Significant Divergence in Capital Flows! US Equity Funds See Highest Net Outflow of the Year, While Tech-Themed Funds Attract Inflows Amid the Astra Frenzy

English Finance APP reports that global capital flow data compiled by LSEG Lipper, along with stock market trading trends, indicate a “risk-off in aggregate, selective opportunity-seeking” divergence in global equity markets over the past two weeks. Based on closing prices from August 28 to September 10, Korea’s benchmark KOSPI Index—widely regarded as an “AI computing power indicator”—rose about 3.61%, with its two heavyweight stocks, Samsung Electronics and SK Hynix, climbing approximately 4.67% and 12.10% respectively; meanwhile, the Philadelphia Semiconductor Index gained about 1.26% during the same period, but surged 3.37% on September 4 only to decline 2.66% on September 10, reflecting that the global rebound in AI computing power-related stocks and the wider AI computing power sector remains under pressure from interest rate shocks. Nonetheless, capital continues to actively flow into Asian equity funds and technology sector funds.

In line with these market movements, the latest global equity and bond fund inflow and redemption data compiled by LSEG Lipper reveals investors are significantly reducing exposure to US large-cap equity funds while continuing to purchase European, Asian equity funds as well as global technology sector funds. The expansion of high-performance AI computing power necessities triggered by the introduction of OpenAI Astra, coupled with the dominance of the RSI (Recursive Self-Improvement) training paradigm in AI training, has led to explosive growth in AI computing demand and associated profit opportunities, but market tolerance for its valuation and financing costs is waning.

Wall Street powerhouse Morgan Stanley, referencing the emergence of OpenAI Astra—described by Nvidia CEO Jensen Huang as ushering in the “AGI era”—has highlighted a key trend: the dramatic enhancement of large AI model capabilities is making more workloads economically viable, thereby strengthening constraints across the AI computing power, data center electricity chains, substrate, and storage manufacturing supply. In its scenario projections, hyperscale cloud operators’ computing capacity deployment is set to expand power capacity from about 35 gigawatts in 2025 to roughly 145 gigawatts by 2028, an increase of approximately 4.1 times.

GPT-6 Astra was unveiled on September 3, 2026 amidst heightened anticipation, with Astra’s release unprecedentedly boosting market enthusiasm for the imminent arrival of the AGI (Artificial General Intelligence) era—leading to an even stronger trajectory of computing power demand. Notably, the new “output-based billing” growth model is expected to drive much greater overall computing power demand, with the most direct evidence of surging demand coming from the AI R&D process itself—namely, the “AI building AI” track defined by Recursive Self-Improvement (RSI).

As Astra leads the way in advanced, front-line large models and brings increasingly robust computing power demand, Morgan Stanley forecasts that combined capital expenditure by North America’s four largest hyperscale cloud and AI application operators will climb from $917 billion in 2026 to $1.47 trillion in 2027 and $1.64 trillion in 2028, with deployed capacity projected to grow from 35 gigawatts in 2025 to 145 gigawatts in 2028 within the same period.

The crucial industrial signal from Astra is that more complex tasks are now commercially viable for AI execution. On September 10, media reported that OpenAI released a ChatGPT product tailored to the financial services sector, combining GPT-6 Astra with professional data sources to support research, financial modeling, and client materials creation. Projecting forward, the growth variables for AI demand will further extend to the number of concurrent agents, task duration, tool invocation frequency, and context scale: as the cost of completing a task drops and success rates improve, companies have good reason to deploy more workflows. This opens up vast incremental space for cloud-based AI inference computing power and high-performance storage demand related to AI, providing the latest foundation for the market’s reassessment of the durability of AI infrastructure growth.

Massive Outflows from US Equity Funds, Tech Funds Buck the Trend: Global Capital Narrows Exposure and Selects Sectors

The latest fund inflow and redemption data shows investors have sharply cut exposure to US large-cap equity funds, while continuing to acquire European, Asian equity funds and global technology industry funds. Value stocks, high free cash flow equities, and bonds are not proving to be universal safe havens. During the same period, using total return closing prices of the corresponding index ETFs as the observation standard, the IWD ETF tracking the Russell 1000 Value Index fell about 1.94%, COWZ—focusing on high free cash flow yield stocks—declined roughly 3.79%, AGG tracking the Bloomberg US Aggregate Bond Index dropped around 1.13%, while the S&P 500 ETF—SPY—was down about 1.50%.

