According to information compiled by Woofun AI, the US financial market in September 2026 displayed a highly tense divergence: as QuantStreet Capital founder Harry Mamaysky issued a warning in his latest monthly investment letter, the market is facing a rare split between a sharp selloff in US Treasuries and a powerful rally in AI assets, and this structural profit mismatch may be eroding the sustainability of the AI investment thesis under a high interest rate environment. The fixed income market suffered broad valuation repricing pressure during the month, with changes in its core indicators far exceeding historical norms. Specifically, the 10-year US Treasury yield climbed rapidly from 4.75% on August 31 to 5.29% on September 30, accumulating an increase of 54 basis points in a single month. This sharp steepening of the yield curve directly led to a significant decline in bond prices. Some US fixed income assets fell between 2.3% and 5% in September. A pullback of this magnitude is not an isolated event, but a direct reflection of profound changes in the long-term pricing mechanism for capital. When the risk-free rate rises quickly, the capital loss pressure borne by long-duration assets amplifies non-linearly, forcing investors to reassess the cost-effectiveness of asset allocation at current rate levels. It is worth noting that this selling was not triggered by a single liquidity shock, but by the resonance of multiple macro variables, including concerns about fiscal sustainability and an implicit repricing of the future economic growth path. For institutional investors who rely on fixed income as a ballast, this round of adjustment means that duration exposure strategies built over the past few years are facing a severe test, and the psychological threshold of yields breaking above 5% marks a new high range for financing costs. In sharp contrast to the turmoil in the bond market, extreme structural divergence appeared within the stock market. The technology sector, especially semiconductors, continued to maintain strong performance in September, and the Nasdaq index showed relative resilience even as the broader market came under pressure. At the same time, Bitcoin and momentum strategy products with technology and semiconductor stocks as their core holdings also delivered outstanding returns. However, this rally was not broad-based but highly concentrated in a handful of leading assets. By contrast, the equal-weighted S&P 500 index, which represents broader corporate fundamentals, performed weakly, showing that the market's upward momentum did not effectively spread to small and mid-cap stocks. In addition, rate-sensitive sectors such as real estate investment trusts (REITs), utilities, and financials generally came under enormous downward pressure. Even more puzzling was that the US dollar and commodities rose simultaneously in September, breaking their usual inverse relationship. This "double rise" pattern suggests that the market environment investors face is becoming more complex and chaotic, and traditional macro hedging logic appears to have temporarily failed, with capital flows across asset classes showing irrational crowding characteristics that further intensified market divergence. Regarding the drivers behind the surge in US Treasury yields, the market holds multiple interpretations, but most traditional narratives appear inadequate in the face of empirical data. The first mainstream view holds that investors are losing confidence in the US dollar and are therefore selling dollar assets. However, Harry Mamaysky is skeptical of this judgment, because the dollar actually appreciated in September, which is inconsistent with the narrative logic of a broad selloff in dollar assets. Of course, a stronger dollar does not completely rule out the existence of long-term credit risk, but it at least shows that the market has not experienced a one-way crisis of confidence in the dollar. The second explanation focuses on the deterioration of US fiscal conditions, with investors demanding higher long-term risk compensation. The author also believes that existing evidence is not yet sufficient to support such a strong conclusion. He specifically points out that market-implied inflation breakevens remained relatively stable and did not move by the same magnitude as the rise in yields. This indicator reflects the yield difference between nominal Treasuries and inflation-protected Treasuries, and is a key window for observing how investors price future inflation and related risk compensation. If nominal yields rise while inflation compensation does not increase substantially in tandem, it is difficult to attribute the entire change to runaway inflation expectations. However, this does not mean fiscal risk can be completely excluded, because long-term yields are also affected by a complex mix of Treasury supply, term premiums, real rates, and market liquidity, and stable inflation compensation alone cannot prove that US debt is risk-free. By contrast, Mamaysky prefers a third explanation: the market is repricing for stronger future economic growth and the enormous funding needs brought by AI infrastructure construction. Large cloud service providers and technology companies continue to expand capital expenditures (Capex) to build data centers, purchase chips, and invest in supporting power infrastructure. These massive outlays mean companies need to tie up more capital and may also increase external financing needs. If investors simultaneously expect AI to bring higher productivity and future earnings, then rising long-term rates would not necessarily be understood only as worsening economic risk, but could also partly reflect changes in growth expectations and capital demand. This shift in perspective is crucial because it links the bond market selloff to the real economy's investment boom. In other words, the rise in US Treasury yields may be a market "vote" for AI-driven economic growth, even though that vote comes with high financing costs. This logic partly explains why some technology stocks can still rise while the bond market is being sold off, because both share the same macro narrative: AI will reshape the global economic landscape, and the current rate level is the "entry fee" paid for that grand vision. Data compiled by Woofun AI shows that at the level of equity valuation, this logical contest is reflected in a fierce battle between a rising discount rate and expectations for AI's future earnings. According to traditional valuation models, all else being equal, the higher the market interest rate, the higher the return investors demand, and the lower the present value of companies' future earnings. This is why high-valuation growth stocks are usually more sensitive to rising rates. However, the market in September did not operate entirely according to this linear logic. Mamaysky's explanation is that investors may believe AI will create sufficiently strong future earnings growth to offset the valuation pressure from a higher discount rate. From the valuation formula's perspective, this amounts to two forces competing with each other: one is a rising discount rate, which lowers the present value of future earnings, and the other is higher expected earnings, which raises the intrinsic value of stocks. If the latter's increase is large enough, stock prices may continue to rise even in the face of higher