US Equities

Micron reported fiscal fourth-quarter revenue of $54.23 billion against $41.46 billion in the prior quarter and $11.32 billion a year earlier, with non-GAAP earnings of $33.42 a diluted share against a consensus of $31.61 and a gross margin of 87.0 percent. Guidance for the first quarter of fiscal 2027 is revenue of $61.5 billion plus or minus $1.5 billion against a consensus of $57.02 billion. Core data centre revenue rose to $18.00 billion from $1.58 billion a year earlier, full-year revenue reached $133.19 billion against $37.38 billion, and noncurrent customer contract liabilities rose to $12.90 billion from $142 million as customer commitments under long-term supply agreements reached $32 billion against $22 billion in June.

Why it matters here: This is the hard evidence that the artificial-intelligence spending cycle converts into cash rather than promises, and it arrives in the most commoditised part of the supply chain, where pricing power has historically been weakest, which is what makes it a read on the whole theme rather than on one issuer. For this class it reaches the semiconductor exposure directly, the technology benchmark through the memory and infrastructure complex it carries, and the broad benchmark through research published the same day naming technology a major driver of index performance. The balance sheet is the strongest part of the case, because $12.90 billion of noncurrent customer contract liabilities against $142 million means customers are prepaying years ahead to secure supply.

Counterpoint: The shares actually fell in the session after the release despite the beat, because expectations had already run ahead and margin guidance of 86.25 percent came in below the street on a worker pay increase; market commentary noted the stock sank even as sales nearly quadrupled because much of the news was already priced in. A memory cycle with an 87.0 percent gross margin also invites the capacity additions that have historically ended such cycles abruptly.

Accenture reported fiscal fourth-quarter revenue of $18.68 billion, above its own guidance range of $17.8 billion to $18.4 billion and an analyst consensus of $18.3 billion, with earnings of $3.29 a share against a $3.18 consensus. The chief executive said the company capped off another year of broad-based growth, grew adjusted earnings per share 8 percent, returned a record $11.5 billion to shareholders and reached a new high of 141 quarterly client bookings of $100 million or more, alongside a dividend increase and an upbeat fiscal 2027 outlook. Shares closed up 15.78 percent.

Why it matters here: Large-deal bookings are the best available forward indicator of enterprise technology spending because they are contracted rather than forecast, so a new record says corporate clients are still committing multi-year budgets despite 5 percent yields and triple-digit oil. For this class it supports the technology benchmark on demand and the broad benchmark on cash returns, but the exposure worth naming is the equal-weight index, because a near-16 percent single-day gain in a large but non-megacap constituent is exactly the kind of move that cohort has been starved of.

Counterpoint: A 15.78 percent single-day move in a mature services company implies expectations were extremely low rather than results being extraordinary. A reading that enterprise budgets are healthy also sits awkwardly beside manufacturing respondents in the same week describing capital expenditure pushed out indefinitely and capital commitment lead times extending to 176 days from 171, and one company's bookings cannot carry a macroeconomic conclusion.

Federal Reserve Vice Chairman Philip Jefferson said on 1 October that the economy is being buffeted by a cascade of shocks including rising energy prices, the surging artificial-intelligence build-out and changes to trade policy, that any future policy adjustments should be determined by carefully examining trends in the data, and that reaching a judgement may take more time. The remarks echoed New York Fed President John Williams, who said earlier in the week there was no need for urgency after September's quarter-point increase. Citi's head of U.S. equity strategy said the central bank is probably in pause mode until the middle of next year and its next move could be a cut, and by 2 October traders priced a 72 percent chance of no change in October.

Why it matters here: Equity valuations in a tightening cycle hinge on when the cycle stops, and three officials in one week declining to press for the next move is the market's best available answer to that question. For this class the benefit is ordered by balance-sheet sensitivity rather than by duration: the small-cap and equal-weight cohorts gain most because their earnings are directly levered to borrowing costs rather than to the discount rate on distant cash flows, which is why they are named ahead of the broad benchmark.

Counterpoint: Jefferson explicitly named rising energy prices and trade policy changes as shocks buffeting the economy, which is hardly a supportive growth message, and the index gained very little on the day. Minneapolis Fed President Kashkari's argument that the data-centre build-out is pushing up the neutral rate, if correct, is a permanent increase in the cost of equity that far outweighs a one-meeting pause.

