Review of Last Week (9/21–9/25) and Outlook for This Week (9/28–10/2): The Stock Market Holds Up Despite High Interest Rates, While Capital Flows Back to AI

Looking back at the U.S. market last week, it was a very interesting week.

At the start of the week, buying returned to AI-related stocks against the backdrop of falling crude oil prices and stabilizing long-term interest rates, with the NASDAQ hitting a new high.

On the other hand, in the latter half of the week, U.S. Treasuries were sold off heavily due to strong economic indicators, inflation concerns, and expectations of additional rate hikes, with the 10-year Treasury yield temporarily rising to the 5.2% range.

Normally, such a rapid rise in interest rates would be a harsh factor for growth stocks, especially AI and technology stocks that rely heavily on expectations for future earnings.

Even so, the stock market as a whole did not collapse.

Perhaps this is the most important point to consider when looking back at last week.

“Why isn’t the stock market collapsing even though interest rates are rising?”

One answer to this can be found in the capital flow data released over the weekend.

Capital flows into U.S. stocks for the first time in 5 weeks, with large-cap tech at the center

According to LSEG Lipper, $37.6 billion flowed into U.S. equity funds in the week ending September 25. This is the first inflow in five weeks and the largest since June 17.

Looking at the details, the characteristics become even clearer.

Large-cap funds saw an inflow of $36.62 billion. Meanwhile, mid-cap stocks saw an outflow of $372 million, and small-cap stocks saw an outflow of $1.02 billion.

By sector, $4.89 billion flowed into technology, the largest amount since July 29. On the other hand, $2.53 billion flowed out of financials.

The same trend is seen globally.

Global equity funds saw an inflow of $44.1 billion, reversing two consecutive weeks of capital outflows. $5.29 billion went into technology.

Rather than saying the market has become bullish on stocks as a whole, it is more accurate to view it as:

capital reacting strongly once again to the profit growth of AI and large-cap tech

that seems closer to the reality.

Last week, Meta’s AI agent “Muse” topped the U.S. app download rankings, and the fact that AI has begun to expand from a mere capital investment theme to services actually used by consumers also supported the return of capital.

And Goldman Sachs points out that AI investment is driving nearly half of the S&P 500’s EPS growth this year.

I believe this is a key point in understanding the current stock market.

It can no longer be explained solely by ‘high interest rates = lower stock prices’

Meanwhile, a completely different landscape is unfolding in the bond market.

Last week, the U.S. 10-year Treasury yield broke through 5%, briefly rising to 5.2251%, its highest level in about 19 years.

The 30-year bond yield also rose to around 5.5%.

The backdrop is a strong economy and inflation.

The U.S. PMI released last week was 58.4 overall. Far from showing a stall in the economy, the strength of economic activity was reaffirmed.

Added to this are rising energy prices due to higher crude oil prices, and a series of remarks from Fed officials suggesting additional rate hikes, leading the market to largely price in an additional rate hike at the October FOMC. Looking at the components of the rise in long-term interest rates, we can see that the ‘term premium’ is rising sharply.

Here, let’s break down the rise in the 10-year Treasury yield a step further.

The graph above overlays the U.S. 10-year Treasury yield with the 10-year term premium based on the Fed’s Kim-Wright model.

Simply put, the term premium is the additional interest rate that investors demand for the ‘risk of holding bonds for a long period of 10 years’.

This premium tends to rise as uncertainty about inflation, government bond supply and demand, and interest rate fluctuation risks increase, in addition to future policy interest rates.

*According to the latest official data, the 10-year term premium was 0.9595% as of September 18, and rose to 0.9719% on September 16. This is the highest level in about 15 and a half years since February 2011.

In other words, the recent rise in 10-year Treasury yields cannot be explained solely by the fact that ‘the Fed will raise interest rates further’.

In addition to expectations for future short-term interest rates, investors are beginning to demand greater compensation for holding long-term bonds themselves.

I think this is important.

The current rise in long-term interest rates may include not only vigilance against inflation, but also a change in risk assessment for long-term interest rates themselves.

What is interesting here is,

The stock market is reacting positively to strong investment demand driven by AI, while the bond market is wary of the inflation that this strength creates.

That is the current dynamic.

In other words, while “AI” is currently a factor for earnings growth in stocks, it can also be a factor that pushes interest rates higher from the perspective of the bond market through capital expenditure, power demand, labor costs, and funding needs.

As I mentioned in last week’s article, I believe this is an important duality for understanding the current market.

Yet, capital is not fleeing from bonds

What is even more interesting is the capital flow.

Despite the sharp rise in interest rates, US bond funds saw an inflow of $5.93 billion last week.

In particular,

General taxable bond funds $4.15 billion
, short-to-medium term government bonds/Treasuries $2.15 billion
, short-to-medium term investment-grade bonds $1.63 billion
, and loan participation funds $1.31 billion

are among those receiving funds.

Money market funds (MMFs) also saw an inflow of approximately $11 billion.

This is extremely important.

It is not a simple movement of “rising interest rates equals fleeing from bonds.”

With yields exceeding 5%, it can also be considered that bonds themselves are becoming more attractive as an investment target.

In the current market,

capital seeking to capture earnings growth from AI and large-cap tech

and

capital looking to secure interest rates near 5%

exist simultaneously.

In other words, rather than a full-scale risk-on environment, it appears that a very distinct allocation of capital is taking place.

The next problem for AI investment is “memory shortage”

And this week, the sector I want to focus on in particular is semiconductors.

According to TrendForce, the price of PCIe Gen4 8TB server storage is rising from around $600–$700 in the first half of 2025 to $2,759.7, and is estimated to reach $3,580 in the third quarter.

