Dow (DJI) Climbs to 52,806.31 While Nasdaq (IXIC) 26,054.38 and S&P 500 (SPX) 7,559.99 Fade — TSM Sold Down 1.97% on Record $40.2B Print

The tape delivered the same verdict twice this week and the second reading was uglier than the first. A dominant chip franchise printed a blowout quarter, guided higher, and the sector rolled over inside the first hour. It was the second time in three sessions that a beat-and-raise from a sector giant preceded a pullback across semiconductors. A gauge of semiconductor firms dropped 3% while the Nasdaq 100 shed 1%, and the S&P 500 wavered even as most of its members traded green.

By late morning the scoreboard read: S&P 500 (SPX) at 7,559.99, down 12.41 points or 0.16%. Nasdaq Composite (IXIC) at 26,054.38, down 214.85 or 0.82%. Dow Jones Industrial Average (DJI) at 52,806.31, up 147.67 or 0.28%. The Russell 2000 held 2,988.55, up 12.29 or 0.41%. The VIX ticked to 15.87, up 0.20 or 1.27% — no panic bid, just rotation. Gold got smoked at 4,008.80, off 43.00 or 1.06%.

Set that against Wednesday's close, when everything worked. The S&P 500 finished up 0.38% at 7,572.40. The Nasdaq climbed 0.62% to 26,269.23. The Dow added 150.37 points or 0.29% to 52,658.64. Twenty-four hours later the Dow tacked on almost the identical 147.67 points while the Nasdaq handed back 214.85. Same index complex, opposite engines.

The thesis writes itself. Good news has stopped paying inside the AI trade. Capital is not leaving equities — the Dow is green, small caps are green, breadth is wide — it is leaving the corner of the market where the payback period runs to 2028 and moving into the corner that converts revenue into cash this quarter. A 4.60% ten-year and $84 Brent are the discount-rate mechanism doing that work.

Monday framed the range. The S&P 500 lost 0.79% to 7,515.34, the Nasdaq snapped 1.55% lower to 25,873.18, and the Dow settled down 138.37 points or 0.26% at 52,498.64 after the Hormuz blockade headline hit. From that low the S&P has clawed back 44.65 points. The Nasdaq has recovered 181.20. The Dow has ripped 307.67. That spread is the whole story.

TSMC Printed $40.2 Billion and the Tape Sold It

Taiwan Semiconductor Manufacturing (TSM) posted record second-quarter revenue of $40.2 billion and lifted both its capital spending and its revenue outlook for the year. The figure landed at the absolute ceiling of management's own $39.0–$40.2 billion guidance band and blew through the $39.8 billion consensus mark. Street EPS sat at $3.77, a 52.6% jump on the year-ago print, against revenue growth pegged at 32.2%. The company delivered a 77% annual earnings gain. Gross margin guidance had been staked at 65.5%–67.5%.

The stock got smoked anyway. TSM moved 4% lower before the bell and by mid-session sat at 411.20, down 8.28 or 1.97%, after printing as low as -3.4% in the early tape. Shares are up 40% year to date and holders took the profit. TSM closed at $426.62 into the report with resistance mapped at $437.98 and a 52% year-to-date gain already banked. The stock now sits 15.42 points below that close and 26.78 under the pivot.

The tell was in the guidance language, not the beat. Management warned of rising prices and was explicit about where those prices bite: consumer and price-sensitive end market segments are getting challenged by climbing component costs and macro uncertainty. That is a margin warning wearing a demand jacket. Chairman C.C. Wei paired it with the line the bulls wanted — conviction in the multi-year AI megatrend remains very high — and the tape priced the first half while ignoring the second.

Context sharpens the reaction. TSMC's 2025 revenue grew 31.6% to NT$3.809 trillion. Gross margin expanded from 56.1% to 59.9%. Operating margin ran from 45.7% to 50.8%. The ten largest customers generated 78% of 2025 revenue, with the largest at 19% and the second at 17%. AI chips now account for 61% of the top line. That is not a diversified foundry. That is a levered bet on a handful of hyperscaler capex budgets. When the tape decides those budgets carry duration risk, a 77% earnings gain does not save you.

