Semiconductor Selloff, Calm Index: What the July 2026 Chip Rout Reveals About the Market Beneath the S&P 500

Last updated: July 29, 2026, 5:00 a.m. EDT

The S&P 500 rose 0.21% on Tuesday, July 28, 2026, to close at 7,428.78. On the same day, the Philadelphia Semiconductor Index fell 4.5% after trading down as much as 6.5%, South Korea’s Kospi lost 10.8% and tripped a circuit breaker, Micron shed 8.9%, and the Nasdaq-100 slid toward correction territory. Two numbers, one session, and almost no relationship between them.

That gap is the story. A four-day semiconductor selloff — triggered by a report that China has begun mass-producing its own deep ultraviolet lithography tools, amplified by the blockbuster Shanghai debut of Chinese memory maker CXMT, and accelerated by leverage embedded in newly launched single-stock exchange-traded products — has torn through the most crowded trade in global equities. Yet the headline U.S. index has barely flinched, because money leaving chips has landed in health care, financials, industrials and staples rather than leaving the equity market altogether.

The immediate question investors are asking is simple: is this a rotation or the first leg of a broader unwind? The honest answer, as of the New York open on Wednesday, July 29, is that the evidence supports rotation and does not yet rule out something worse. The rotation case rests on hard data — the equal-weighted S&P 500 reached a record high on July 28, the VIX fell to 18.21, health care and financials sector funds set records, and 26 of the Dow’s 30 members advanced. The unwind case rests on positioning: options markets show hedging demand spiking, dispersion at a six-year high, and a semiconductor-versus-index volatility spread more than five standard deviations above its average. Those are not the fingerprints of a sleepy summer tape.

Three events over the next 72 hours will do more to settle the argument than any amount of technical analysis. The Federal Reserve announces its decision at 2:00 p.m. EDT on Wednesday, July 29, with the target range at 3.50%–3.75% and an unusually live debate about whether the next move is up. Microsoft and Meta Platforms report after the close on the same day; Apple and Amazon follow on Thursday, July 30. And the Bank of Japan meets July 30–31 with the yen near levels last seen in the mid-1980s. Each carries the capacity to convert a sector repricing into something that reaches the index.

Key Takeaways

• Main development: A four-session semiconductor selloff, set off by a July 27 report from The Information that China has begun mass-producing domestic immersion DUV lithography machines, spread from Asia to the United States and pushed the Nasdaq-100 toward correction while the S&P 500 finished higher.
• Key figures: The Philadelphia Semiconductor Index fell 4.5% on July 28 after an intraday decline of as much as 6.5%; the Kospi dropped 10.8%, its eighth circuit-breaker halt of 2026; the S&P 500 closed up 0.21% at 7,428.78 and the Dow Jones Industrial Average rose 1.03% to 52,747.32.
• Market response: Capital rotated rather than exited. The equal-weighted S&P 500 set a record high on July 28, health care and financials sector funds hit records, and the VIX slipped 2.46% to 18.21 — a level more consistent with repositioning than with panic.
• Why it matters: The top 10 stocks account for roughly 43% of S&P 500 market capitalization, a record. When the index masks a violent move in its most-owned names, investors reading only the headline number can badly misjudge the risk sitting inside their own portfolios.
• What comes next: The Federal Open Market Committee announces at 2:00 p.m. EDT on July 29 with the target range at 3.50%–3.75%; Microsoft and Meta report the same afternoon, Apple and Amazon on July 30; the Bank of Japan concludes its meeting July 31.

What actually happened between July 27 and July 29

The sequence matters more than the size of any single move, because each leg of the selloff was triggered by a different piece of news and hit a different part of the market.

It began on Monday, July 27, in two places at once. In Shanghai, CXMT Corp. — formerly ChangXin Memory Technologies, China’s largest producer of dynamic random-access memory — surged 466% on its trading debut. In New York, The Information published a report that a Shanghai-based, state-backed manufacturer had begun mass-producing immersion deep ultraviolet lithography systems, the class of machine that ASML Holding has dominated for two decades. ASML fell roughly 6% to 7% on the day, according to Bloomberg’s reporting, and U.S. chip names followed: Advanced Micro Devices dropped 7.3% and Micron Technology fell nearly 5% by midday.

Overnight into Tuesday, the reaction found its most concentrated expression in Asia. The Korea Exchange suspended program trading as futures collapsed, then halted cash trading on the Kospi for 20 minutes at 10:13 a.m. local time after the benchmark fell 8.02%. It closed down 10.8% at 6,023.66, its weakest finish since April. Samsung Electronics shed 13.4% and SK Hynix lost 14.7%. Japan’s Nikkei 225 fell 3.95%, with Kioxia down more than 10% and Screen Holdings, a chip-equipment supplier, off 17%. Taiwan’s market also broke lower.

By the time New York opened on Tuesday, the damage had a clear shape: memory and semiconductor capital equipment took the hit, and most of the rest of technology did not. Micron fell 8.9%, AMD lost 8.1% and Applied Materials shed 7.8%. Meanwhile Alphabet rose 1.85%, Microsoft gained 1.09% and Apple added 0.94%, briefly touching a $5 trillion market capitalization intraday before closing below it. The Nasdaq Composite finished down only 0.22% at 24,876.91 despite trading as much as 9.3% below its record at one point in the session.

The Dow told the opposite story, rising 537.24 points to 52,747.32 — still short of its July 6 record close of 53,055.91 — on the back of earnings beats from Sherwin-Williams and Coca-Cola and rotation-driven gains in IBM, Salesforce and Amgen. Because the Dow is price-weighted and holds no memory manufacturers, the same session that punished the Nasdaq barely registered there. The Russell 2000 added 0.20% to 2,953.80.

Then, overnight into Wednesday, July 29, the selling stopped. Korean and Japanese equities rebounded, with the Kospi up roughly 2.8% and the Nikkei 225 reclaiming 63,000, after SK Hynix reported record quarterly results and raised its capital spending plans. Whether that bounce holds through the Fed decision and two nights of mega-cap earnings is the open question of the week.

The three catalysts, and how much each one actually explains

China’s DUV report: a supply-chain story dressed as a technology story

The proximate trigger was a July 27 report in The Information stating that China has begun mass production of homegrown DUV chipmaking tools, with the first domestic immersion systems due for delivery this year to SMIC, Hua Hong and CXMT. Subsequent coverage identified Shanghai Yuliangsheng Technology — a startup linked to Huawei’s SiCarrier investment vehicle — as the developer, and put the initial run rate at roughly five immersion DUV systems in 2026 and about 20 in 2027, aimed at 28-nanometer production.

