Our thesis in one sentence. The Fed delivered the expected hike, but the more important investment signal was a higher real cost of capital and a market recovery that did not lift all equity groups equally.

Four findings carry it.

First, the policy debate moved beyond September. The Fed raised its target range by 25 basis points to 3.75%–4.00%, unanimously. The September projections put the median year-end policy rate at 4.1% in both 2026 and 2027. That is a materially different backdrop from assuming this hike will be quickly reversed.

Second, the bond arithmetic strengthened the prior edition’s real-yield thesis. At five years, the nominal yield rose 8 basis points while the real yield rose 17, leaving inflation compensation 9 basis points lower. The ten-year nominal yield finished at 5.01%. Higher real yields can squeeze valuations even when inflation compensation is contained.

Third, the consumer was still spending. August retail sales rose 1.2%, including a 1.2% increase excluding autos and gasoline. Initial claims fell to 196,000. But manufacturing output declined 0.3%. The economy was resilient without being uniformly strong. These releases describe activity before the September hike could materially affect behaviour.

Fourth, the recovery was concentrated in time. The S&P 500 lost 105.17 points over Monday to Wednesday, then recovered 98.69 on Thursday and Friday. That repaired almost 94% of the decline, but it did not establish a broad improvement in financing-sensitive businesses.

The strongest argument against our cautious reading is that earnings growth and productivity can offset higher discount rates. A rising real yield can partly reflect confidence in future growth, rather than only tighter policy. If that confidence spreads into smaller companies and new orders, the market may be absorbing the hike more successfully than the weekly index split suggests.

We would become more constructive if real yields eased, oil supply improved and small caps began outperforming alongside steady employment. We would become more defensive if real yields and inflation compensation rose together while activity weakened. Neither outcome is established by one volatile week.

The Hike Was Only the Beginning

The decision removed uncertainty about the first step. The projections raised a different question: how long must policy stay restrictive before inflation returns to target? That horizon matters to asset values because a temporary increase in borrowing costs is easier to absorb than a persistent one.

A higher path does not guarantee a long hiking cycle

At the new target midpoint of 3.875%, a year-end median of 4.1% is consistent with roughly one further quarter-point increase. The distribution is informative: 12 of 18 participants projected a 4.125% midpoint for 2026, four projected 4.375% and two projected 3.875%. The median is therefore a useful centre of gravity, but it hides disagreement over the next step.

Our interpretation is that markets should test the persistence of policy, rather than trade only the next meeting. The combination of resilient demand and unresolved energy costs makes an immediate reversal harder to justify. Equally, a sustained improvement in supply or a deterioration in employment could change the path. Forecasts are conditional on an economy that will continue to evolve.

What this means for valuation

A business whose cash flows arrive far in the future is sensitive to the discount rate applied over many years. A higher policy path can also raise refinancing costs before revenues respond. The useful distinction is between companies that can fund investment internally and those whose plans depend on repeatedly accessing cheap capital. The same macro rate can have very different consequences for each.

Real Yields Did More Than Nominal Yields

The rise in the headline Treasury yield understates the change in the real return investors demanded. Inflation compensation moved in the opposite direction, partly cushioning the nominal increase. This is consistent with a stronger policy response restraining inflation risk, although the yield decomposition alone cannot prove that explanation.

Why the curve matters more than the round number

The two-to-ten-year slope narrowed from 33 to 25 basis points; the two-to-thirty-year slope narrowed from 72 to 58. These are calculated from the official Treasury par curves. The front end repriced more than the long end. That pattern fits concern about tighter near-term policy better than a simple claim that every maturity was selling off equally.

Five percent is a useful reference point, but it is not a mechanical trigger for an equity crash. Stocks respond to the combination of earnings, discount rates and risk appetite. A gradual yield rise alongside stronger profits can be manageable; a fast rise accompanied by falling estimates is a different proposition. Credit spreads and refinancing terms help distinguish the two.

The limits of the inflation signal

Nominal minus real par yields is an approximation to inflation compensation, not a pure forecast of future inflation. Risk premia, liquidity and differences between the bond curves also matter. Falling compensation is therefore evidence against an uncontrolled inflation repricing this week, but it is not proof that households or firms have stopped worrying about prices.