These figures highlight that high free cash flow boosts companies’ ability to handle financing pressure, but cannot eliminate risks from strong sectoral cycles, portfolio structure, or valuation corrections; net subscriptions in bond funds do not necessarily mean simultaneous bonds price rises. The current clearer positioning shift is that capital is becoming more selective and tends to control the overall interest rate sensitivity across stock-bond portfolio allocations.

LSEG Lipper data shows that for the week ended September 9, global equity funds saw net outflows of $15.52 billion, the largest weekly outflow since March 18. US equity funds pulled out a surprising $32.27 billion, the highest since the week ended December 17, 2025, when net outflows hit $52.45 billion; Europe and Asia equity funds, however, saw net inflows of $11.16 billion and $3.03 billion respectively.

The US market’s divestment scale exceeded the net outflows of global equity funds, indicating that net buying in other regions partly offset losses. This demonstrates regional allocation divergence and does not imply that the same group of investors are shifting US asset proceeds entirely to Europe or Asia; fund inflow and redemption statistics also do not equate to overall cross-border capital flows in stock markets.

Within the US market, withdrawals were concentrated in large-cap equity funds: weekly net outflows of $40.44 billion, a record; mid-cap funds saw net outflows of $682 million, while multi-cap funds and small-cap funds had net inflows of $3.52 billion and $274 million, respectively. Industry selection occurred alongside overall divestment: global sector funds saw net inflows of $2.92 billion, with technology and financial sectors drawing $1.89 billion and $1.25 billion respectively; US sector funds saw net inflows of $1.46 billion, with technology posting $1.71 billion net inflow and finance $720 million.

A combined net influx into technology and financial sectors exceeds overall sector fund flows, largely indicating substantial single-direction capital withdrawals from other industries. Consequently, funds demonstrate a shrinking broad equity exposure while retaining specific sector opportunities; it is worth noting that region, industry, and capitalization classifications are separate LSEG Lipper statistical dimensions and cannot be simply summed or subtracted.

Inflationary Pressure Redirects Funds! Short-Term Bonds Build a Defensive Wall, Energy Allocation Rises

The macro pressure driving this round of adjustments is still energy supply shocks and their rate consequences: Brent crude surpassed $100 per barrel on September 9, reaching $109.97 during September 11 trading; WTI broke July’s $93.50 high during the week and hit $104.46 on September 11, both marking four-month highs. US PPI and CPI data released on September 10, along with previous warming prices in Europe and Japan driven by energy inflation, further strengthen inflation’s stickiness and rate hike concerns.

The clearest shift in bond allocations is investors’ increased willingness to take on short-term risk. For the week, global bond funds saw net inflows of $8.95 billion, the lowest weekly inflow since July 29; of this, short-term bond funds attracted $6.65 billion, the second-highest single-week inflow in three months. Loan participation funds and government bond funds each brought in $1.01 billion and $743 million respectively, while corporate bond funds had net outflows of $2.37 billion. US bond funds saw their 21st consecutive week of net inflows, bringing in $6.56 billion; short-to-intermediate investment-grade bond funds attracted $3.75 billion, a nine-week high, while short-to-intermediate government bond and US Treasury funds absorbed $2.78 billion.

Overall, investors still require bond coupon income, but are placing greater emphasis on duration control and credit quality; the outflow from global corporate bond funds and inflow into US short-to-intermediate investment-grade funds also reflect differences in region and term structure.

Money market fund flows further demonstrate that “safe-haven” trading is not globally synchronized. Global money market funds saw net inflows of $10.72 billion for a second consecutive week; however, US money market funds saw net outflows of $10.41 billion, after net inflows of roughly $48.76 billion the previous week. Thus, massive redemptions from US equity funds did not mechanically convert into net subscriptions for US money market funds in the same week.

The pace of inflows and redemptions, capital use, and regional distribution may all affect outcomes; current data is insufficient to track every redemption’s final destination, but it is clear: global liquidity preference is rising, which can coincide with phase-outflows from US cash-equivalent funds.

Fund distribution in commodities and emerging markets is also highly selective. Global gold and precious metals funds ended eight consecutive weeks of net inflows, seeing a net outflow of $537 million for the week; energy funds attracted $211 million. The allocation increase to energy is consistent with investors’ focus on supply shocks and potential beneficiaries, but this does not imply that precious metal funds are directly channeling capital into energy. Data covering 28,984 funds also shows emerging market equity funds ended eight consecutive weeks of net inflows with a net outflow of $1.56 billion for the week; emerging market bond funds saw net inflows for the sixth week in a row, attracting $537 million.