The market treated this as a competitive shock to ASML and, by extension, to the entire semiconductor capital-equipment complex. That reading deserves scrutiny on three counts.

First, scale. ASML said it plans to produce roughly 130 DUV immersion machines in 2026 and to add about 30% to that in 2027. Five domestic Chinese units this year is not a meaningful share of that. Even 20 units in 2027 would sit well inside the growth ASML expects to add.

Second, capability. SMIC already manufactures seven-nanometer-class chips using ASML’s older deep ultraviolet tools through multi-patterning. What Yuliangsheng’s machine appears to add is not a new process node. It adds a domestic supply path to a capability China already possesses. Analysts quoted by CNBC made this point directly. “I would take this with a pinch of salt as what [China does] could be limited to the very low end,” Stephane Houri, head of equity research at ODDO BHF, told the network in its explainer on the reported breakthrough.

Third, the gap that actually matters. Advanced AI logic requires extreme ultraviolet lithography, which remains an ASML monopoly and which export controls place beyond China’s reach. Mastering immersion DUV does not close that gap; it insulates China against sanctions on the tools it already uses. Those are different things with different investment implications.

None of that means the market reaction was irrational. A domestic Chinese DUV supply chain, even a small one, changes the long-run terminal value assumption embedded in ASML’s China revenue and in the broader equipment complex. It also removes a policy lever Washington has relied on. But the move in equipment stocks on July 27 and 28 priced a structural threat off a report describing five machines. The available evidence suggests the market repriced a narrative faster than it repriced any near-term cash flow.

CXMT: a $488 billion signal about Chinese memory ambition

The second catalyst arrived through the equity market rather than the trade press. CXMT, based in Hefei, priced its Shanghai STAR Market listing at 8.66 yuan per share and closed its first session at 49 yuan — a 466% gain that made it, by CNBC’s account of the debut, the most valuable China-listed company at roughly 3.3 trillion yuan, or about $488 billion. Reported IPO proceeds vary across outlets between roughly $8.6 billion and $9.8 billion depending on whether over-allotment is included; readers should treat the precise figure as unsettled. Either way it was Asia’s largest listing of 2026 and the largest semiconductor listing in STAR Market history.

CXMT held a 7.67% share of the global DRAM market in 2025, according to its own IPO prospectus — a company disclosure rather than an independent measurement. That share is not large enough to break the Samsung–SK Hynix–Micron oligopoly today. What unnerved investors was the combination: a domestic tool supply chain, a domestic memory champion with an enormous new war chest, and a Chinese equity market willing to fund both at valuations that make Western capital cost look expensive.

Memory is a commodity business dressed in growth-stock clothing. Its economics depend almost entirely on supply discipline. Three players who all remember the 2018–2019 and 2022–2023 downcycles can maintain that discipline. A fourth player with state backing, a strategic mandate and no obligation to earn a return on capital cannot be relied upon to do so. That is the real content of the CXMT scare, and it is a genuine long-cycle risk — even though nothing in CXMT’s current capacity threatens 2026 or 2027 DRAM pricing.

The AI-debt strand

The third catalyst had nothing to do with China. Bloomberg framed Tuesday’s rout as deepening on “AI debt jitters” alongside Chinese competition, and that framing captures something the China headlines obscure.

The four largest hyperscalers — Alphabet, Amazon, Meta and Microsoft — are on track for combined capital expenditure approaching $725 billion in 2026, up roughly 77% from about $410 billion in 2025. The mechanics of how that gets funded have changed materially. FactSet’s research team found that incremental annual debt rose from about 9% of capex in fiscal 2024 to roughly 32% on a trailing-twelve-month basis by mid-2026, as internal cash flow stopped scaling with spending. Capital intensity at these companies now runs at levels that would once have been associated with telecom operators or utilities, not asset-light software franchises.

That shift changes the risk character of the trade. When capex is funded from operating cash flow, a demand disappointment produces slower growth. When it is funded from debt, project finance and leasing structures, a demand disappointment produces slower growth and a refinancing question. Investors have begun charging for that distinction, and memory — the most capex-levered, most price-sensitive link in the AI supply chain — is where the charge shows up first.

Why Korea broke first, and why that is not only a Korean problem

The Kospi’s 10.8% single-day loss was the most violent expression of a global repricing, and the reasons are structural rather than sentimental.

Samsung Electronics and SK Hynix together account for roughly half the Kospi’s total market capitalization. That is a concentration that no major developed-market index carries. When memory sells off, the Korean benchmark is not diversified against it in any meaningful sense — it is a leveraged memory index with a few banks and carmakers attached.

Layer onto that the retail leverage. Korean retail margin debt stood at roughly 37.74 trillion won as of early June, near historical highs, and reports out of Seoul indicated that roughly 120,000 retail accounts faced margin calls during the meltdown. Margin calls do not care about valuation; they force selling into falling prices, which forces more selling. The July 28 halt was the eighth Kospi circuit-breaker activation of 2026 and the 14th on record, following halts on March 4 and 9, June 8, 23 and 26, and July 7 and 13. A market that trips its emergency brake eight times in seven months is telling you something about the stability of the positioning underneath it, not just about the news flow above it.

The reason this matters beyond Seoul is the SK Hynix listing. On July 10, 2026, SK Hynix priced 177.9 million American depositary shares at $149 each, raising about $26.5 billion in what was reported as the largest U.S. listing ever by a foreign company, surpassing Alibaba’s 2014 debut. The book was reportedly oversubscribed sevenfold.

Eighteen days later, those ADSs closed at $143 — below the offer price. An IPO breaking issue is a common enough event. An IPO of that size breaking issue within three weeks, in the middle of what its underwriters described as a structural memory supercycle, is a more pointed signal about where marginal demand for the trade actually sits.

What the Schwab strategists argued, and how the argument holds up against the data

On Schwab Network’s Morning Trade Live on July 28, Kasey McCurdy, chief portfolio strategist at Schwab Wealth Advisory, and Joe Mazzola, Schwab’s head trading and derivatives strategist, made a set of specific, testable claims about the session. Worth noting at the outset: Schwab Network is affiliated with a brokerage and asset-management business, and both speakers work for a firm whose revenue depends on client trading and advisory relationships. That does not make their analysis wrong. It does mean the claims are worth checking rather than repeating.

Claim one: the index looks calm even when the market isn’t

McCurdy’s framing was that many clients experience the market through the index level, and that the index level is currently a poor proxy for what is happening to individual holdings. He pointed to top-performing stocks of 2026 sitting 20%, 30% and in some cases close to 50% below their highs while still ranking among the year’s best performers.