DDR5 64GB server memory is similar, rising from the $200 range in the first half of 2025 to $1,289, and is estimated at $1,500 in the third quarter.

With the expansion of AI data centers, the wave of demand is spreading not just to GPUs, but

HBM → DRAM → NAND → SSD → Power → Cooling → Networking

to the entire infrastructure.

This is a major tailwind for memory manufacturers like Micron.

However, at the same time, it also means rising costs for those building AI data centers.

As AI investment expands, component prices rise, and financing interest rates also increase.

Therefore, going forward, I believe

not just ‘whether there is AI demand,’ but at what cost that demand can be realized

will also become an important issue.

The biggest corporate event this week is the Micron earnings report

In that sense, the earnings report I am most looking forward to this week is Micron Technology (MU).

Micron will announce its fiscal 2026 fourth-quarter earnings on September 30. The company has also scheduled an earnings call for 4:30 PM (ET) that same day.

In the previous Q3, they reported record-breaking results with revenue of $41.46 billion and Non-GAAP EPS of $25.11.

This time, beyond just revenue and EPS, what I want to confirm is

to what extent the rise in DRAM and NAND prices is contributing to profit margins

and

To what extent is demand for AI visible through 2027?

I suppose.

There are other earnings reports this week from companies like Accenture, Nike, and Jabil, but considering their connection to current market themes, I think it is fair to say that Micron’s earnings will confirm not just semiconductors, but AI capital expenditure itself.

Meanwhile, new risks have begun to emerge for AI.

There was another important piece of news regarding AI last week.

Multiple cases are being investigated where OpenAI agents took unintended actions, such as accessing government websites and leaking user images.

The important point here is not pessimism toward AI itself.

Rather, as AI evolves from “software that generates answers” into

agents that can access the web, research, make decisions, and take action on their own,

“control” becomes just as important as “performance.”

Going forward, when evaluating AI companies, not only model performance but also

operational capabilities, including safety, security, auditing, and governance,

could impact corporate value.

Precisely because the AI market is expanding, new challenges are also growing.

US-China Summit: Continued Dialogue Over “Major Resolution”

There were also new developments in US-China relations over the weekend.

Following the summit in Washington, the US and China announced an agreement to reduce tariffs on $30 billion worth of non-sensitive items from both sides and the start of a dialogue on AI. A framework will also be established to discuss risks and safety regarding AI.

What is important for the market is not that the “US-China issue has been resolved.”

Difficult issues such as semiconductors, Taiwan, critical minerals, and export controls still remain.

However, the fact that the world’s two largest economies have maintained a framework for continued dialogue on trade and AI is a factor worth noting, at least when considering short-term uncertainty.

The main focus this week is “Employment → PCE”

And this week, the spotlight shifts back to macroeconomics.

First, the August JOLTS job openings report will be released on Tuesday, September 29. The official BLS schedule also lists it for 10:00 AM (ET) on September 29.

And September 30 will be a very important day.

The ADP employment report will be released at 9:15 PM.

At 9:30 PM (JST),

August PCE Price Index,
Core PCE,
Personal Income,
Personal Consumption Expenditures,
and Q2 GDP

will all be released simultaneously.

The BEA has also officially scheduled the release of the August Personal Income and Outlays report for 8:30 AM (ET) on September 30.

According to the economic calendar, market expectations are:

Core PCE MoM +0.3%
Previous +0.2%

PCE YoY +3.4%
Previous +3.3%

PCE MoM +0.4%
Previous +0.2%

.

If the results are close to these figures, PCE will accelerate once again.

This will likely be the most important point this week.

Will a “strong economy” once again become a burden on the stock market?

Summarizing the trends that have driven the market until last week:

Strong economy
↓
Inflation remains sticky
↓
Expectations of additional Fed rate hikes
↓
Treasury selling
↓
Rising long-term interest rates
↓
Pressure on stock valuations

was the structure.

Therefore, if JOLTS and ADP data are strong, and PCE also exceeds market expectations, the 10-year Treasury yield may test the upside once again.

Conversely, if employment demand softens slightly and PCE also comes in below expectations, it will provide a reason for the rapid bond selling seen last week to pause.

For the current stock market, more than whether the economy is good or bad,

how that data changes the Fed’s next move

is what matters.

Things to consider this week

Last week, the U.S. 10-year Treasury yield significantly exceeded 5%.

Even so, the stock market did not collapse.

I believe one reason for this is that AI investment is translating into actual corporate profits.

And looking at capital flows, investors are not running away from AI; rather, they are pouring money back into large-cap technology.

However, that does not mean that interest rates are no longer an issue.

Rather, what is happening now is,

a state where profit growth driven by AI is outweighing the burden of rising interest rates

is the more natural way to think about it.

That is precisely why it is important to see how long this equilibrium will last.

If AI investment strengthens further, it will be a tailwind for companies like Micron.

On the other hand, if costs for memory, storage, power, data centers, and even capital financing rise, the next issue of the profitability of AI investment itself will emerge.

In that sense, I do not think it is necessary to view the current market as a binary choice of,

“whether the AI market will continue or end.”

I do not believe it is necessary to see it that way.

What we should pay attention to is,

to what extent AI demand is actually translating into revenue and profit.
And, how well that profit growth can absorb interest rates exceeding 5%.

The Micron earnings report and the PCE data being released this week may seem like completely different events at first glance, but they are actually materials to verify these two factors respectively.

Micron represents profit growth on the AI side.

PCE represents pressure from interest rates.

And how the stock market will react in between those two.

This week is likely to be a week to confirm the balance of the two forces currently supporting the market.