What the Market May Be Missing

Oil can stop rising and still remain a problem

WSJ’s settlement summary put WTI at $100.30 and Brent at $103.87, with little net weekly movement. The retreat from midweek pressure followed reports that partial pipeline flows could resume. These are contract-specific settlements; they should not be substituted with earlier intraday quotes.

The investment inference is about duration. A flat weekly oil return does not mean the cost shock has disappeared. Businesses may still face higher freight, fuel and working-capital costs than their budgets assumed. What matters for margins is the cumulative cost burden and the ability to pass it on, not just the latest five-day price change.

The Bank of England supplied a useful counterpoint

The BoE held Bank Rate at 3.75% by a 6–3 vote, with the minority preferring a hike. It reported little evidence so far of material second-round wage and price effects, while judging inflation risks to have increased. The contrast with the Fed illustrates how the same energy shock can meet different domestic conditions. [11]

For a UK-based investor, the implication is to separate global shocks from local policy responses. Domestic activity, wage growth and currency exposure can matter as much as the global commodity headline. Sterling returns on overseas assets also depend on exchange-rate movements, so the dollar-based equity scoreboard is not a UK investor’s realised portfolio return.

A credible central bank does not guarantee rising asset prices

Policy can reduce inflation risk while increasing the real return required on investments. Both outcomes can occur together. It is therefore possible for monetary credibility to improve while equity multiples contract. Investors should ask whether prospective cash-flow growth adequately compensates for that higher hurdle.

What We Are Watching

The real cost of capital

Five-year real yield: 2.55%. A sustained move above 2.65% would reinforce valuation and refinancing pressure. A move below 2.40% with stable activity would offer more convincing relief. The point is the real financing environment, not a single intraday print. [5]

The inflation component

Five-year compensation: 2.31%. A rise above roughly 2.45% while real yields also climb would be more concerning than either move alone. It would suggest investors require both more real return and more compensation for inflation risk. The measure includes risk and liquidity effects, so it needs corroboration. [4–5]

The long bond benchmark

Ten-year par yield: 5.01%. A persistent move above 5.10% alongside weaker earnings expectations would increase pressure on equity valuations. A fall below 4.85% would need interpretation: supply relief is helpful, while recession-driven demand for bonds may be much less so. [4]

The breadth of the equity response

The Russell 2000 versus the S&P 500. A durable period of small-cap outperformance, supported by improving earnings and stable credit, would challenge the view that tighter financing remains the dominant constraint. Relative gains caused only by large-cap losses would be a weaker signal.

Energy supply rather than headlines alone

Oil below $95 would be a meaningful change only if it reflected sustainable supply relief rather than collapsing demand. Conversely, prolonged triple-digit prices can sustain cost pressure even without another weekly surge. Track physical flows, refined products and company commentary together.

Employment and investment

Claims below 200,000 and stronger orders would support resilience; a sustained four-week claims average above 225,000 accompanied by weaker investment would make the growth cushion less convincing. These are our monitoring bands, not official recession criteria. Initial claims must also be read alongside hiring and continuing claims.

The research question for the next edition

Can better cash-flow expectations broaden the market while the real cost of capital remains high? That is the link between the macro data and investment outcomes. The Fed has acted; the next evidence must come from households, businesses and the prices investors are prepared to pay for their future earnings.

The prior edition ended with payrolls as the event that reopened the September-hike question. This week answered it with inflation. The two-year yield moved from 4.37% to 4.63%, the ten-year approached 5%, and the implied hike probability moved above 80%. Oil also crossed from a high-$80s/low-$90s problem into a triple-digit macro variable. At the same time, the inflation-compensation move was small relative to the rise in real yields, which means the market treated the Fed response as part of the solution rather than simply pricing an uncontrolled inflation shock.

Figure 1. Oil rose while every major U.S. equity index fell

Key takeaway: The week was not a generic risk-off move. It was a specific energy-and-rates shock: crude gained close to 9% while small caps fell more than twice as much as the S&P 500.