The data supports this strongly. The Philadelphia Semiconductor Index peaked at 14,655 in June and traded down to 11,194.60 in July — a drawdown of roughly 24% inside a matter of weeks. Micron, up roughly 229% year to date through July 20 by one market tally, fell 8.9% in a single session on July 28. Meanwhile the S&P 500 closed June 30 at 7,499.36 and finished July 28 at 7,428.78, a move of less than 1% across nearly a month that contained all of this.

The most direct confirmation comes from the options market. Cboe’s DSPX Index — a 30-day measure of expected dispersion among S&P 500 constituents, derived from index and single-stock option prices using a modified VIX methodology — jumped to a six-year high of 47% in the week ending July 10, according to Cboe’s Macro Volatility Digest published July 13 by Mandy Xu. That reading exceeded the peak recorded during the April 2025 “Liberation Day” selloff, when the VIX touched 60. In other words: option markets are pricing more disagreement among individual stocks now than they did during a genuine index panic, while the VIX itself sat at 15%.

McCurdy’s claim is not a talking point. It is arguably the single most measurable fact about this market.

Claim two: breadth is strong beneath the surface

Mazzola cited advancers outpacing decliners by nearly three to one in the S&P 500 during Tuesday’s session — an intraday characterization made on air, not a closing statistic, and one this article has not independently verified against exchange tape data. The closing evidence, however, points the same direction. Twenty-six of the Dow’s 30 components finished higher. The equal-weighted S&P 500 closed at a record. Health care and financials sector funds reached record territory while the technology sector fund fell to a multi-month low.

Cboe’s data adds independent support from a different angle. In the July 20 Macro Volatility Digest, Xu noted that five S&P sectors rose during a week in which the broad index fell, as investors moved from expensive technology into cheaper value names in energy, financials and staples. That is the textbook signature of rotation.

There is a caveat the on-air discussion did not raise. The Russell 2000 and the S&P 500 each gained about 0.2% on July 28, a fraction of the Dow’s 1.03%. Six Dow components supplied roughly 460 of the index’s 537 points, and only three of those six reported earnings that morning. One decliner, Caterpillar, subtracted close to 193 points on its own. Breadth was positive. It was not overwhelming, and a good part of the Dow’s headline number was an artifact of price weighting rather than a referendum on the economy.

Claim three: leveraged single-stock ETFs amplified the move

This was Mazzola’s most specific mechanical claim, and it is the one with the most direct supporting evidence. The proliferation of leveraged and inverse single-stock products around the memory complex in 2026 has been extraordinary. Bloomberg reported on July 1 that a $13 billion leveraged ETF was driving volatility in a key AI memory stock. Within days of SK Hynix’s U.S. debut, at least six leveraged and inverse products tracking the new ADSs launched. At least nine more tied to Japan’s Kioxia were reported to be preparing to list, with some expected as early as August.

The mechanics are not controversial. A 2x daily leveraged fund must rebalance its exposure at the close to maintain its stated multiple. When the underlying falls, the fund must sell. The larger the asset base relative to the underlying’s tradable float, the more that mandatory selling moves the price — and the more it moves the price, the more the fund must sell. In a rising market this works in reverse and looks like genius. In a falling one it is a mechanical accelerant with no view on valuation.

Mazzola’s practical conclusion — that traders chasing winners need tighter stops because the reversals arrive fast — follows from the structure rather than from any market forecast. It is worth separating that observation from his broader question, which he correctly declined to answer: whether the momentum unwind becomes a broad-based selloff depends on whether investors who have profited from concentration are willing to add at lower prices or are already fully committed. Nobody has that data in real time.

Dispersion, explained: how a market can be violent and quiet simultaneously

The concept sitting underneath this entire episode is dispersion, and it is worth a short explanation because it is doing the analytical work.

Index volatility measures how much the S&P 500 moves. Single-stock volatility measures how much its members move. The difference between them is dispersion. When stocks move together — everything up, everything down — index volatility approaches average single-stock volatility and dispersion is low. When stocks move in opposite directions, their moves offset inside the index and dispersion is high. The index can sit still while enormous amounts of capital change hands underneath.

That is precisely the July 2026 condition. Cboe’s July 20 digest quantified how extreme it has become. One-month implied volatility on SMH, the VanEck semiconductor ETF, jumped five points to a one-year high of 59%. One-month implied volatility on QQQ rose 3.8 points. The SMH-versus-SPX one-month implied volatility spread widened to a record 44%, described by Cboe as more than five standard deviations above its historical average. Meanwhile Russell 2000 one-month implied volatility rose just one point to 19% and remained in the 15th percentile of its trailing-year range. The RTY-QQQ volatility spread fell to a five-year low of negative 6.6%, against a long-run average of positive 1.4% — meaning small caps, historically the more volatile asset, are now priced as calmer than large-cap technology.

For an investor, the practical implication is uncomfortable. A portfolio benchmarked to the S&P 500 has looked stable in 2026. A portfolio concentrated in the winners of 2023 through 2025 has not. The index has been providing false comfort to exactly the investors least entitled to it.

There is one further detail in the Cboe data that cuts against the sanguine reading. In the week ending July 17, S&P 500 one-month skew — a measure of how much more investors are paying for downside protection than for upside exposure — surged from the 14th percentile to the 82nd percentile. Institutions were buying index insurance even as the index itself held up. Dispersion explains why the market looks calm. Skew suggests that professional investors do not believe it will stay that way for free.

The Magnificent Seven’s lost year

To understand why rotation has been able to absorb a semiconductor rout without breaking the index, you have to look at what the market’s largest stocks have already done in 2026 — because the answer is: not much, and mostly downward.

By one widely cited tally reported in late July, the Magnificent Seven were down roughly 3.7% for 2026 while the other 493 S&P 500 constituents were up about 12.9%, a gap of nearly 17 percentage points. The equal-weighted S&P 500 has outperformed the capitalization-weighted version by roughly two percentage points on the year. Goldman Sachs had forecast at the start of 2026 that the Magnificent Seven would trail the equal-weighted index, citing diverging AI strategies and rising stock-level dispersion — a call that has, so far, worked.

The dispersion inside the group is more striking than the group average.

The Microsoft number is the one that should stop readers. A company that was the second most valuable in the world entering 2026 has fallen to fourth, behind Apple at roughly $5 trillion, Nvidia at roughly $4.7 trillion and Alphabet at roughly $3.9 trillion. Reported year-to-date declines cluster between 19% and 23% depending on the measurement date; by one account the stock closed at $389.10 on July 27, down 19.54% for the year.

