[Feature: Weekly Market Wrap] The Week Good News Split in Two — Records in New York, a Sixth Circuit Breaker in Seoul

This week (July 1–7) the market got the good news it had waited for all cycle. June US jobs, out July 2, came in far below expectations (nonfarm payrolls +57,000 vs. consensus of about 110,000), releasing the "rate hike" fear the market had dreaded — the odds of a July hike collapsed to 7%. Yet the same good news split the assets. The Dow crossed 53,000 for the first time ever and the S&P 500 wrote a record high, but the Nasdaq alone fell among the three major indices, and in Seoul the KOSPI took two circuit breakers in a single week — its sixth of the year. The dividing line was one thing — how tethered you were to AI and semiconductors. That Samsung Electronics' stock collapsed on the very day it announced a record quarterly operating profit of 89.4 trillion won condenses that line. That the good news didn't work is a signal that the market has started pricing not the present (earnings, rates) but the future (the possibility that the AI cycle has peaked). The money didn't vanish — it just moved from an overpriced future into defensives and metals.

🇺🇸 US — The Good News Came, the Indices Split

The June jobs report, released July 2, was a shock. Nonfarm payrolls came in at 57,000 — far below consensus (about 110,000–115,000), a four-month low — and the unemployment rate was 4.2%. Weak jobs are usually bad news, but what this cycle's market feared was not a delayed cut but the risk of a hike — the Fed had erased its easing tilt at the June meeting and left the hike option open. The weak jobs erased that hike scenario; the odds of a July hike collapsed to 7%, and the expectation of a hold at 3.50–3.75% firmed up.

That relief pushed the indices up.

The Dow wrote a record high on July 2 at +1.10% (52,865.24), then crossed 53,000 for the first time ever on July 6 (53,055.91). The S&P 500 also set an all-time high of 7,537.43 on July 6. The July 6 rebound was reinforced by a strong AI-demand signal from Foxconn, and semiconductors and Big Tech jumped together (AMD +6.6%, among others).

Yet the Nasdaq alone fell — −1.51% from June 30, the only decline among the three major indices, with no record high anywhere in the span. The split was sealed on July 7: disappointment over Samsung Electronics' results and reports of China's DeepSeek developing its own AI chip combined to push the semiconductor index down −4.5% and the IT sector down −2.7%. In the same week, the broad Dow held its record high, but the AI-heavy Nasdaq and semiconductors couldn't hold on.

One stock sums up this split: Nvidia. Its fundamentals are still the strongest, yet its return this year stands at just +3.2%, far behind AMD (+171%), Micron (+305%), and Intel (+278%). Even its market-cap throne is under chase from Apple (Apple about $4.5 trillion vs. Nvidia about $4.7 trillion, a gap of about $190 billion, Apple +15% this year). This is a stretch where "the best company" becomes "the worst-performing stock."

Yields moved against common sense. Weak jobs usually mean lower yields, yet the 10-year rose from about 4.44% at the end of June to about 4.47–4.50% on July 7 — a two-week high — and the 30-year cleared 5%. What pushed long-term yields up was not jobs but oil — Iran's July 7 attack on a merchant ship in the Strait of Hormuz briefly revived a geopolitical premium in oil and stirred inflation worries (we return to this oil thread in the commodities section below).

What moved — For the US, this was a week that held both "good news pushed the indices up" and "only the Nasdaq couldn't rise" on one screen. Once the jobs shock released the hike fear, the broad Dow and S&P went to record highs, but the AI-heavy Nasdaq and semiconductors weren't rescued even by the same good news. The Nvidia paradox condenses that line — the moment the best results become the worst-performing stock, the market's eye is already on the next cycle, not the present.

🇰🇷 Korea — Where the Good News Didn't Reach

The good news that led the US to record highs didn't reach Seoul. Seoul's week was a roller coaster.

The KOSPI was already down −2.04% on July 1, then plunged −7.89% (−655 pts) on "Black Thursday," July 2, tripping the circuit breaker. The trigger was not US jobs but AI — news that Meta would resell its data-center compute capacity to outside parties spread "AI overinvestment, semiconductor overvaluation" fears, and chips were sold across the board. The prior year's surge — Samsung Electronics +490%, SK Hynix +1,030% — came back as profit-taking supply.

On July 3 it reclaimed the 8,000 line in a single day, +5.76% (+440.25 pts) — a roller coaster that ran from an intraday low of 7,378 to a high of 8,136. Talk of Anthropic–Samsung AI-chip production cooperation and reports that SK Hynix's long-term supply contract had its price cap removed were the rebound's material. On July 6, expectations of government support for semiconductors and AI added on, and it rebounded roughly +1% more (the exact close is uncertain).

Then on July 7, it collapsed again. −4.91% (−395.02 pts), pushed to −8.03% intraday, tripping the year's sixth circuit breaker. That single day also brought the 32nd sidecar, and the six circuit breakers are half of 2026's record of twelve. For the week, the KOSPI fell −7.8% and the KOSDAQ −10.6%. Among the materials that dragged semiconductors down again that day were reports of China's DeepSeek developing its own AI chip.