The $100 Billion Arizona Line and a Capex Number Nobody Would Underwrite

The capex figure is where the session turned. TSMC now expects capital expenditures of $60 billion to $64 billion in 2026, at least $4 billion above its previous forecast, and announced an additional $100 billion investment in Arizona inside that envelope. It also raised its revenue growth projection to slightly more than 40%, well above the 30%-plus increase it had carried into the print.

Read that against the setup. The question walking in was whether the 2026 capital budget got revised up from the $52–$56 billion range, because that number represents TSMC's own bet on where demand sits in 2028 and beyond. The answer landed at $60–$64 billion — a $4 billion to $12 billion overshoot depending on where you anchor. The prior language had management drifting toward the high end of $52–56 billion because supply was tight and demand kept climbing. Now the ceiling itself moved.

The framing was straightforward: a strong beat paired with raised full-year guidance sends the stock sharply higher, and any sign the AI capex cycle is cooling triggers a pullback across the sector. TSMC delivered the first scenario and collected the second outcome. That inversion is the single most important data point on the tape today.

The mechanism is duration. Lead times at TSMC's packaging facilities run 52 to 78 weeks with bookings extending into 2027, and CoWoS availability is the binding constraint on how fast AI accelerators deploy. A $64 billion budget funds capacity that monetizes in 2028. With the ten-year at 4.60% and crude bid, that cash flow discounts back harder with every basis point.

It did not happen in a vacuum. The results came after ASML raised its outlook — the second hike to its sales forecast this year. Two of the three most critical suppliers in the AI supply chain raised guidance inside seven sessions and both got faded. Earnings blowouts do not lift AI names the way they did six months ago. The market is no longer reading the spend as demand confirmation. It is reading it as cash out the door.

Micron and the Memory Complex Rolling Over Under Its Own Weight

Memory took the worst of it, and it took the worst of it yesterday too. Micron Technology (MU) traded at 857.10, down 47.18 or 5.22%, stacked on Wednesday's 8.02% collapse to 904.28 — a 78.84-point single-session loss. Shares had slid 4.6% before the bell to $862.70 after dropping 8% in the prior session. Two days, 13.2% of market value, and no earnings event to blame.

The selling ran straight through a piece of good news. Micron signed new long-term automotive supply deals covering 20% of DRAM volume — DRAM made up 76% of last quarter's revenue — and a third of NAND output, with a target of half of revenue coming from take-or-pay contracts by 2030 to steady the earnings line. The stock got sold anyway, because the bulk of DRAM sales still carry spot market risk. Same pattern as TSMC: structural de-risking announced, structural de-risking ignored.

The damage is global. SK hynix lost 11.5% and Samsung Electronics dropped 8.8%, both pressured by regulatory moves in Korea, with supply and AI demand uncertainty sitting on the sector. SK Hynix had already tumbled more than 15% in Seoul on Monday — its worst day after a strong Nasdaq debut — dragging the Kospi almost 9% to 8,806.93 and the Kosdaq 4.6% lower to 799.36. The chipmaker had ripped 13% in its Wall Street debut on Friday. Round trip in three sessions.

Sandisk (SNDK) got hit for 5.4% on the day after an 8.12% drop to 1,615.00, a 142.82-point loss. Corning (GLW) shed 6.7%. ServiceNow (NOW) fell 4.7%. Western Digital (WDC) dropped 7.2% before the bell and Coherent (COHR) lost 5%. The bid faded across every node of the AI hardware stack — memory, optics, glass, software.

Breadth inside the complex tells you this is not idiosyncratic. When the leader beats by 77% and the second-largest supplier raises twice and the whole chain still bleeds, the multiple is the problem, not the fundamentals.

The Dow Did the Work at 52,806.31 While the Nasdaq Bled

The Dow was the lone major index in the green and it earned the tape. At 52,806.31, up 147.67 or 0.28%, the index ripped to within striking distance of the week's highs while the Nasdaq handed back 214.85 points. That divergence is not noise. It is the rotation printing in real time.

Look at the composition. UnitedHealth (UNH) at 437.61, up 19.09 or 4.56%. Abbott (ABT) up 12%. Those two carried the price-weighted index while the AI complex got smoked. Healthcare, industrials, financials and energy have absorbed the money coming out of semis, and the shift has legs beyond a single session.

The rotation had already shown up in the small-cap tape. The Russell 2000 at 2,988.55, up 12.29 or 0.41%, outperformed the S&P 500 by 57 basis points on the day. Small caps do not lead when a risk-off wave is running. They lead when capital is redistributing inside equities. That is the signature.