That did not happen because Azure stopped growing. Reporting on the decline indicates Azure reaccelerated to roughly 40% growth and that AI-attributed revenue was running at an annualized rate around $37 billion. It happened because the market re-rated the cost of getting there. Microsoft is projected to spend roughly $190 billion on capital expenditure in calendar 2026, up about 61% year over year, and analysis of its commercial backlog indicates that roughly 45% of a reported $627 billion figure is tied to a single counterparty — OpenAI — which is actively diversifying its cloud arrangements. Those are estimates drawn from analyst work and reporting rather than audited disclosures, and readers should treat them accordingly. But directionally they explain the de-rating: investors spent 2024 and 2025 pricing AI revenue as if it would arrive quickly and cleanly, and are now pricing a multi-year infrastructure cycle with heavy front-loaded cash costs and meaningful customer concentration.

Apple’s ascent is the mirror image. Apple has comparatively modest AI capital commitments, an enormous installed base, and a business model that benefits from AI deployed on devices it already sells rather than from data centers it has to build. In a market rotating away from capital-intensive AI infrastructure, that profile has been rewarded — which is why Apple, not Nvidia, touched $5 trillion on July 28.

Concentration math: why 43% at the top changes what “the index” means

The top 10 S&P 500 constituents now account for approximately 43% of the index’s market capitalization, a record. RBC Wealth Management has documented this trend under the label “the great narrowing,” noting that the historical average between 1990 and 2015 ran between roughly 18% and 23%. The top-10 weight has close to doubled in a decade.

Concentration is not automatically dangerous. It reflects a real economic fact: a small number of platforms have captured an outsized share of profit growth. But it changes the statistical properties of the index in ways that matter for anyone using it as a risk benchmark.

At 20% top-10 weight, an index is a diversified claim on corporate America with a large-cap tilt. At 43%, it is a concentrated bet on ten businesses with a diversified hedge attached. The correlation structure inside a portfolio built to track it is therefore not what the number of holdings implies. Investors who own an S&P 500 fund, a Nasdaq-100 fund and a technology sector fund believe they hold three positions. They largely hold one, three times.

This is the practical content of McCurdy’s warning. The unusual feature of 2026 is that concentration has, for once, worked in investors’ favor at the index level — because the concentrated names have been flat to down while the remaining 493 have carried the benchmark. That is not how the risk usually resolves. It is a reminder that concentration cuts in both directions, and that a year in which the biggest stocks lag is not evidence that concentration risk has disappeared. It is evidence that this particular episode happened to be absorbable.

The memory supercycle is real. That is exactly why it is dangerous.

It would be easy to read the past week as a bubble deflating. The earnings say otherwise, and the distinction matters.

SK Hynix reported second-quarter 2026 results on July 29, and the figures are among the most extreme ever posted by a large manufacturer. Per the company’s own preliminary earnings release, revenue reached 79.3187 trillion won, operating profit 60.5426 trillion won and net profit 93.9226 trillion won. Operating margin was 76%. Revenue rose 51% sequentially and 257% year over year; operating profit rose 557% year over year. First-half revenue crossed 100 trillion won for the first time in company history. Cash and equivalents reached 88 trillion won, up 33.6 trillion won in a single quarter, while total debt fell to 18.6 trillion won for a net cash position of 69.4 trillion won.

Two accounting notes deserve flagging. Net profit exceeding operating profit — producing a reported net margin of 118% — implies substantial non-operating income, and the company itself states the figures are preliminary, prepared under K-IFRS, and subject to change during the independent audit. Readers should not treat the net income line as final.

The operational detail matters as much as the headline. HBM4 entered mass shipment in the second quarter, with production ramping in the second half. HBM4E samples completed shipment in the first half. The company finalized long-term agreements with around 10 customers to secure multi-year supply visibility — a structural change from an industry that historically sold spot. Sales of SOCAMM2 grew significantly, and 321-layer NAND already represents the largest share of production, targeted to reach roughly 50% of domestic capacity by year-end.

Capacity expansion is accelerating: the M15X ramp is being pulled forward, Yongin Phase 1’s cleanroom opens in early 2027, and the P&T7 advanced packaging facility and M17 NAND base are queued behind it. Reported full-year 2026 capital expenditure guidance sits in the high 40 trillion won range. Notably, SK Hynix framed all of this under the banner of “CapEx discipline” — language that sits somewhat awkwardly beside a nearly 50 trillion won spending plan, and which investors should read as management signaling awareness of the cycle risk rather than as evidence of restraint.

Micron’s trajectory has been similar in shape. Reporting on its fiscal third quarter described revenue of $41.5 billion, up 74% sequentially and 346% year over year, with DRAM contract prices reported up roughly 90% in the first quarter of 2026. Those figures come from secondary coverage rather than from the filing itself and should be verified against Micron’s own disclosures before being relied upon.

Here is the tension at the center of the whole trade. Memory is now generating margins that no commodity industry sustains indefinitely. A 76% operating margin is an invitation — to competitors, to customers seeking second sources, to governments building strategic capacity. SK Hynix is spending nearly 50 trillion won to add capacity. Samsung and Micron are doing the equivalent. CXMT has just raised roughly $9 billion from public markets to do the same with state backing.

Every memory downcycle in history has been manufactured by exactly this: extraordinary profits attracting capacity that arrives after demand normalizes. The bull case is that AI demand is structural and that long-term agreements with hyperscalers change the industry’s behavior permanently. The skeptical case is that the industry has told itself a version of that story before. Neither will be settled by a four-day selloff.

The Fed walks into this with a hawkish problem

The Federal Open Market Committee announces at 2:00 p.m. EDT on Wednesday, July 29, with Chair Kevin Warsh’s press conference at 2:30 p.m. EDT. The target range for the federal funds rate stands at 3.50%–3.75%, where it has been since December 2025. This is one of the four 2026 meetings without a Summary of Economic Projections, so no updated dot plot accompanies the statement.

The unusual feature is the direction of the debate. Economists polled by FactSet expect a hold, which would be the fifth consecutive meeting without change. But the June meeting’s dot plot showed a median year-end 2026 rate of 3.8%, up from 3.4% in March — a shift from an implied cut to an implied hike. Forbes reported on July 23 that markets were pricing rising odds of a July increase. Cleveland Fed President Beth Hammack has argued the Fed may need to raise rates to address persistent inflation, and Governor Christopher Waller’s recent remarks have been read as a decisive shift in emphasis from labor-market risk to inflation containment. Reuters-sourced coverage put implied odds of a 25-basis-point increase at roughly 30% ahead of the meeting — market pricing, not Fed guidance.