The Federal Reserve's preferred inflation gauge showed core prices rising 0.2 percent in August and 3.0 percent year on year, well below a 3.3 percent expectation, with headline prices up 0.3 percent monthly and 3.4 percent annually against a 3.7 percent expectation. Market-implied odds of an October increase fell to around 35 percent from roughly 51.5 percent beforehand and about 70 percent earlier in the week, and by 2 October traders priced a 72 percent chance of no change. The 2-year Treasury yield fell about 10 basis points during the 1 October session as the increase was priced out.

Why it matters here: Equities in this regime trade as a direct inverse of yields and the session proved it, with the market lower at midday while yields sat at 24-year highs and higher by the close once the front end rallied. Small-cap outperformance is the tell for this class, because that cohort carries the most floating-rate debt and therefore the most sensitivity to the policy path specifically rather than to the term premium, while the technology benchmark gains through long-duration growth valuations and financials through reduced mark-to-market pressure on securities portfolios.

Counterpoint: The gains were trivial across the major averages and came after a September in which the equal-weight index trailed the cap-weighted index by 4.6 percentage points, one of only three such months in data back to 2003. A single month's inflation print cannot offset a manufacturing prices index at 77.9, crude above $100 and a 30-year yield at 5.683 percent, and published accounts of the release itself disagree on the levels.

Private nonresidential construction spending rose 1.0 percent in August to a seasonally adjusted annual rate of $773.0 billion, a fifth consecutive monthly increase, with data centre construction up 7.5 percent in the month and 73.2 percent year on year and now above 10 percent of total private nonresidential construction spending. Total construction spending was estimated at $2,203.1 billion, 0.9 percent above the revised July estimate of $2,184.5 billion but 1.7 percent below August 2025.

Why it matters here: This converts the artificial-intelligence narrative into poured concrete, which is why it is carried on the growth factor for this class rather than on business fundamentals. Five consecutive monthly increases in nonresidential building, led by a category growing 73.2 percent annually, is a real order book for the industrial exposure that supplies electrical equipment, cooling and construction services, and forward confirmation of the compute demand that the semiconductor exposure sells into. It also corroborates the policymaker who argued this build-out is large enough to move the equilibrium interest rate.

Counterpoint: Private nonresidential spending remains 1.0 percent below the prior year in aggregate and total construction spending is 1.7 percent lower than a year ago, so the data centre boom is substituting for weakness elsewhere rather than adding to the total. Concentrating a growing share of construction demand in one tenant category is a fragility rather than a strength, and the data is two months stale by publication.

Alphabet unveiled Gemini 4 Argon on 30 September, describing it as its most advanced artificial-intelligence model yet with major improvements in coding, cybersecurity and complex professional work, and saying it sets a new record in real-world software engineering, ties for first in cybersecurity and leads another benchmark measuring performance across finance, legal and other professional tasks. Shares rose more than 1 percent after hours and were up about 1.5 to 1.9 percent in premarket trading on 1 October, before closing the session 1.7 percent lower. The release came nearly a year after its predecessor.

Why it matters here: Model competition is the engine of the capital expenditure cycle for this class: each frontier release forces rivals to spend more on compute, and that spending is the demand semiconductor and memory earnings capitalise, which is why the semiconductor exposure appears alongside the technology benchmark here rather than only the issuer's own index. The claimed gains in coding and professional work matter specifically because those are the first applications with demonstrable enterprise willingness to pay, which is what turns the build-out into revenue rather than depreciation.

Counterpoint: The stock finished the session 1.7 percent lower despite the release, so the market did not treat it as value-creating, and the benchmark claims are the company's own with no independent verification. A competitive arms race that forces ever-higher spending is as easily read as a margin threat as an opportunity, and one analyst warned the same day about increasing artificial-intelligence turbulence as a factor investors must weigh.

The Institute for Supply Management's Manufacturing PMI registered 54.5 percent in September, 0.1 point below August and slightly below a consensus of 54.9, marking a ninth consecutive month of expansion and a 23rd month of overall economic growth. New orders rose 1.6 points to 55.3, backlogs jumped 4.6 points to 56.4 and employment rose 1.5 points to 52.7, while production fell 1.6 points to 56.7 and inventories dropped 2 points into contraction at 48.6. The reading corresponds to a 2.4 percent annualised increase in real gross domestic product.