The anchor is Samsung Electronics. On July 7, Samsung announced preliminary Q2 operating profit of 89.4 trillion won, a record. Yet the stock plunged that day, and the "₩300,000 Samsung" level broke. Record present earnings were read, instead, as a "peak signal."

Open up the flows and the agent of defense is clear. On July 7, foreigners net-sold 2.934 trillion won and institutions 309 billion won, which retail absorbed with net buying of 3.317 trillion won. Foreigners had already been net sellers for 11 straight sessions as of July 3, and had sold about 4.3 trillion won on July 2 as well. Among the featured names, Samsung Electronics fell −9.06% on July 2 and, on July 7, went from an intraday −9.75% (₩287,000) to a close of about −6%; SK Hynix fell −14.57% on July 2 and closed about −6% on July 7.

The same macro left its mark on FX too. KRW/USD flipped direction, turning from 1,555–1,558 won early in the span (a weak won) to the 1,520s late in the span (a stronger won). Alongside the dollar weakness we'll see in the next section, the won turned stronger too — though with July 8 still in session, it's too early to call it a settled close.

What moved — What broke Seoul was not the US jobs shock but AI. July 2 Meta, July 7 DeepSeek — two "AI capacity glut" narratives dragged down the two semiconductor names that hold half the market cap, and because of that concentrated structure the index collapsed with them. The paradox of Samsung's stock falling on the day it posted its largest-ever results reveals that line most sharply — the moment good results become a bad stock price, the market was pricing not the present but the future.

🛢️🥇 Commodities and the Dollar — Where the Money Went, and the Oil That Traded on Its Own

On the other side of the jobs shock, there were beneficiaries. As Fed hike expectations receded, the dollar weakened (dollar index from about 101.28 early July to below 101 on July 7, its largest weekly drop since April), gold rose +4.03% ($4,026 on July 1 → $4,122 on July 7, briefly about $4,182), silver gained roughly +6% on the week (clearing $62 on Friday), and copper too was pushed up into historically high territory. The June jobs shock was the common driver of a weaker dollar, higher gold, higher silver, higher copper — a macro path separate from the AI narrative that broke Korea's market.

Oil alone stood outside this story. WTI was little changed, from about $70 on July 1 to the $69 line on July 7 (the starting figure is uncertain); Brent was in four-month-low territory at $73.29 on July 7. What moved oil was not macro but its own supply and demand — OPEC+ accelerated its August output increase on July 5 (188,000 barrels a day) and Saudi Arabia cut its Asian selling prices, easing supply, while the geopolitical premium that had spiked on the late-June Hormuz tanker strike mostly bled out over the week. The July 7 merchant-ship strike that briefly pushed up US long-term yields was only that drained premium lifting its head for a moment — it didn't reverse the weekly trend. Oil moved on a different axis — "more output vs. geopolitics," not "concentration."

What moved — Look at where the money went and you see this week's grammar. As the jobs shock pressed the dollar down, money flowed into broad physical assets like gold, silver, and copper. Oil alone dropped out of that flow because it had its own narrative — output increases and geopolitics. It's a control case showing that an asset not caught up in the concentration story has to be read through supply and demand, not macro.

What Ties This Week Together

What ties this week together is not causation but contrast. The US jobs shock did not cause Korea's plunge — Seoul's driver was AI oversupply (Meta, DeepSeek), while the jobs shock was a separate force that moved the dollar, metals, and US indices. Two narratives ran at once in the same week, and the result is "a market that split even after receiving the year's friendliest good news."

The dividing line was one thing — how tethered you were to AI and semiconductors. Once the hike fear lifted, broad assets (the Dow, the S&P, gold, silver, copper) all rose, but that good news couldn't save the Nasdaq, semiconductors, or the KOSPI. The more concentrated the place, the deeper the cut, and even in the US the index itself (the Dow) held while semiconductors couldn't.

The Samsung paradox tells us why. That record results become a bad stock price means the market has started pricing not the present (earnings, rates) but the future (the possibility that the AI cycle has peaked). And the money didn't vanish. Physical assets like gold and silver already had the macro tailwind of the jobs shock at their back (the commodities section above), and on top of that, money leaving AI landed in the same direction — the two forces overlapped, gathering money into defensives (healthcare +2.6%, utilities +2.2%) and metals. The rotation is not selling but reallocation.

Next week's watch (a schedule, not a certainty):

• Samsung Electronics' final results — how the preliminary 89.4 trillion won gets confirmed; the next hard measurement in the "memory peak" debate.
• Mid-July CPI and late-July FOMC — the price and monetary events absent from this span come next. At stake is whether the hike fear revives and when the window for a cut opens.
• A repricing of US semiconductors and the KOSPI — whether the DeepSeek- and Meta-driven "AI capacity glut" narrative continues or reverses.

What this week left is not an answer but a frame. The market has started pricing the future, not the present, and the places where conviction about that future sat heaviest were shaken the hardest.