The inflation picture is genuinely mixed, which is why the debate is live. Headline CPI fell to 3.5% year over year in June, down from 4.2% in May and below forecasts of 3.8% — the first decline in five months. Core CPI was flat month over month at 0.0% against expectations of 0.2%, with the annual rate at 2.6%. That looked like a decisive turn.

The problem is composition. Essentially all of the June relief came from energy, where costs rose 15.7% year over year against 23.5% in May as the U.S.–Iran ceasefire eased pressure. Strip energy out and the disinflation largely disappears. The June FOMC materials showed PCE inflation at 3.6% and core PCE at 3.3%, both well above the 2% target. And policymakers have flagged a second inflation channel that has nothing to do with oil: the electricity and construction cost pressure generated by the AI data-center buildout itself.

The transmission to equities is more complicated than the usual rate story. A hawkish Fed raises the discount rate applied to long-duration cash flows, which hurts exactly the high-multiple AI names already under pressure. But it also removes a support that value and cyclical sectors do not need as badly, since their earnings arrive sooner. That asymmetry is one reason the rotation has run as far as it has, and it is why a hawkish surprise on Wednesday afternoon would likely deepen the sector divergence rather than reverse it.

A dovish surprise would do the opposite — and, counterintuitively, might be the more destabilizing outcome for the rotation trade, because it would remove the rate-differential logic that has driven money out of growth and into value since the spring.

Oil, the Strait of Hormuz, and the inflation channel nobody controls

The Fed’s problem in 2026 has been substantially an energy problem, and the energy problem has been a geopolitics problem.

The Strait of Hormuz has been effectively disrupted since late February 2026 following the outbreak of U.S.–Iran hostilities. A memorandum of understanding signed June 18 briefly reopened flows and pushed oil sharply lower — Al Jazeera reported oil falling and equities rallying on the framework announcement. That ceasefire then deteriorated in July, with Brent topping $90 on July 19 as Houthi attacks on tankers opened a new front, and briefly trading above $100 for the first time since 2022.

Then it collapsed again. Brent settled around $84.09 on July 28, down 4.8% on the day, with West Texas Intermediate near $79.26, down about 4%. CNBC reported prices sliding as a pause in U.S.–Iran hostilities appeared to hold, with Iranian Foreign Minister Seyyed Abbas Araqchi holding separate calls with Saudi and Omani counterparts about Hormuz security. Bloomberg described the move as oil’s worst three-day stretch in more than six years.

A 20% round trip in the world’s benchmark crude inside ten days is not a functioning price signal. It is a market trading headlines about a negotiation whose outcome nobody can forecast. For the Fed, that means the single largest input to its inflation forecast is effectively a coin flip on diplomacy. For equity investors, it means the “inflation is falling” case that supported multiple expansion in July rests on a foundation that could reverse within one trading session.

It also complicates the read on consumer demand. The Conference Board’s Consumer Confidence Index fell 1.4 points to 90.8 in July from an upwardly revised 92.2 in June, below the 92.3 consensus. The Present Situation Index dropped 3.6 points to 114.9, its third consecutive monthly decline, while the Expectations Index held at 74.7 — still in territory the Conference Board associates with recession signals. Chief Economist Dana M. Peterson noted confidence has been on a “general downward sloping trajectory since late 2021.” The survey period ran July 1–22, covering the oil spike but not its subsequent collapse. Respondents cited grocery and fuel costs.

Currencies: why “risk-off” did not produce a stronger yen

A conventional risk-aversion episode produces a familiar currency pattern: dollar and Swiss franc up, Australian dollar and Korean won down, yen up. July 2026 has broken the last of those.

USD/JPY printed a fresh 52-week high of 163.24, a level not seen since the mid-1980s. The arithmetic is straightforward: the Fed’s policy rate sits at 3.50%–3.75% while the Bank of Japan’s overnight guideline is around 1.00%, leaving a differential of roughly 250 to 275 basis points. That gap has overwhelmed the yen’s traditional safe-haven behavior. ING’s currency team has described the pair as having gone “back to the 1980s” — and the yen has been unable to strengthen even while the BOJ conducts the most significant normalization in its modern history.

Japanese authorities reportedly sold just over $70 billion in late April and early May at levels just above 160. With the pair now above 163 and the BOJ meeting July 30–31, intervention risk is elevated. This is not an ordinary technical level. It is a policy zone where price action can become discontinuous, and where a stop-loss order may not fill anywhere near where it was placed.

Two practical implications follow. First, the traditional bond-equity-currency hedging relationships are not functioning normally; investors reaching for yen exposure as a risk offset are also taking a large carry loss and an intervention risk. Second, a disorderly unwind of yen-funded carry positions remains one of the more plausible mechanisms by which a sector selloff could become a cross-asset event — as it did in August 2024, when a modest BOJ move triggered a global deleveraging cascade far larger than the policy change warranted.

Elsewhere, the Australian dollar has weakened on softer domestic inflation that reduced expectations of further Reserve Bank of Australia tightening, compounded by exposure to Chinese and Asian risk sentiment. The Korean won faces direct pressure from the equity rout. Sterling has been soft ahead of a Bank of England decision on July 30, with Bank Rate at 3.75%.

Four earnings reports that will settle more than the quarter

Microsoft and Meta Platforms report after the close on Wednesday, July 29. Apple and Amazon follow on Thursday, July 30. Between them they represent an enormous share of index weight, and — more importantly for this particular market — the entirety of the demand assumption underpinning the memory and semiconductor complex.

Consensus expectations reported ahead of the releases include adjusted earnings per share of roughly $4.24 for Microsoft on revenue of about $87.62 billion, representing growth of roughly 16.2% and 14.6% respectively; Meta earnings per share in the $7.18–$7.24 range; Apple fiscal third-quarter earnings per share around $1.86; and Amazon around $1.85. These are analyst consensus figures compiled by third parties, not company guidance, and consensus composition varies by provider.

The numbers themselves are close to irrelevant. What matters is the capital expenditure commentary. Combined 2026 AI capex plans across the four approach $725 billion, up roughly 77% from about $410 billion in 2025. Every dollar of that budget is a dollar of forward revenue for the memory, networking, power and equipment suppliers whose shares just fell 20% to 25% from their June highs.

There are three plausible outcomes, and they point in very different directions.

