing those debts. That, in turn, means that, to service the debt, these companies will need to generate more cash than previously anticipated. This could create additional pressure at a time when US-based tech companies are facing increased competition from Chinese suppliers of AI services.
And what about the government? Higher real yields mean that the true cost of servicing the government’s massive debt is now higher than previously. Plus, it likely means that any decision to accelerate government borrowing (such as in response to an economic downturn) could be met by a further rise in yields. It means that the government may have less flexibility to make new fiscal choices.
The United States is not alone: Real yields have risen sharply in many other countries. The market for sovereign debt is a global market that crosses borders. A rise in demand for credit by US technology companies can influence the global supply and demand conditions for bonds, thereby boosting yields around the world. Plus, concerns about fiscal sustainability are not unique to the United States. Such concerns are also prevalent in Japan, the United Kingdom, France, and Italy, to name a few countries.
If leaders want to reduce real bond yields, one approach would be to announce credible plans to shift the trajectory of fiscal policy, leading to less government borrowing in the near future. This is not happening at the moment.
What about the rest of the region? In Indonesia, investment in information technology is accelerating, having nearly quadrupled from 2023 to 2025. For both Indonesia and Malaysia, much of the demand for the services from data centers appears to be coming from nearby Singapore, which is a major center of finance and biotechnology.
Thailand is also getting in on the game: It is not surprising that, in September, Thailand’s purchasing managers’ index for manufacturing was the highest of the 32 countries analyzed by S&P Global. There has been an acceleration in investment in data centers. In 2025 alone, investors sought permission to build digital projects valued at US$24 billion. Finally, Vietnam is also accelerating investment in data centers.
France is by no means alone in dealing with these issues. Yet, it has become the poster child for such concerns. Although France is nowhere close to facing default risk, some analysts are comparing this situation to the eurozone debt crisis in 2012, when the so-called PIIGS nations (Portugal, Ireland, Italy, Greece, and Spain), all faced significant default risk due to unbalanced fiscal policy and soaring debt-servicing costs.
The crisis was only resolved when Mario Draghi, president of the European Central Bank, said that the ECB will do “whatever it takes” to save the euro. Then yields fell, countries reformed their finances, and the crisis dissipated. Today, Greece—which faced major fiscal challenges during the crisis in 2012—has lower borrowing costs than the United States.
The problem now is that crisis begets crisis. That is, the perception that a crisis is brewing has led investors to sell French bonds for the safety of German bonds. This, in turn, can exacerbate the difficulty in servicing French debt. What is needed is a credible plan by France to boost fiscal probity. Yet, given the divided and fragmented legislature, it is hard to see how this will happen anytime soon.
However, investors are concerned about rising inflation in Europe: They likely view elevated energy prices as here to stay. That suggests that energy prices may gradually seep into other prices, thereby boosting underlying inflation. As such, investors now expect the ECB to tighten monetary policy further. Plus, the sharp rise in energy prices is likely contributing to the rise in bond yields across Europe.
The inflation report found that, from a year earlier, consumer prices were up 3.3% in Germany, 3.4% in France, 4.1% in Italy, 5% in Spain, 3% in the Netherlands, and 4.6% in Belgium.