Why Are AI Stocks Still Rising Despite Soaring US Treasury Yields?

Editor’s Note: In September, the US financial market witnessed a notable divergence: US Treasuries faced significant selling pressure, with the yield on the 10-year Treasury rising by over 50 basis points in a single month, and some fixed-income assets dropping between 2.3% and 5%. Meanwhile, AI trading remained robust, semiconductor stocks continued to rise, and the Nasdaq index performed relatively well, while small-cap stocks, the equal-weighted S&P 500 index, and other interest rate-sensitive sectors generally faced pressure.

According to traditional valuation logic, rising long-term interest rates imply that future cash flows need to be discounted at a higher rate, which typically depresses stock valuations. However, tech stocks have not experienced the same level of adjustment as the bond market. Is the market genuinely worried about the US fiscal and inflation outlook, or is it betting that AI will drive economic growth sufficient to offset the impact of high interest rates?

Harry Mamaysky, founder of QuantStreet Capital, raised another layer of questions in his latest monthly investment letter: If the market believes AI can enhance overall economic productivity, why are the primary beneficiaries still infrastructure suppliers like semiconductors, rather than the broad range of companies that should benefit from AI in the future? The author does not believe that AI has formed a bubble, but remains cautious about the divergence between these earnings expectations and stock performance.

This question relates to whether AI investments can form a sustainable business loop. The capital markets are already willing to pay high costs for computing power, chips, and data centers, but these investments ultimately need to translate into corporate profits to support long-term returns. As financing costs continue to rise, the investment logic for AI may increasingly depend on one question: Who bears the construction costs, and who can truly reap the profits generated by AI?

Below is the original text compilation:

In September 2026, the most significant change in the US market was not the rise of AI stocks, but the rare divergence between bonds and stocks.

On one hand, the fixed-income market faced widespread selling. Data from the US Treasury showed that the yield on the 10-year Treasury rose from 4.75% on August 31 to 5.29% on September 30, a cumulative increase of 54 basis points. According to QuantStreet’s statistics, some US fixed-income assets dropped between 2.3% and 5% during the month.

On the other hand, tech stocks, particularly in the semiconductor sector, remained strong. Bitcoin, momentum strategies heavily invested in tech and semiconductor stocks, and the Nasdaq index all performed relatively well. In contrast, US small-cap stocks, the equal-weighted S&P 500 index, as well as real estate investment trusts (REITs), utilities, and financial sectors sensitive to interest rates generally performed weakly.

Even more unusually, both the dollar and commodities rose in September. This is inconsistent with the typical inverse relationship between the two and indicates that the market environment investors face is becoming increasingly complex.

For QuantStreet, there are two questions worth probing behind this round of market divergence: Why have long-term interest rates risen sharply while tech stocks remain strong? If investors are betting on economic growth driven by AI, why has this expectation not been reflected in the broader stock market?

1. Soaring US Treasury Yields: What Is the Market Worried About?

The yield on the 10-year Treasury rose by over 50 basis points in a month, indicating a significant change in the pricing of long-term funds.

Bond prices and yields typically move inversely. A rise in yields means existing bond prices fall, and the longer the duration and the more sensitive the bond is to interest rate changes, the more pronounced the price pressure usually is.

However, there is no consensus explanation in the market for the reasons driving this round of US Treasury selling.

The first explanation is that investors are losing confidence in the dollar.

However, Mamaysky is cautious about this judgment. The dollar actually appreciated in September, which is inconsistent with the narrative of a comprehensive sell-off of dollar assets. Of course, the rise in the dollar does not completely rule out long-term credit risk, but at least it indicates that the market has not experienced a one-sided crisis of confidence in the dollar.

The second explanation is that the US fiscal situation is prompting investors to demand higher long-term risk compensation.

The author also believes that existing evidence is insufficient to support this strong conclusion. He specifically mentions that the market-implied inflation compensation indicators (inflation breakevens) have remained relatively stable and have not shown changes commensurate with the rise in yields.

This indicator reflects the yield difference between nominal Treasuries and inflation-protected Treasuries and can be used to observe how investors price future inflation and related risk compensation. If nominal yields rise while inflation compensation does not increase significantly, it is difficult to attribute all changes to uncontrolled inflation expectations.

However, this does not mean that fiscal risks can be ruled out. Long-term yields are also affected by factors such as Treasury supply, term premiums, real rates, and market liquidity; stable inflation compensation does not alone prove that US debt is risk-free.