Why it matters here: This release contains two mechanisms pointing opposite ways, and separating them matters for this class: the demand signal is genuinely good and the cost signal is genuinely bad, so they are allocated as distinct projections rather than netted. On the demand side, orders and backlogs both accelerated while customers' inventories stayed in too-low territory, which the institute treats as positive for future production, and that combination supports domestic cyclical earnings, which is why the industrial and small-cap exposures lead here with the semiconductor exposure participating through computer and electronic products expanding on every measure.

Counterpoint: Sentiment is deteriorating beneath the index. Negative comments outnumbered positive ones, the headline missed consensus at 54.5 against 54.9, inventories dropped 2 points into contraction at 48.6, and capital expenditure commitment lead times extended to 176 days from 171, which reads as hesitation rather than confidence about forward demand.

Initial claims for U.S. unemployment benefits totalled a seasonally adjusted 197,000 in the week ending 26 September, down 1,000 from the prior week and below a consensus of 200,000, the lowest level since July, while continuing claims dropped to a three-year low. Placement firm Challenger, Gray and Christmas reported that companies announced 43,281 job cuts in September, a 39 percent decrease from August and 20 percent below September 2025, with planned layoffs year to date at their lowest since 2022.

Why it matters here: Employment income is the foundation of the consumption that most of this class's earnings rest on, and both the claims series and the announcement series say employers are holding onto workers. For this class that matters most to the consumer discretionary exposure, whose sales depend directly on household cash flow, and to the domestic-cyclical small-cap and equal-weight cohorts that have no overseas offset, with analysts explicitly identifying data resilience as a major factor supporting U.S. risk assets.

Counterpoint: The same resilience is precisely what gave the central bank room to start raising rates, so good labour data is a rates risk as much as an earnings support, and a strong September employment report would restore the October increase that was just priced out. Reporting the same week also noted it has rarely been harder for people to find new jobs, which means low layoff announcements may mask weak hiring rather than signal strength.

Tariffs were cited in 34 percent of negative comments in the September manufacturing survey, second only to pricing volatility at 46 percent and ahead of the Iran war at 30 percent, with respondents naming Canada specifically and repeatedly. A transportation equipment respondent said the trade war with Canada is getting worse every day, causing prices to rise and creating massive disruption, with buying pushed out indefinitely as customers avoid capital expenditure until costs and demand are clearer. An electrical equipment respondent said new tariffs against Canada have drastically increased costs for capital expenses and assemblies. Average capital expenditure commitment lead times extended to 176 days from 171 and new export orders fell to 50.9 from 53.2.

Why it matters here: The mechanism for this class is not the tariff itself but the uncertainty it creates, because respondents describe customers refusing to commit capital until costs and demand are clearer, which is a deferral of revenue that no price increase recovers. The 176-day capital commitment lead time and the fall in export orders are the quantitative fingerprints of that behaviour. Industrials bear it first through order books, the equal-weight cohort bears it hardest because restructuring a cross-border supply chain takes scale, and consumer discretionary absorbs the end of the chain in retail prices.

Counterpoint: The headline manufacturing index still printed 54.5 with new orders accelerating to 55.3 and backlogs jumping 4.6 points to 56.4, which is not the profile of an economy frozen by trade uncertainty, and industrials actually led the market on the day of this release. Longer capital expenditure lead times can equally reflect confident long-horizon ordering rather than hesitation, and officials on both sides continue negotiating.

The September manufacturing prices index rose 6.8 points to 77.9, close to the 78.3 reading at the start of the Iran war, with 58.6 percent of respondents reporting higher prices, up 12.4 points from August. Twenty-four commodities were reported up in price and none down, including aluminium for a 34th consecutive month, steel for an 11th and copper for a 15th. Every one of the six largest manufacturing industries reported price increases, and the issuing institute attributed the surge to steel and aluminium prices through the value chain, tariffs on imported goods, and petroleum products from the Middle East conflict.