Wednesday hinted at it. Money pared exposure to key semiconductor names and moved into select mega-cap tech. Amazon and Alphabet ripped 3%. Microsoft climbed 3%. Apple gained 4% to a fresh all-time high, closing at 327.50, up 12.64 or 4.01%. Micron dropped 8%. Lam Research fell 3%. Advanced Micro Devices fell 3%. The tape was already sorting AI infrastructure from AI monetization twenty-four hours before TSMC printed.

Under the surface the churn is heavier than the index levels imply. For a stretch, the crowd rode positive momentum generated by chips alongside the AI build-out names — computer hardware, power, construction. The fact that chip stocks are not moving up this week despite strong results from sector giants argues the rotation into financials, industrials and energy has staying power. With earnings expectations pitched this high in tech, the year's biggest winners face the steepest headwinds.

The VIX at 15.87, up 0.20 or 1.27%, confirms it. This is not a market bracing for a drawdown. This is a market re-weighting, and the Dow's 147.67 points is where the money landed.

UnitedHealth's $6.38 Blew a 30% Hole Through Consensus

UnitedHealth Group (UNH) delivered the cleanest beat on the tape and the stock ripped 4.56% to 437.61, adding 19.09 points. Second-quarter revenue hit $112.03 billion against $111.62 billion in the year-ago quarter. Earnings from operations landed at $8.0 billion. GAAP EPS printed $6.04 and adjusted EPS came in at $6.38 against a $4.84 consensus — a 29.9% beat. Net income reached $5.48 billion versus $3.41 billion, or $3.74 per share, a year earlier.

The guidance raise was the real fuel. UnitedHealth now expects full-year 2026 adjusted earnings of $19.50 to $20.00 per share, up from a prior outlook above $18.25. GAAP full-year guidance moved to $18.45 to $18.95. Revenue guidance held above $439 billion, though CFO Wayne DeVeydt signaled the company should beat that figure given the second-quarter result. For reference, 2025 revenue landed at $447.6 billion.

Optum did the heavy lifting. The unit generated second-quarter revenue of $65.7 billion and earnings from operations of $4.0 billion, delivering 160 basis points of year-over-year margin expansion while supporting more than 120 million consumers. That margin line is what separates this print from a cost-control story.

The membership data cuts the other way and management did not hide it. UnitedHealthcare served 48.5 million people in the quarter, down 525,000 from the prior quarter. The company forecast full-year losses of 500,000 ACA exchange members and 1.1 million Medicare Advantage members. Premium increases have been steep enough to hold revenue flat despite the attrition. DeVeydt was blunt that medical costs stayed elevated over historical levels and that the results reflect pushing down an already-high number rather than any bend in the underlying trend.

Set it against the Q1 print — adjusted EPS of $7.23 against a $6.57 estimate — and the pattern is two consecutive quarters of double-digit beats from a name that spent 2025 as a broken story. That is the rotation in one ticker: no AI exposure, no capex cycle, no 2028 payback, and $8.0 billion of operating earnings landing this quarter.

GE Aerospace Raised Everything and Got Sold Down 4%

GE Aerospace (GE) delivered a beat-and-raise across every line that matters and the stock got smoked for 4.00%, dropping 14.41 points to 345.94. Second-quarter revenue hit $13.35 billion, up 21%, against an $11.87 billion consensus. Net income from continuing operations printed $2.30 per share versus $1.87 a year ago. Adjusted earnings landed at $2.02 against $1.66 in the year-ago quarter and an estimate of $1.86 — an 8.6% beat on the bottom line and a $1.48 billion beat on the top.

The raise was aggressive. Full-year 2026 adjusted profit guidance moved to $7.65 to $7.85 per share from $7.10 to $7.40. Adjusted operating profit guidance jumped to $10.55 billion to $10.75 billion from $9.85 billion to $10.25 billion — a $700 million lift at the midpoint. The company now projects high-teens adjusted revenue growth. Against 2025, which delivered adjusted income of $6.37 per share and adjusted operating profit of $9.1 billion on adjusted revenue of $42.3 billion, the 2026 guide implies 20% earnings growth and 18% operating profit expansion.

