橙子Joyce

Will the Federal Reserve raise rates as expected this week? Wall Street is debating: will it end the U.S. stock bull market?

After an unexpectedly strong U.S. CPI report came out last Friday, traders generally expect the Federal Reserve to begin raising rates at this week’s policy meeting—marking the first rate hike in more than three years.

Historically, previous rounds of rate hikes have offered a reference point for today’s market. Based on past experience (though history of course can’t guarantee the future), U.S. stocks may first weaken, then rebound.

Among the six tightening cycles since 1994, during the first four months after the rate-hike cycle began, the S&P 500’s average return was negative.

This suggests that once the “rate-hike shoe” drops, U.S. stocks may look lackluster through the beginning of next year.

As of the close last Friday, the benchmark U.S. equity index, the S&P 500, is up nearly 12% year to date. Strong corporate earnings and a fairly resilient economy have provided solid support for bulls in the stock market.

If you extend the time horizon, the S&P 500’s performance tends to improve gradually: in the 12 months after the start of a rate-hiking cycle, the index’s average return is close to 7%, with a median return of about 11%. (Using median-based statistics helps remove distortions from extreme outliers—for example, the index surged more than 40% after hikes began in March 1997.)

If the Federal Reserve implements a rate hike this Wednesday, it will be the first hike since July 2023—when the Fed raised rates to a range of 5.25% to 5.50%.

Currently, the federal funds rate in the U.S. is at 3.50% to 3.75%. According to the CME Group’s FedWatch tool, futures traders currently assign an 86% probability to a 25-basis-point hike this week.

One positive factor for the market is that mega-scale cloud service providers are still driving growth in excess returns through large-scale AI spending. The S&P 500 component stocks’ forecast for earnings growth in 2027 is expected to reach double digits. If the outlook for AI spending remains unchanged, it may be enough to offset any cooling in optimistic sentiment caused by the rate hikes.

Another bright spot for equities is that although inflation remains sticky, it appears to be slowing. The inflation rate has fallen from a May peak of 4.2%. This should allow the Federal Reserve to take a more gradual approach, and the data shows that the pace of rate hikes is crucial for stock performance—slower pacing gives investors more time to absorb policy changes!
$BZ
$CL
Energy

Elon Musk just threw another heavyweight bomb. When replying to a user on X, the head of SpaceX (SPCX.US) said clearly: “I am highly confident that SpaceX will deploy an NVIDIA (NVDA.US) VR NLV72 AI computer in space next year.”

Musk’s VR NLV72 refers to NVIDIA’s Vera Rubin NVL72, which this year has entered full-scale production. This rack-level AI supercomputer integrates 72 Rubin GPUs and 36 Vera CPUs. Its single-rack inference computing power reaches 3.6 EFLOPS, while its training computing power is 2.5 EFLOPS. The core Rubin GPU is based on TSMC (TSM.US) 3nm process technology, integrates 336 billion transistors, and comes with 288GB of HBM4 memory, delivering 22TB/s bandwidth. Per-card inference performance is 5x that of the previous-generation Blackwell. The total memory plus VRAM capacity of the entire cabinet is as high as 74.7TB—roughly equivalent to the total memory of 4,500 mainstream smartphones. NVIDIA’s own claim is that, compared with the GB200 NVL72, the inference cost per million tokens is only one-tenth.

Not just putting a chip in space

Sending a device like this to orbit is on a completely different scale from prior experiments that “ran a GPU in orbit.”

SpaceX’s roadmap is more specific than the outside world imagines. According to CFO Bret Johnsen’s remarks at a Goldman Sachs conference, the company will launch its first batch of Starmind AI1 satellites in the fourth quarter of 2027, and will significantly expand deployments in 2028. In essence, these satellites are “racks in space”—repurposing the Starlink V3 satellite platform, removing the communications phased-array antenna, swapping in a computing payload and a larger solar array, and adding a 110-square-meter deployable liquid-cooling heat dissipation unit. The first-generation AI1 satellite will have an approximately 20-meter deployment height, a 70-meter wingspan, a 210-kilowatt solar cell array, an average compute power of 120 kilowatts, and a peak power of 250 kilowatts.

Production is also moving forward. SpaceX’s AI satellite factory in Bastrop, Texas aims to achieve large-scale production by the end of 2027, with a long-term plan to deploy about 1 million AI satellites.

At the chip level, SpaceX is already NVIDIA’s “die-hard” customer. In an earnings call, Musk said bluntly: “We think the Vera Rubin architecture is the best architecture, the best AI computer, so we only choose NVIDIA.” Johnsen added that the partnership with NVIDIA helps SpaceX secure scarce production capacity allocations for GPUs amid current supply constraints.
I continue to invest in SPCX and GOOGL
$SPCXB

$BTC

1

Is a high interest rate the “terminator” for the U.S. stock market? Or are earnings the real factor?

This round of rate hikes is still some distance away from truly suppressing valuations. The key factor determining the valuation ceiling is not the interest rate level itself, but the earnings growth rate.

More broadly, Wall Street strategists have also not treated the Fed’s possible return to rate hikes as a signal ending the bull market. As long as economic growth and corporate earnings remain resilient, market pullbacks caused by moderate hikes may be only short-term volatility.

The threshold where valuations truly come under pressure is 5%-6%

The first tier is a super-growth environment where earnings growth exceeds 20%. In such a case, the valuation multiple can be supported as high as about 24x, corresponding to a 10-year U.S. Treasury yield of roughly 6%. The second tier is a above-trend growth environment with earnings growth of 10%-20%, where the valuation multiple is about 20x, corresponding to a yield of around 5%. Overall, the lower the earnings growth rate, the lower the interest-rate level the market can tolerate.

Currently, the S&P 500 trades at about 22x 2026 EPS, implying adjusted 2026 earnings growth of roughly 28%. For 2027, the valuation is about 18x, implying earnings growth of about 21% (excluding one-off investment gains/losses). This means that as long as earnings growth can be maintained at 15% or above, there is still room for further valuation re-rating in 2027.

There’s another yardstick for whether valuations are expensive. A two-stage dividend discount model shows that the implied equity risk premium is currently about 7.2%, which is around the 69th percentile historically; the long-term PEG is about 2x. In other words, as long as companies can deliver average annual earnings growth of 13%-15%, today’s valuation level still has fundamental support.

The 30 AI leaders currently trade at about 30x forward valuations. That compares with roughly 19x for the other 470 constituents in the S&P 500 and about 14.3x for MSCI ACWI peers. This valuation premium mainly comes from stronger visibility into earnings, lower leverage levels, and more stable shareholder returns.

Productivity is another buffer. If productivity stays in the 1.5%-2.5% range, the current yield can still support roughly a 20x valuation multiple. If AI further drives productivity above 2.5%, the valuation support would be even stronger.

In the short term, these two forces are enough to partially offset the pressure from rising financing costs: first, improved profitability in the financial sector; second, companies still hold about $2.4 trillion, and these funds can earn higher interest income. From a market-cap style perspective, large-cap stocks have stronger ability to absorb pressure.
$GOOG.US