If capex guidance rises, the semiconductor selloff looks like a positioning washout and the memory names likely retrace quickly — the demand thesis will have been confirmed by the customers themselves. If capex guidance holds but commentary emphasizes efficiency, discipline or optimization, the market will read a plateau, and the derating in equipment and memory continues at a slower pace. If any of the four trims capex or signals a pause, the past week stops being a rotation story and becomes the opening of a genuine repricing across the AI supply chain.

Microsoft carries the heaviest burden, because it enters the print already down roughly 20% for the year with the market openly questioning the return profile of a $190 billion spending program and the concentration of its backlog. A reassuring quarter from Microsoft would do more to stabilize the AI complex than any Fed outcome. A disappointing one, delivered into a market that has just watched the Kospi trip a circuit breaker, would land on unusually thin ice.

The strongest case that this is healthy

The constructive interpretation, which McCurdy articulated on air and which the data substantially supports, runs as follows.

For three years, index returns depended on a handful of names. Analysts warned repeatedly that this was fragile. In 2026, the market has been running the experiment: the Magnificent Seven have gone nowhere or fallen, and the index has still produced a positive return because 493 other companies picked up the load. The equal-weighted S&P 500 hitting a record high on the same day the semiconductor index fell 4.5% is not a coincidence. It is the mechanism working.

Rotation without exit is the healthiest form of a drawdown. Investors are moving from expensive assets to cheaper ones inside the equity market rather than moving to cash. Cboe’s data confirms five sectors rising in a down week for the index. Health care and financials sector funds set records on July 28. Small caps have outperformed and their implied volatility sits in the 15th percentile of its trailing-year range — hardly the profile of a market bracing for recession.

The volatility evidence is consistent with repositioning rather than stress. The VIX at 18.21 is elevated relative to the mid-teens of early July but nowhere near crisis territory. Credit spreads have not been reported as materially widening. The dollar has been firm but not spiking in the way that signals a scramble for liquidity.

And the fundamental backdrop supporting the rotation is real. Sherwin-Williams raised full-year guidance. Coca-Cola lifted its comparable earnings growth outlook to 9%–10% from 8%–9%. Boeing’s deliveries rose 14% to 171 aircraft despite a reported core loss including a $280 million charge on the Air Force One program. Money is not fleeing to defensive assets out of fear; it is buying earnings growth that had been ignored while everything AI-adjacent commanded a premium.

The strongest case that this is not

The skeptical reading has equally specific evidence behind it, and dismissing it requires ignoring several things at once.

Start with the options market, which is where professional risk is priced. Index skew moving from the 14th percentile to the 82nd percentile in a single week is not a rotation signal. It is institutions buying downside protection on the whole market while the whole market holds up. The SMH-SPX volatility spread at a record 44%, more than five standard deviations above average, describes a market that has isolated a hazard rather than eliminated it. Isolation works until correlations rise, and correlations rise precisely when they are least convenient.

Next, the mechanical fragility. Leveraged single-stock products around memory names have grown to a scale where their rebalancing flows are a meaningful share of daily volume. Korean retail margin debt near 37.74 trillion won produced roughly 120,000 margin calls in one session. Eight Kospi circuit breakers in seven months is a structural signal, not a run of bad luck. These are amplifiers, and they are indifferent to whether the initiating shock is large or small.

Then the concentration arithmetic. A 43% top-10 weight means the index cannot absorb a genuine repricing of its largest constituents no matter how well the other 490 perform. What happened in July is that the AI infrastructure names fell while the AI platform names — Apple, Alphabet — held up. If the earnings reports on July 29 and 30 impair the platform names too, the offsetting mechanism disappears. Rotation requires somewhere to rotate to.

Consider also what has already happened without anyone calling it a bear market. The Nasdaq-100 has entered correction territory, reaching a 10% drawdown in 38 trading days. The semiconductor index fell roughly 24% from its June peak. The largest foreign IPO in U.S. history broke issue within three weeks. Microsoft, one of the two or three most-owned stocks on earth, is down roughly 20% year to date. These are not the statistics of a market in a summer lull. They are the statistics of a bear market in one sector that the index has successfully camouflaged.

Finally, the macro overlay is not supportive. The Fed is debating a hike, not a cut. Consumer confidence has fallen for three consecutive months on the present-situation measure. Oil has moved 20% in ten days on diplomatic headlines. The yen sits at a level that invites intervention two days before the Bank of Japan meets. Any one of these is manageable. The combination reduces the market’s capacity to absorb a fifth surprise.

Historical comparisons, and where each one breaks down

Three episodes get cited most often in discussions of the current tape. Each illuminates something, and each misleads in a specific way.

2000. The comparison is to a concentrated technology complex trading at extreme multiples on capital spending that turned out to be forward-pulled. The similarity is genuine: telecom operators in 1999 and 2000 were building capacity against demand forecasts that did not materialize, financed increasingly with debt. Where it breaks down is profitability. The AI infrastructure buildout is being funded by companies generating enormous operating cash flow, and the suppliers are posting 76% operating margins on shipped product. The 2000 comparison describes a possible ending, not the current condition.

2021–2022 growth-to-value rotation. The similarity is the mechanism: rising rates compressing long-duration multiples while cyclical and value earnings hold up. The difference is the starting point. In late 2021 the entire market was expensive and the rotation coincided with an index decline of more than 20%. In 2026 the rotation has occurred with the index roughly flat, because the value side was cheap enough to absorb the flow. That is a materially better setup.

August 2024 yen carry unwind. This is the most relevant structural analogue and the least discussed. A modest Bank of Japan policy adjustment triggered a violent global deleveraging because leveraged positions funded in yen had to be unwound simultaneously. The mechanism — cheap funding currency, crowded positioning, forced liquidation — is present again, with the added ingredient of leveraged single-stock ETFs that did not exist at anything like the current scale. The BOJ meets July 30–31. That is the calendar risk most likely to convert a sector story into a cross-asset one.

A fourth comparison deserves mention because the Cboe data invokes it directly: April 2025, when the “Liberation Day” tariff selloff drove the VIX to 60. Dispersion in July 2026 exceeded the level reached during that episode — while the VIX sat at 15%. Whatever is happening now is not a repeat of April 2025. It is a structurally different kind of stress, one that lives inside the index rather than on top of it.