In contrast, Mamaysky is more focused on a third possibility: the market is repricing for stronger future economic growth and the enormous capital demand brought about by AI infrastructure construction.

Large cloud service providers and tech companies continue to expand capital expenditures (Capex), building data centers, purchasing chips, and investing in supporting power infrastructure. These expenditures mean that companies need to occupy more capital, which may also increase financing needs.

If investors simultaneously expect AI to bring higher productivity and future profits, then the rise in long-term interest rates may not necessarily be understood solely as a deterioration in economic risk; it may also partially reflect changes in growth expectations and capital demand.

This does not mean that AI investment has been proven to be the main reason for the rise in Treasury yields, but the author attempts to explain why some tech stocks can still rise when the bond market is facing selling pressure.

2. Why Do Higher Rates Strengthen AI Stocks?

Stock valuation can be simply understood as the discounted value of future cash flows.

All else being equal, the higher the market interest rates, the higher the return rate demanded by investors, and the lower the present value of a company's future profits. This is why high-valuation growth stocks are usually more sensitive to rising interest rates.

However, the market in September did not operate entirely according to this logic. Mamaysky’s explanation is that investors may believe that AI will create sufficiently strong future profit growth to offset the valuation pressure brought about by rising discount rates.

From the valuation formula perspective, this is equivalent to two forces competing against each other: one is the rising discount rate, which lowers the present value of future profits; the other is the expected increase in profits, which raises the intrinsic value of stocks.

If the growth rate of the latter is sufficiently large, stock prices may continue to rise even in the face of higher interest rates. In other words, the market may not be ignoring high rates but rather believes that future AI profits are sufficient to cover higher capital costs.

This explanation has some rationality. AI infrastructure investment is forming a massive demand. Chip manufacturers, semiconductor equipment suppliers, and related tech companies can relatively directly gain revenue from construction expenditures. Companies like AMD, Micron, Intel, Cisco, and Applied Materials also appear among the major holdings of the momentum ETFs that the author is focusing on. As long as the market believes that AI capital expenditures will remain high, the profit expectations of upstream suppliers may continue to be supported.

But the problem is that the revenue growth of these companies initially comes from other companies increasing capital expenditures, which does not necessarily mean that the entire economy has achieved corresponding productivity gains.

For companies purchasing chips and building data centers, expenditures first form costs or capital assets. Only when these assets ultimately help companies increase revenue, reduce costs, or improve profits can investments generate sustainable economic returns.

Therefore, the rise in semiconductor stocks can only indicate that the market is optimistic about the profit prospects of AI infrastructure-related companies, but it is not sufficient to prove that the ultimate economic returns of AI investments have been realized.

3. The Biggest Question About AI: Chip Companies Are Making Money, But What About Other Companies?

This is also the contradiction that Mamaysky is most concerned about in his investment letter.

In September, the semiconductor industry performed strongly, but the equal-weighted S&P 500 index performed weakly.

Compared to market-cap-weighted indices, the equal-weighted S&P 500 index gives each constituent stock roughly the same weight, making it easier to observe whether the market's rise is broadly spread rather than primarily driven by a few large tech companies.

The author refers to the vast array of companies outside the semiconductor sector as ROCS (Rest of the Corporate Sector).

In his view, AI investments have a logic that requires time to validate.

Companies are currently purchasing chips, servers, and software because they anticipate that these technologies will bring productivity improvements in the future. The market is willing to provide funding for this construction process in advance because it believes that unfulfilled profits will emerge in the future.

Therefore, it is not surprising that we do not see synchronized profit growth across all industries at this stage.

The question is that the stock market itself is forward-looking. If investors are convinced that AI will significantly enhance the future profitability of other companies, theoretically, this expectation should gradually be reflected in the stock prices of the relevant companies.

However, such widespread increases did not occur in September. Semiconductor stocks continued to strengthen, while other companies did not receive similar valuation support. This raises the author’s question: If the ultimate buyers of chips cannot obtain sufficient new profits, how can they long-term bear the increasingly large AI procurement and construction costs?

This question touches on the profit distribution mechanism within the AI investment chain.

In the short term, infrastructure suppliers may achieve high profits through order growth and tight demand. In the medium term, cloud service providers need to recoup investments through renting computing power and providing AI services. In the long term, ordinary companies need to convert AI into higher productivity, lower operating costs, or new revenue sources.