Why it matters here: A list of 24 commodities rising with none falling means there is nowhere in the input basket to hide, which is a different statement from an average price increase. For this class the compression lands hardest on the equal-weight benchmark, because the median industrial or consumer-facing company cannot offset a prices index at 77.9 the way a megacap can, and respondent commentary makes it concrete at the consumer end, with food and beverage respondents citing fuel costs affecting transportation and the overall cost of goods.

Counterpoint: Gross margins across the market have held up well so far, and the same survey shows backlogs and new orders accelerating, which gives firms the pricing power to pass costs on. Several respondents also described lead times normalising and supply chain performance improving versus prior years, and the August core inflation reading came in at 3.0 percent against a 3.3 percent expectation, which is the actual outcome a prices survey is supposed to predict.

Freddie Mac's survey published 1 October showed the 30-year fixed-rate mortgage averaging 7.28 percent, up from 7.03 percent a week earlier and 6.34 percent a year earlier, with the 15-year at 6.60 percent against 6.42 percent and 5.55 percent. A daily tracker showed the average 30-year rate reaching 7.6 percent on 30 September, the highest since 1 November 2023, before easing to 7.54 percent, as the 10-year Treasury yield hit 5.306 percent. In the same week's manufacturing survey a machinery respondent said higher interest rates slow the growth of new construction projects.

Why it matters here: Housing is the largest single channel through which interest rates reach the real economy, and for this class the effect fans out in three directions at once: households carry less discretionary cash flow, which hits the consumer exposure; banks originate and refinance less, which hits fee income in the financial exposure; and construction-exposed industrials see orders deferred, which the survey commentary makes concrete rather than theoretical. A rate 94 basis points above a year earlier is a large move in all three.

Counterpoint: Nonresidential construction spending rose for a fifth straight month and residential spending rose 1.1 percent in August, so the order-book damage is not yet visible in the hard data. Rising rates also expand bank net interest margins, and the labour market data released the same day showed layoff announcements at their lowest since 2022, which protects the household cash flow this force assumes is being squeezed.

U.S. diesel prices hit record highs last month and remained at an average $6.40 a gallon on 1 October, after Ukrainian strikes forced Russia to ban diesel exports and attacks on Middle East refineries constrained regional fuel shipments. In the September manufacturing survey a food and beverage respondent said fuel costs are still affecting transportation costs and the overall cost of goods, with fuel and freight among the commodities reported up in price. A U.S. diesel export ban remains under consideration, with a research house estimating it would cut U.S. prices about 4 percent.

Why it matters here: Diesel is the price of moving physical goods, so a record level is a margin tax reaching almost every non-financial company in this class rather than a sector-specific cost. The industrial and transport exposure pays it directly, the consumer exposure absorbs it in the delivered cost of goods and in household budgets, and the equal-weight benchmark carries it most because the median company has no pricing-power offset, which is why that exposure is named rather than the cap-weighted index.

Counterpoint: Gross margins across the market have held up, and a U.S. diesel export ban under consideration would cut prices by an estimated 4 percent, relieving exactly this pressure. A bank also argued the same week that the overall demand environment appears stable and that economic resilience should alleviate these risks for the transport sector, reiterating a buy rating on a major carrier.

The S&P 500 equal-weight index trailed the cap-weighted index by 4.6 percentage points in September, which research firm Bespoke Investment Group said was among only three months with wider underperformance in data going back to 2003. The Russell 2000 fell more than 5 percent in September, bringing its third-quarter decline to about 7.5 percent, while the broad benchmark slipped 0.5 percent in September but added 2 percent for the quarter. The firm wrote that for the broadening trade to start working again investors had better start rooting for lower oil prices.

Why it matters here: This is the most important thing the index level is not telling you about this class. The benchmark finished September down only 0.5 percent while the median constituent was hammered, which means the apparent calm is a weighting artefact, and the research firm's framing closes the loop with the rest of the record: the rotation into megacaps began with the war and unwinds when oil falls rather than when the benchmark rises. The equal-weight, small-cap and blue-chip exposures are the losing cohorts and the megacap technology benchmark is the destination, which is why it carries the opposite direction inside the same force.

Counterpoint: On the first day of October the pattern reversed, with small caps outperforming and the Russell 2000 closing at 2,806.625, up 0.35 percent, against a smaller gain for the large-cap index, and the small-cap benchmark has still beaten the large-cap one over 2026 as a whole with a gain of more than 12 percent. Extreme breadth divergences are mean-reverting signals as often as they are warnings, and narrow leadership is also fragile leadership.