rates. Companies such as AMD (AMD.US), Micron (MU.US), Intel (INTC.US), Cisco (CSCO.US), and Applied Materials (AMAT.US), as core suppliers of AI infrastructure, appeared among the main holdings of the momentum ETFs the author follows. As long as the market believes AI capital expenditure will remain elevated, earnings expectations for upstream suppliers may continue to receive support. This logic of "earnings growth offsetting rate pressure" provides a theoretical basis for the strength of technology stocks, but it also plants enormous uncertainty. However, there is a core question in this logical chain: why has ROCS (Rest of the Corporate Sector) not benefited in tandem? In September, the semiconductor industry performed strongly, but the equal-weighted S&P 500 index was weak. Compared with the market-cap-weighted index, the equal-weighted S&P 500 gives each constituent roughly the same weight, making it better for observing whether gains are spreading broadly rather than being driven mainly by a small number of large technology companies. The author refers to the broad group of companies outside semiconductors as ROCS, and in his view, AI investment contains a logical loop that needs time to verify. Companies are buying chips, servers, and software now because they expect these technologies to deliver productivity gains in the future. The market is willing to fund this buildout in advance because it believes future profits that have not yet materialized will appear. Therefore, it is not surprising that at this stage not all industries are seeing synchronized profit growth. The problem is that the stock market itself is forward-looking. If investors are convinced that AI will significantly improve the future profitability of other companies, then in theory this expectation should gradually be reflected in the share prices of the relevant companies. But in September there was no such broad rally; semiconductor stocks kept strengthening while other companies did not receive similar valuation support. This led the author to ask: if the ultimate buyers of chips cannot obtain enough additional profit, how can they bear the increasingly large AI procurement and construction costs over the long term? This question touches on the profit distribution mechanism in the AI investment chain and is also the root of the current market disagreement. To verify whether AI is truly bringing productivity gains, Mamaysky began paying closer attention to macro data. Revised data released by the US Bureau of Labor Statistics (BLS) on September 3, 2026 showed that nonfarm business sector labor productivity rose at a quarter-over-quarter annualized rate of 1.4% in the second quarter and 2.2% year over year. Over a longer horizon, from the fourth quarter of 2019 to the second quarter of 2026, US nonfarm business sector labor productivity grew at an average annual rate of about 2.1%, higher than the roughly 1.5% level of the previous business cycle. This is consistent with the productivity improvement trend observed by the author. But a distinction must be made: a rise in macro labor productivity does not equal confirmation that AI's contribution has been established. Capital input, labor allocation, technological progress, and cyclical factors can all affect this indicator, and at present it is not possible to directly calculate from it how much profit AI has created for companies. Mamaysky therefore proposes a further verification standard: productivity improvements ultimately need to show up in corporate earnings outside the technology sector. Only when non-technology companies can prove that they have achieved cost savings or revenue growth through AI applications will the commercial loop of AI investment truly be formed. Otherwise, the current boom may remain merely at the stage of "selling shovels," while the "gold diggers" have not yet found enough gold to pay for the shovels. This cautious attitude toward the causal relationships behind the data reflects professional institutions' calm thinking in a frenzied market. Based on the above judgments, QuantStreet made slight adjustments to its investment strategy, seeking a more balanced risk-return profile in an uncertain environment. Although value stocks and low-volatility stocks performed poorly in the previous quarter, the institution maintained a relative overweight in these two categories, hoping to retain exposure to the broader corporate sector to capture potential mean-reversion opportunities. At the same time, in portfolios with higher risk tolerance, it continued to retain some technology stock investments to share in the benefits of AI infrastructure construction. In the fixed income market, the institution also began adjusting duration. Duration is used to measure how sensitive bond prices are to changes in yields. The author believes that when the 10-year US Treasury yield reaches about 5.25%, the potential investment appeal of bonds has begun to improve. Therefore, QuantStreet slightly increased bond duration in low-risk portfolios, a relatively clear directional adjustment for the institution in more than a year. However, this does not mean the institution has turned fully bullish on long bonds; its models still do not favor high-duration fixed income assets, and overall portfolio duration remains below benchmark, though the degree of underweighting has narrowed. In addition, for suitable investors, the author also mentioned the diversification role of alternative assets such as evergreen private equity funds. According to the performance of some products he listed, the relevant funds rose about 0.5% to 0.75% in September, providing some portfolio diversification effect in a month when most stocks came under pressure. However, such products still have limitations related to valuation frequency, liquidity, and underlying asset risk, and a single month's return cannot prove their long-term defensive capability. What truly needs to be watched in the future are three key signals, which will determine whether the AI investment narrative can continue. First, whether AI infrastructure spending can continue and whether revenue growth at upstream suppliers still has sufficiently strong demand support. Second, whether AI has begun to improve the profitability of non-technology 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 US Treasury yields reflects more of growth expectations or more of inflation, fiscal supply, and term risk compensation. If growth fails to improve as expected while financing costs remain elevated, corporate investment returns will face greater pressure. The question the current AI trade truly needs to answer is no longer just how long demand for chips and computing power can keep growing, but how much additional profit these investments can ultimately create for the entire economy. Every dollar of bonds issued by the US Treasury corresponds to a financing activity in the real economy. If these funds cannot be transformed into broad productivity gains, then debt accumulation under high interest rates will become a sword of Damocles hanging over the market. Only when AI's productivity gains gradually translate into real profits outside the technology sector can the market obtain more complete evidence to support the current large-scale investment; otherwise, this growth狂欢 driven by capital expenditure will eventually face a severe test in the reality of profit mismatch.