Only when this process gradually materializes can the value created by AI investments spread to broader economic activities.

Of course, the lack of synchronized stock price increases does not mean that the productivity gains from AI will not materialize. High interest rates, operational pressures within the industry, and different starting points for company valuations may all obscure the market's expectations for future profit improvements.

But at least from the stock price performance in September, investors' confidence in upstream AI suppliers is clearly stronger than their pricing of other companies' ultimate benefits.

For this phenomenon, Mamaysky does not conclude that AI investments are bound to fail. He still believes in the long-term value of AI and does not think the current market constitutes a bubble.

However, for this round of increases to gain broader fundamental support, we still need to see the profits created by AI no longer limited to a few tech companies.

4. What to Watch Next? Can Productivity Gains Be Translated into Corporate Profits?

In assessing the economic value of AI, Mamaysky has begun to pay more attention to productivity data.

The revised data released by the US Bureau of Labor Statistics (BLS) on September 3, 2026, showed that the labor productivity of the non-farm business sector grew at an annualized rate of 1.4% quarter-on-quarter and 2.2% year-on-year in the second quarter.

From a longer-term perspective, from the fourth quarter of 2019 to the second quarter of 2026, the annual average growth rate of labor productivity in the non-farm business sector in the US was about 2.1%, higher than the previous business cycle's level of about 1.5%.

This aligns with the productivity improvement trend observed by the author.

However, it is important to distinguish that an increase in macro labor productivity does not equate to the confirmation of AI's contribution. Capital investment, labor allocation, technological progress, and cyclical factors can all influence this indicator, and it cannot currently be directly calculated how much profit AI has created for companies.

Therefore, Mamaysky proposed a further verification standard: productivity improvements ultimately need to be reflected in the profits of companies outside the tech industry.

This is also reflected in QuantStreet's portfolio adjustments. Despite value stocks and low-volatility stocks performing poorly in the previous quarter, the institution still maintains a relative overweight in these two categories, hoping to retain exposure to a broader corporate sector. Meanwhile, in portfolios with higher risk tolerance, it continues to hold some tech stock investments.

In the fixed-income market, the institution has also begun to adjust duration. Duration is used to measure the sensitivity of bond prices to changes in yields. The longer the duration, the greater the price loss typically faced when interest rates rise; however, if yields fall, the potential price gains are also more pronounced.

The author believes that when the yield on the 10-year Treasury reaches around 5.25%, the potential investment attractiveness of bonds has begun to improve. Therefore, QuantStreet has slightly increased the duration of bonds in low-risk portfolios, marking a noticeable directional adjustment for the institution in over a year.

However, this does not mean that the institution is fully bullish on long bonds. Its model still does not favor high-duration fixed-income assets, and the overall duration of the portfolio remains below the benchmark, although the degree of underweight has narrowed.

For suitable investors, the author also mentioned the diversification role of alternative assets such as Evergreen Private Equity Funds. According to some product performances he listed, related funds rose about 0.5% to 0.75% in September, providing a certain diversification effect in a month when most stocks were under pressure. However, these products still face limitations such as valuation frequency, liquidity, and underlying asset risks, and monthly returns cannot prove their long-term defensive capabilities.

From these adjustments, it can be seen that QuantStreet has not chosen to completely exit AI trading, nor has it significantly shifted towards long bonds due to rising Treasury yields; instead, it is seeking a more balanced risk-return profile among different assets.

What truly needs to be observed in the future are three sets of signals.

First, whether AI infrastructure spending can be sustained and whether the revenue growth of upstream suppliers still has sufficiently strong demand support.

Second, whether AI has begun to improve the profitability of non-tech companies. Productivity data can provide early clues, but corporate profit margins, cost savings, and new revenue are more direct evidence of whether investment returns can be realized.

Finally, whether the rise in long-term Treasury yields reflects more economic growth expectations or inflation, fiscal supply, and term risk compensation. If growth does not improve as expected while financing costs remain high, corporate investment returns will face greater pressure.

The real question that AI trading needs to answer now is not just how long chip and computing power demand can continue to grow, but how much new profit these investments can ultimately create for the entire economy.

Semiconductor companies have already gained visible revenue from capital expenditures, but broader corporate profit improvements still need to be validated. Only when the productivity gains from AI gradually translate into real profits outside the tech industry can the market obtain more complete evidence to support the current large-scale investments.