[Weekly Strategy] 5% Yield on Government Bonds, 30% Profit (Week 39 of 2026 | Issue No. 289)

Last week, the 10-year US Treasury yield rose by about 20 basis points to around 5.2%, its highest level since 2007.

In the same week, the Nasdaq 100 ETF gained 3.2%.

The bond market is sounding the alarm, while the stock market is crunching the numbers. The calculation centers on the two figures in the headline: a 5.2% government bond yield and 30% earnings growth.

Rising rates and stocks together isn't new

First, some reassurance: rising rates alongside a rising stock market aren't rare in US history. During the 2013 'taper tantrum', the 10-year went from 1.6% to 3%, and the S&P 500 gained nearly 30% for the year. In the second half of 2016, the 10-year rose from 1.4% to 2.6%, and small-cap stocks surged 14% in a month post-election.

Counterexamples are just as common. The S&P was flat throughout 1994, fell 19% in 2022, and in autumn 2023, as the 10-year hit 5%, the S&P pulled back 10%.

There's only one watershed: if rates rise because the economy is too strong and investment is booming, the stock market can handle it; if they rise due to runaway inflation, the market cannot.

The closest parallel is 1999. On June 30 of that year, Greenspan began hiking rates, raising them six times through May 2000 from 4.75% to 6.50%. The 10-year Treasury climbed from around 4.2% in October 1998 to roughly 6.8% in January 2000. Over the same period, the Nasdaq surged 85.6% in 1999, and the S&P gained 19.5%.

Where does it resemble today? Both feature a wave of new tech investment, the Fed just starting to hike, oil prices doubling (WTI rose from ~$12 to ~$25 in 1999), and tech leaders leading the charge.

On September 16 this year, Warsh delivered the first rate hike.

Who pushed the 5.2% up?

To determine which scenario we're in, we need to break down this 5.2%.

The 10-year Treasury yield consists of two parts: an inflation compensation component and the real interest rate (the true cost of borrowing) after stripping out inflation. Since February, inflation compensation has risen by only about 7 basis points, while the real rate has jumped by roughly 65 basis points. The real yield on the 10-year TIPS auctioned on Sept 17 was 2.653%, the highest since October 2008. At the same maturity in March, it was just 1.896%.

This round isn't driven by inflation panic, but by money genuinely becoming more expensive. This is the key difference from 1994 and 2022.

Three main forces are pushing up the real rate.

The first is the Federal Reserve. After the September hike, no dot plot member expects a cut. When Wednesday's PMI data came out, the probability of another hike in October jumped from ~55% to 70%. Since taking office, Warsh has caused larger volatility in the 2-year Treasury around FOMC meetings than his predecessors; this uncertainty itself commands a premium.

The second is AI infrastructure. Many assume massive corporate debt issuance by tech firms crowded out Treasuries. However, asset managers who specifically tested the six large, above-guidance AI bond issuances over the past year found no significant reaction in Treasury yields. The actual channel is simpler: major cloud providers' capex totals ~$700bn this year, consuming nearly all their operating cash flow and competing for the same power grid, labor, and construction crews. When investment appetite outstrips savings, the price of money must rise. Oracle's force majeure notice for its New Mexico data center project on Thursday serves as a footnote to these resource constraints.

The third is organic growth. Atlanta Fed GDPNow estimates Q3 at 5.0%, while sell-side tracking sits at 3.4%, with initial jobless claims at a low 197k.

There's also global long-end resonance: Japanese government bond yields have hit levels unseen since 1996, and the oil price floor remains above $90.

Two differences from 1999 should be noted. In 1999, US productivity had already taken off, and the fiscal budget was in surplus. Today, total factor productivity growth remains negative; AI has sparked an investment boom, but productivity dividends haven't materialized yet, and the fiscal stance is deficit-driven. Thus, the room for further rate hikes may be smaller than in 1999.

30% earnings can withstand 5.2% rates

Having covered the rate side, let's look at the other.

Latest data from last Friday: S&P 500 Q3 earnings are expected to grow 29.1%, exceeding 25% for the third consecutive quarter. Revenue growth stands at 12.1%, with a net profit margin of 15.0%, the second-highest since records began in 2009. Full-year earnings are projected to grow 32%.

Earnings guidance tells the same story. 72 companies issued positive Q3 guidance versus 44 negative, far above historical norms. Of those positive, 44 are tech firms.