Nike reported fiscal 2027 first-quarter revenue of $11.21 billion, down 4 percent reported and 5 percent currency-neutral and short of about $11.32 billion expected, with diluted earnings of $0.48 a share above a $0.43 consensus and net income down 2 percent to $712 million. Gross margin expanded 60 basis points to 42.8 percent and selling and administrative expense fell 3 percent to $3.9 billion, while digital revenue fell 13 percent. The company guided fiscal 2027 revenue to decline high-single digits against a prior forecast of a low-to-mid-single-digit decline in the first half, and announced a transformation programme targeting about $2.5 billion of cumulative savings through fiscal 2031 against roughly $1.0 billion of pre-tax charges.

Why it matters here: This is a demand signal rather than an execution excuse for this class, because the margin and cost lines both improved while revenue missed: gross margin expanded 60 basis points and overhead fell, yet the company still guided the year down. Digital revenue falling 13 percent is where consumer intent shows up first. It reaches the consumer discretionary exposure as the clearest available discretionary demand reading, the equal-weight benchmark because consumer brands carry a far larger effective weight there than in the cap-weighted index, and the price-weighted blue-chip index because it holds the company directly.

Counterpoint: Earnings of $0.48 beat the $0.43 consensus, gross margin expanded, North America actually grew 2 percent, and the weakness is concentrated in Greater China and a struggling subsidiary brand rather than being broad-based. This is substantially a company-specific brand repositioning problem, which the transformation programme is designed to address, and reading it as a macroeconomic consumer signal overstates it.

Brent settled 4.4 percent higher at $102.31 a barrel and West Texas Intermediate 2.7 percent higher at $92.87 on 1 October after the carrier deployment report, while U.S. diesel averaged a still-elevated $6.40 a gallon. The session wrap noted that further gains in oil kept the equity market's enthusiasm in check even as Treasury yields retreated from their highs, and the September manufacturing survey listed fuel and freight among the commodities rising in price.

Why it matters here: The index level hides the mechanism here. Equities rallied because yields fell, then gave most of it back because oil rose, so energy costs compress the median company's margin while rewarding one sector, and the effect shows up as narrow leadership rather than as an index decline. That is why the equal-weight exposure is the honest read on an oil shock for this class, with the cyclical and consumer sectors named ahead of the headline benchmark, and why the energy sector carries the opposite direction inside the same force as the offsetting leg.

Counterpoint: The index still finished higher, small caps outperformed, and the energy weighting inside the large-cap benchmark means part of the oil move is simply recycled into index earnings. If the deployment proves to be deterrence and crude retraces, as it began to do on 2 October with front-month crude down almost 3 percent, this headwind reverses into a tailwind for exactly the breadth trade it is said to be blocking.

Long-dated U.S. Treasury yields reached their highest levels in 24 years on 1 October, with the 10-year touching 5.342 percent and the 30-year reaching 5.683 percent, the 10-year's highest since April 2002, before reversing by late morning with the 10-year and 30-year ending lower by around 5 and 3 basis points and the 2-year tumbling about 10 basis points. The move was synchronised internationally, with Germany's 10-year climbing 5.5 basis points to 3.633 percent and the French spread over Germany widening as far as 132 basis points, and it is occurring against U.S. government borrowing that has crossed $40 trillion.

Why it matters here: Equities in this regime are a straightforward inverse function of yields, and the session demonstrated it: stocks fell while yields rose in the morning and recovered when yields turned. For this class the mechanism works through three channels at once, the discount rate on future cash flows, the competing income from risk-free bonds that bears directly on blue-chip dividend payers, and the actual cost of the debt funding the artificial-intelligence capital cycle. The equal-weight and small-cap exposures bite hardest because they lack balance-sheet shelter, and financials face mark-to-market losses on securities portfolios even where margins widen.

Counterpoint: All three major averages still closed higher on 1 October and small caps outperformed, which is the opposite of what a pure rate shock should produce. A strategy head at Citi argued the central bank may be done raising rates and could cut next, traders priced a 72 percent chance of no change in October, and the long-end move reversed within the session, so the market is plainly not treating 5 percent yields as a solvency event.