Material risks from here

• Capex guidance risk. Any of Microsoft, Meta, Apple or Amazon trimming or pausing AI capital spending would directly impair the revenue assumptions embedded in memory, networking and equipment valuations. This is the single largest identifiable risk in the next 48 hours.
• Policy risk. A Fed hike, or a hold accompanied by explicitly hawkish language, would raise the discount rate applied to the assets already under most pressure. The Bank of England decides July 30; the Bank of Japan July 31.
• Currency and intervention risk. USD/JPY above 163 with an intervention-sensitive authority meeting within days creates the possibility of discontinuous price action and a forced unwind of yen-funded positions.
• Energy and geopolitical risk. The U.S.–Iran situation has reversed twice in six weeks. Renewed Hormuz disruption would mechanically reverse the June inflation improvement inside a single CPI print.
• Leverage and market structure risk. Leveraged single-stock ETFs, retail margin debt in Korea and elsewhere, and concentrated index construction all amplify rather than dampen shocks. None of these were designed with a simultaneous unwind in mind.
• Memory cycle risk. Extraordinary margins are attracting capacity from all four major producers plus state-backed Chinese entrants. Historically, memory downcycles are manufactured during upcycles.
• Concentration risk. A record 43% top-10 index weight limits how much a broad rotation can offset if the largest names decline together.
• Consumer risk. Three consecutive monthly declines in the Conference Board’s Present Situation Index, with the Expectations Index at 74.7, suggests the demand side is not as robust as index levels imply.

What to watch, with dates

• Wednesday, July 29, 2:00 p.m. EDT: FOMC statement; 2:30 p.m. EDT, Chair Warsh’s press conference. No Summary of Economic Projections at this meeting.
• Wednesday, July 29, after the close: Microsoft and Meta Platforms fiscal results, with capital expenditure guidance the decisive variable.
• Thursday, July 30, 12:00 noon London time: Bank of England decision, minutes and Monetary Policy Report. Bank Rate currently 3.75%.
• Thursday, July 30, after the close: Apple and Amazon fiscal results.
• Friday, July 31: Bank of Japan decision and Outlook Report, following the July 30–31 meeting. Overnight rate guideline approximately 1.00%.
• Ongoing: Whether Korean and Japanese semiconductor shares hold the July 29 rebound; whether the equal-weighted S&P 500 sustains its record; whether credit spreads widen; and whether the Nasdaq-100 recovers without the Dow surrendering its gains.

Frequently asked questions

Why did semiconductor stocks fall so sharply in late July 2026?

Three things converged. On July 27, The Information reported that China had begun mass-producing domestic immersion DUV lithography tools, which hit ASML and the chip-equipment complex. The same day, Chinese memory maker CXMT surged 466% on its Shanghai debut, raising competitive concerns for Micron, Samsung and SK Hynix. Separately, investors have been reassessing the debt-funded portion of hyperscaler AI capital spending. Leveraged single-stock ETFs and Korean retail margin debt amplified the resulting selling.

Did the S&P 500 fall during the semiconductor selloff?

No. The S&P 500 closed up 0.21% at 7,428.78 on July 28, 2026, and the Dow rose 1.03% to 52,747.32. Money rotated into health care, financials, industrials and staples rather than leaving equities. The equal-weighted S&P 500 set a record high the same day.

What is the Cboe DSPX index and why does it matter now?

DSPX measures expected dispersion among S&P 500 constituents over the next 30 days, derived from index and single-stock option prices. High dispersion means individual stocks are expected to move in different directions, offsetting each other inside the index. It reached a six-year high of 47% in July 2026 while the VIX sat at 15% — quantifying exactly why the index has looked calm while individual holdings have not.

Has the Federal Reserve raised interest rates?

Not as of this article’s publication. The target range is 3.50%–3.75%, unchanged since December 2025, and the July 29 decision had not been announced. Economists polled by FactSet expected a hold, though market pricing implied roughly 30% odds of a 25-basis-point increase and several policymakers have publicly argued for a hike.

Why did the Kospi fall more than 10% in one day?

Samsung Electronics and SK Hynix together account for roughly half of the Kospi’s market capitalization, so a memory selloff hits the Korean benchmark almost undiluted. Retail margin debt near 37.74 trillion won produced forced selling, with reports of roughly 120,000 accounts facing margin calls. The July 28 halt was the eighth circuit-breaker activation of 2026.

Is SK Hynix trading below its U.S. IPO price?

Yes. SK Hynix priced 177.9 million American depositary shares at $149 on July 10, 2026, raising about $26.5 billion. The ADSs closed at $143 on July 28, below the offer price, despite the company reporting record quarterly results on July 29.

How much are the big technology companies spending on AI in 2026?

Alphabet, Amazon, Meta and Microsoft are on track for combined capital expenditure approaching $725 billion in 2026, up roughly 77% from about $410 billion in 2025. Microsoft alone is projected at roughly $190 billion, up about 61% year over year. An increasing share is debt-funded rather than financed from operating cash flow.

Why has Microsoft stock fallen in 2026?

Reported year-to-date declines cluster between 19% and 23%. The drivers cited in coverage are the scale and cash-flow impact of AI capital spending, slower-than-hoped monetization of Copilot, and concentration in its commercial backlog — analysis suggests a large share is tied to OpenAI, which is diversifying its cloud arrangements. Azure growth itself has been reported as reaccelerating, which is why the decline reflects a change in how investors value the spending rather than a collapse in the business.

Does China’s DUV development threaten ASML?

Not materially in the near term, based on the reported figures. The Chinese effort targets roughly five immersion systems in 2026 and about 20 in 2027 at 28 nanometers, against ASML’s plan for approximately 130 DUV immersion machines in 2026. The tools also do not approach extreme ultraviolet capability, which remains required for leading-edge AI logic. The longer-term risk is to ASML’s China revenue base and to the leverage of export controls.

Is the memory supercycle over?

Nothing in the current data says so. SK Hynix reported record second-quarter revenue of 79.3 trillion won with a 76% operating margin and raised capacity plans, and signed long-term agreements with around 10 customers. The risk is the opposite of a demand collapse: margins at these levels attract capacity from all four major producers plus state-backed Chinese entrants, which is historically how memory downcycles get built.

What would turn this rotation into a broad selloff?

Watch for selling spreading to non-technology sectors, credit spreads widening, further deterioration in the Korean won or Australian dollar, sustained equity weakness accompanied by a stronger dollar, and volatility that persists rather than reversing intraday. A capex cut from a major hyperscaler or a disorderly yen move around the Bank of Japan meeting are the most plausible triggers.

What should investors watch first this week?

The Fed statement and press conference on July 29, then Microsoft’s and Meta’s capital expenditure commentary the same evening, then Apple and Amazon on July 30, then the Bank of Japan on July 31. Whether Asian semiconductor shares hold their July 29 rebound is the cleanest single indicator of whether the past week was positioning or repricing.

Final assessment

The most defensible reading of late July 2026 is that the market has successfully quarantined a serious sector event — and that the quarantine has not yet been tested by anything that touches the whole market at once.