Crucially, valuations. The S&P 500's forward P/E is 19.2x, lower than its 5-year average of 19.8x and comparable to its 10-year average of 19.0x. In March 2000, this figure was 24.4x.

In 1999, valuation expansion drove the rally; today, earnings growth is doing the heavy lifting.

At this stage, there's no need to imagine a crash. With earnings growing at 30% against a 5.2% risk-free rate, the math still works for equities.

The ledger does show some ugly numbers. Michigan Consumer Sentiment hit a four-month low of 48.1, and one-year inflation expectations jumped from 4.0% to 4.6%. The 30-year mortgage rate is nearing 7.5%, with mortgage applications down 11% YoY. Consumption and real estate are under pressure, but these aren't the engines driving S&P earnings this cycle.

One more thing to remember: after the 1999 hiking cycle began, the Nasdaq rallied for another eight months before peaking. As time goes on, gains concentrate in fewer leaders, making the final leg both strongest and most fragile. Highlighting risks doesn't mean being bearish; pullbacks are opportunities to enter.

White House state dinner, trader's big screen

On Thursday, Xi Jinping met with Donald Trump at the White House, marking the second head-to-head visit between the two leaders within six months.

Paper results: a tariff truce extended by two months until Jan 10, 2027; mutual tariff preferences on ~$300bn in non-sensitive goods; China to import at least 10 million tons of US coal annually in 2027 and 2028; establishment of trade and investment committees; inclusion of fentanyl precursors in controls; and a 'Super Intelligence' dialogue scheduled for November in Shenzhen.

What's omitted matters too. Rare earths get only a line about 'continuing to address US concerns,' and Taiwan issues remain 各自为政。

Before the summit, Craig Singleton of the Foundation for Defense of Democracies characterized US-China relations as a 'manageable stalemate,' predicting the summit would be 'form over substance.'

Markets reacted directly. The S&P was flat on meeting day, while the KWEB (China Internet ETF) dropped 0.6%. Weekly gains were concentrated Mon-Tue, driven by semiconductors, Meta, and oil—unrelated to the summit.

For US equities, the summit's significance lies in compressing tail risk: the probability of further tariff escalation before Jan 10 next year is low. It acts as a shock absorber, not an engine.

September 30: Two exams in one day

Next Wednesday, the 5.2% and the 30% each face their test.

Morning brings August PCE. If core PCE MoM exceeds 0.3%, an October hike is virtually certain, pushing the 5.2% higher.

After-hours features Micron earnings. The focus isn't on this quarter's numbers, but next quarter's guidance. It answers the article's core question: Is AI capex translating into profits, or merely demand?

[Next Week Outlook]

Next week is the grading week for US equities.

Last week's cards are played: bonds pushed the 10-year to 5.2%, stocks held firm on AI leaders, and the summit yielded no new leverage.

This week's table: Monday details on the US-China deal, Wednesday PCE and Micron earnings, Friday Non-Farm Payrolls. Being quarter-end, fund rebalancing will amplify volatility.

Bonds test inflation; stocks test earnings.

Implied volatility for Micron will climb ahead of earnings. 1040 is the level recently breached; holding it is one thing, falling back is another. The 5.2%-5.25% zone on the 10-year is the high-water mark of this cycle; a breakout means re-rating valuations.

Ultimately, stock prices follow profits. Rates dictate the multiple investors are willing to pay; earnings determine the base number that multiple multiplies.

[For next week's outlook and specific holdings, please visit our website]

(This article represents personal views and does not constitute investment advice. Data sources: Market data from Longbridge (closing prices, as of Sept 25); Treasury yields from TheStreet and Yahoo Finance daily closes; Real rate/inflation breakeven split from Russell Investments; TIPS auctions from US Treasury; Oct hike probabilities from CME FedWatch; AI bond impact analysis from PIMCO; Cloud capex from Tech Times citing Morgan Stanley estimates; Q3 GDP estimates from Atlanta Fed GDPNow and Goldman Sachs; Consumer sentiment from University of Michigan; Earnings/guidance/valuation from FactSet; 2000 P/E from JPMorgan 'Guide to the Markets'; Summit outcomes from White House fact sheet and Xinhua; Craig Singleton view from Yahoo Finance; PCE schedule from BEA.)