None of it held. GE hit an all-time high of $382.97 earlier this month and now sits 37.03 points below it, a 9.7% drawdown from the peak. The stock carried a forward multiple of 48.01 times into the print. That is the whole answer. When you are paying 48x forward for an industrial, a $700 million guidance raise gets you nothing because the raise was already in the price.

The operating momentum is not in dispute. First-quarter orders exploded 87% to $23.0 billion, with Commercial Engines & Services orders nearly doubling and Defense & Propulsion Technologies orders climbing 67% — the strongest defense intake of the decade. Q1 revenue rose 24.74% to $12.39 billion, adjusted EPS ripped 25% to $1.86 against a $1.60 consensus, and free cash flow climbed 27.44% to $1.66 billion. Backlog cleared $210 billion.

GE has now beaten in five straight quarters with surprise margins between 9.79% and 17.32%. The tape sold the sixth. Same disease as TSMC: the fundamentals are not the constraint, the multiple is.

Abbott's $12.6 Billion and the Exact Sciences Distortion

Abbott (ABT) was the day's standout, ripping 12% on a print that cleared consensus and lifted the full-year floor. Second-quarter sales hit $12.6 billion, up 13.0% on a reported basis and 4.8% on a comparable basis, against an estimate of $12.48 billion. GAAP diluted EPS came in at $0.53 and adjusted diluted EPS printed $1.31 against a $1.28 consensus.

The guidance moved the right way. Abbott raised full-year 2026 adjusted diluted EPS guidance to $5.45 to $5.60 from $5.38 to $5.58 and reaffirmed comparable sales growth guidance of 6.5% to 7.5%. Third-quarter adjusted EPS guidance landed at $1.38 to $1.46. CEO Robert Ford framed the second half as accelerating on both the sales and earnings lines. The company returned $2.1 billion to shareholders in the quarter.

The segment data is where the gap between headline and substance opens. Diagnostics sales ripped 42.3% on a reported basis but only 2.9% on a comparable basis once the Exact Sciences acquisition — closed March 23, 2026 — is folded into prior-year figures. That 39.4-point spread is pure M&A arithmetic, not organic demand. Cancer diagnostics did deliver 13.3% comparable growth, with Cologuard posting mid-teens gains as the colorectal screening test pulled first-time and returning customers.

Medical Devices, the largest segment, grew 9.0% reported and 8.4% comparable, led by electrophysiology at 13.4% comparable and high-single-digit gains across rhythm management, diabetes care and heart failure. Established Pharmaceuticals grew 8.4% reported and 8.7% comparable on double-digit expansion across Latin America and Asia Pacific emerging markets. Nutrition was the sole drag, falling 3.1% reported and 3.6% comparable against reduced volumes and fourth-quarter 2025 pricing decisions, though the segment added $127 million sequentially off the first quarter.

Strip the acquisition and Abbott grew 4.8%. That is the number. The 13.0% reported line is what got the stock a 12% move, and DexCom (DXCM) rode the sector strength for a 7% gain of its own. Healthcare took the baton from semis and did not fumble it.

Retail Sales at 0.2% Was a Gas-Pump Illusion

June retail sales rose 0.2% month over month against a 0.3% consensus, following a 1.0% gain in May that was revised up from 0.9%. On the headline the print looks soft. Under it the consumer is doing better than the number implies, and the gap is gasoline.

Excluding gas stations, retail sales climbed 0.7%. That 50-basis-point spread is the entire miss. Retail sales are not adjusted for inflation, so the drop in pump prices during June mechanically dragged the headline while volumes held. The control group — the components that feed directly into GDP — rose 0.5%, the sixth consecutive monthly increase, against a 0.7% prior reading and a 0.5% forecast.

The composition is informative. Online sales climbed 1.9%, fueled by spending around the June 23–26 Prime Day window. Sales at sporting goods, hobby, musical instrument and bookstores rose 1.3% on World Cup-related spending. Clothing and accessories stores slipped 0.3%. The consumer is spending on events and screens and skipping the mall.

The bear case is pull-forward. May's 1.0% gain may have borrowed from June, and the government already revised first-quarter Personal Consumption Expenditures growth down to 0.5%. Consumer spending is 70% of the US economy. The last Atlanta Fed GDPNow estimate for second-quarter GDP growth sat at 1.3% with an update due today, and the initial second-quarter PCE estimate lands July 30.