The evidence for a healthy rotation is strong and specific. The equal-weighted S&P 500 set a record on the day the semiconductor index fell 4.5%. Five sectors rose during a week the index fell. Health care and financials reached record territory. The VIX stayed at 18. Capital moved between assets rather than out of them, and it moved toward companies raising guidance rather than toward cash. That is what a functioning market looks like when it revalues one part of itself.

The evidence for concern is equally specific and lives mostly in the plumbing. Index skew jumped from the 14th to the 82nd percentile in a week — institutions bought protection on the whole market while the whole market held up. The semiconductor-versus-index volatility spread hit a record more than five standard deviations above average. Leveraged single-stock products, Korean retail margin debt and a record 43% index concentration are all amplifiers, and all three were active in the same week. Eight Kospi circuit breakers in seven months is a structural fact, not noise.

What changed over these four sessions is not the AI thesis. SK Hynix just posted a 76% operating margin and raised capacity plans; Micron’s revenue has more than quadrupled year over year; hyperscaler capex is still budgeted at nearly three-quarters of a trillion dollars. What changed is that investors began charging separately for three things they had previously priced as one: AI demand, the capital intensity required to serve it, and the financing structure used to fund that capital. Those are different risks with different owners, and the market spent late July learning to distinguish them.

What remains genuinely uncertain is whether the rotation has anywhere left to go. It has worked so far because the AI platform companies — Apple, Alphabet — held up while the AI infrastructure companies fell. Apple touching $5 trillion on the same day Micron lost 8.9% is the clearest illustration of that split. If the earnings reports on July 29 and 30 impair the platform side as well, the offset disappears and the index stops being able to hide the move.

The practical takeaway for readers is narrower than any market call. The S&P 500’s 2026 performance has been a poor description of what has happened inside most portfolios, and dispersion at a six-year high says that gap is expected to persist for at least another month. Investors who have not looked at their individual holdings because the index looked fine should look. That is not a forecast. It is arithmetic.

This article is provided for general informational purposes and does not constitute financial, investment, tax, or legal advice.

Sources

• SK hynix, “SK hynix Announces 2Q26 Financial Results,” July 29, 2026
• Board of Governors of the Federal Reserve System, FOMC statement, June 17, 2026
• Board of Governors of the Federal Reserve System, FOMC minutes, June 16–17, 2026
• Cboe Macro Volatility Digest, “DSPX Index Jumps to 6-Year High Ahead of Earnings,” Mandy Xu, July 13, 2026
• Cboe Macro Volatility Digest, “Hedging Demand Spikes Amid AI-Driven Market Rotation,” Mandy Xu, July 20, 2026
• S&P Dow Jones Indices, Cboe S&P 500 Dispersion Index methodology and data
• The Information, “China Begins Mass Production of Homegrown DUV Chip Tools,” July 27, 2026
• Bloomberg, “ASML Shares Drop After Report of China Producing DUV Chipmaking Tools,” July 27, 2026
• CNBC, “China’s reported chip breakthrough comes with some big caveats,” July 28, 2026
• CNBC, “Chipmaker CXMT’s 466% market debut surge makes it the most valuable China-listed company,” July 27, 2026
• Bloomberg, “Korean Markets Hit by Turmoil as Chip Rout Forces Trading Halts,” July 28, 2026
• Bloomberg, “Chip Stock Rout Deepens on AI Debt Jitters, China Competition,” July 28, 2026
• Seoul Economic Daily, “KOSPI Triggers Circuit Breaker, 8th This Year,” July 28, 2026
• Korea JoongAng Daily, “Chip-led selloff sends Kospi below 6,300, triggers circuit breaker,” July 28, 2026
• Al Jazeera, “South Korea’s SK Hynix raises $26.5bn in record-breaking US IPO,” July 10, 2026
• CNN Business, “SK Hynix IPO: A once-obscure chip maker has landed the largest US listing by a foreign company,” July 10, 2026
• Bloomberg, “SK Hynix’s US Trading Debut Unleashes Wave of New Leveraged ETFs,” July 10, 2026
• Bloomberg, “How a $13 Billion Leveraged ETF Drives Volatility in a Key AI Memory Stock,” July 1, 2026
• CNBC, “Apple touches $5 trillion market cap for first time,” July 28, 2026
• Forbes, “Apple Briefly Surpasses $5 Trillion Market Value—Joining Nvidia,” July 28, 2026
• CNBC, June 2026 Consumer Price Index report, July 14, 2026
• The Conference Board, “US Consumer Confidence Edged Down in July,” July 28, 2026
• The Conference Board Consumer Confidence Index
• CNBC, “Oil prices slide, Brent crude below $90 as pause to U.S.-Iran hostilities appears to hold,” July 27, 2026
• Al Jazeera, “Oil prices fall, stocks rally as US, Iran sign framework to end war,” June 18, 2026
• Forbes, “Markets Price In Rising Odds Of July Fed Rate Hike,” July 23, 2026
• The Spokesman-Review, “Fed rate-hike voices swell before July decision, rates still seen on hold,” July 17, 2026
• Morningstar, “What to Expect from the July Fed Meeting,” July 2026
• CBS News, “Will the Federal Reserve raise interest rates? Here is what experts predict for July’s meeting”
• FactSet Insight, “Hyperscalers Tap External Financing as AI Capex Outruns Cash Flow”
• RBC Wealth Management, “The ‘Great Narrowing’: S&P 500 concentration”
• Seeking Alpha, “Mag 7 will continue to trail equal-weight S&P this year – Goldman”
• Benzinga, “S&P 493 Hammers Magnificent Seven in the Year of the Underdog,” July 2026
• Investing.com, “Why the Mag 7 stocks are underperforming the S&P 500 in 2026”
• Schwab Center for Financial Research, “Nasdaq Tumbles Early as Chip Selloff Deepens,” July 28, 2026
• Charles Schwab, biography of Joe Mazzola, head trading and derivatives strategist
• Charles Schwab, biography of Kasey McCurdy, CFA, chief portfolio strategist, Schwab Wealth Advisory
• ING Think, “USD/JPY: Back to the 1980s”
• CNBC, stock market live updates for July 28, 2026
• UPI, “South Korea halts stock trading as Kospi plunges more than 8%,” July 28, 2026
• Microsoft Investor Relations, fiscal 2026 fourth-quarter earnings event details

Affiliate disclosure: Businessfinance.news may earn compensation from qualifying actions completed through selected links on this website, at no additional cost to the reader. Affiliate relationships do not influence our editorial reporting, analysis, or conclusions.