The bond market read it as strength, not weakness. Net of the fuel distortion, June sales grew at a solid clip, and the six-month streak in the control group is the cleanest signal in the release. That is not a consumer rolling over. That is a consumer absorbing a 3.5% annual inflation rate and continuing to transact.

Consensus for retail sales ex-autos had been staked at -0.1% against a 0.8% prior. The Philadelphia Fed Business Outlook Survey was forecast to improve to 13.0 to 15.0 from 10.3, with the employment component sitting at 7.9. Pending home sales were forecast at -0.5% against a 3.8% prior. Nothing in that slate argues for a cut.

Claims at 208,000 Snapped the Rate-Cut Argument

Initial jobless claims landed at 208,000 for the week ended July 11, down 8,000 from the prior week's revised level and well below the 216,000 consensus. The previous week was revised up by 1,000, from 215,000 to 216,000. The four-week moving average fell 4,750 to 211,250. Continuing claims held at 1.814 million.

That 208,000 print is an over-two-month low. It is also the number that broke the September easing case.

Walk the sequence. Claims ran 230,000 in the June 6 week, the cycle high for the quarter. They eased to 225,000, then 212,000, then 210,000, then 215,000, then 216,000, and now 208,000. The four-week average at 211,250 sits below where it printed in April, May and June. Federal employee claims have stayed contained near 400 despite the administration's headcount reductions.

The labor backdrop has been described as low-firing, low-hiring for two quarters. The firing side of that equation just got tighter. Continuing claims at 1.814 million marked the highest reading since late March, which cuts the other way — people who lose jobs are taking longer to find new ones — but the flow into unemployment is not accelerating.

Pair it with the consumer data and the picture sharpens. Retail sales ex-gas at 0.7%. Control group at 0.5% for a sixth straight month. Claims at a two-month low. Bank earnings that reinforced corporate fundamentals holding up. That is not an economy that needs help.

The Fed has been explicit. The FOMC projected a rate hike this year, and rate derivatives are positioned consistent with that. Implied odds of a September increase sit at 44%, down from 50% a day earlier after the soft PPI print, but the direction of travel in the dot plot is up, not down. June CPI fell 0.4% month over month with the annual rate easing to 3.5% and core holding at 2.6%. PPI fell 0.3%, the first drop in nearly a year, driven by lower energy costs.

Energy is the swing factor, and energy is going the wrong way.

The 10-Year at 4.60% and a Fed Positioned to Hike

The yield on the 10-year Treasury ripped to 4.60% on Thursday, closing on the near two-month high of 4.62% set July 13. That is a 5-basis-point move off Wednesday's 4.55% close and the direct transmission channel from the Hormuz headlines into equity multiples.

The path matters. The 10-year finished July 10 at 4.56% with the 2-year at 4.21%. It edged down to 4.55% Wednesday after the producer price report gave fresh evidence of easing inflation. Then retail sales came in solid net of fuel, claims hit a two-month low, and crude stayed bid — and the whole move reversed inside a session. The 10-2 spread sat at 0.42 on July 15. The 52-week range runs 3.95% to 4.69%, putting the current print 81% of the way up the band.

Technically the yield has cleared every moving average that matters. The 20-day sits at 4.48%, the 50-day at 4.49%, the 200-day at 4.25%. Resistance stacks at 4.59% and 4.64%. Support runs 4.52% and 4.49%. A break of 4.62% opens 4.69% and the cycle high.

The Fed backdrop is the part the equity tape keeps underpricing. The wave of attacks between the US and Iran lifted benchmark fuel prices and rekindled the energy-inflation concern that drove sharp price increases from March onward. That coincided with fresh signals of a strong economy. The inflationary outlook and the tight labor backdrop pushed the FOMC to project a hike this year, and rate derivatives are positioned accordingly.

Elevated yields, a hawkish Fed pivot and a resilient economy keep the dollar supported near term. The dollar index closed at 100.49, down 0.27%.

Now run the math on the AI trade. TSMC just committed $60 billion to $64 billion of capital against cash flows landing in 2028. Every basis point on the 10-year cuts the present value of that stream. At 4.25% — the 200-day — the build-out pencils. At 4.60% with a hiking bias, it pencils worse. That is why a 77% earnings gain got faded.