Deep Dive I: CPI Made the Hike the Base Case

Friday’s CPI report mattered because the headline and the core told different stories. The headline increase of 0.4% was exactly what economists expected and was heavily influenced by gasoline. Gasoline rose 3.9% in August, energy rose 2.1%, and the BLS said gasoline accounted for more than one-third of the monthly all-items increase. That is the part a central bank can plausibly describe as an external relative-price shock.

The core was harder to dismiss. CPI excluding food and energy rose 0.3% in August after 0.2% in July, above the 0.2% Reuters consensus. Shelter rose 0.3%, airline fares 2.7% and communication 2.3%. Annual core inflation slowed to 2.4% from 2.5%, but the monthly acceleration arrived one week after a 162,000 payroll print and one day after a firm PPI report.

Figure 2. August inflation was energy-heavy, but core also surprised higher

Key takeaway: Gasoline did much of the headline work, but a 0.3% core monthly print against 0.2% expected was enough to turn the September decision from a coin flip into an 85%-plus market base case.

Why a stock rally can coexist with higher hike odds

The S&P 500 rose 0.86% on Friday even as short-rate markets moved further toward a hike. That is not automatically contradictory. A near-term hike can raise the front-end discount rate while lowering the perceived probability of a larger inflation problem later. Friday’s Treasury curve moved in that direction: short and intermediate yields rose, but the 30-year finished below Thursday’s 5.37% level at 5.35%. The equity rally also benefited from oil falling nearly 3% from Thursday’s highs.

What would make CPI less hawkish than it looks

If the August core rise proves concentrated in volatile categories such as lodging and airfares, and September energy prices reverse, the Fed’s preferred PCE measure could still look more benign. Reuters-cited economists put August core PCE around 0.2% to 0.3%. That makes the September decision likely, but the path after September remains more open than Friday’s headline odds imply.

Deep Dive II: The Curve Was Real, Even With Oil Above $100

The most useful arithmetic of the week is the Treasury decomposition. Five-year nominal yields rose 24 basis points from 4.54% to 4.78%. Five-year real yields rose 21 from 2.17% to 2.38%. Inflation compensation - nominal minus real at the same maturity - therefore rose only 3 basis points, from 2.37% to 2.40%. At ten years the same calculation gives +18 nominal, +17 real and only +1 compensation. At thirty years compensation was unchanged.

Figure 3. Most of the Treasury selloff was a real-yield move

Key takeaway: Even in an oil shock, the medium- and long-end weekly repricing was overwhelmingly real. The market raised the price of real money far more than its medium-term inflation premium.

This is why the week is more than an oil story. If investors had simply concluded that $100 oil would feed continuously into inflation, inflation compensation should have widened much more. Instead, the market also priced a more forceful policy response and a higher real discount rate. That combination is particularly difficult for smaller companies, leveraged balance sheets and long-duration growth assets whose valuations depend on distant cash flows.

The 10-year near 5% is the next threshold

The ten-year Treasury finished Friday at 4.96%, its highest level since 2023 in contemporary market reports. Five percent is not economically magical, but it is behaviourally important: it is a visible alternative return against equities and a reference point for mortgage, corporate and private-market discount rates. A clean break above 5% with credit spreads stable would be a rates problem; a break above 5% with widening high-yield spreads would be a broader financial-conditions problem.

Deep Dive III: Oil Moved from Commodity to Policy Variable

Brent ended the week at $104.61 and WTI at $100.05 after supply disruptions and escalating Middle East hostilities. Reuters calculated weekly gains of more than 8% for both benchmarks. U.S. diesel prices also reached record territory. The point is no longer simply that energy equities receive an earnings tailwind. Oil now sits inside the Fed reaction function through three channels: headline inflation, household expectations and the risk of pass-through into transportation-intensive services and goods.

August CPI already shows the first channel. Gasoline rose 3.9% month on month and 27.4% from a year earlier; the energy index was up 16.3% year on year. Michigan shows the second: one-year inflation expectations jumped to 4.6% from 4.0% as the survey explicitly cited fuel prices and trade tensions. The third channel - broad pass-through - is the one that remains unproven.

Figure 4. Consumers felt worse and expected more inflation

Key takeaway: The survey deterioration matters because central banks can look through a temporary energy shock more easily than a shock that changes wage and price-setting psychology.

What oil would need to do to change the Fed path

A sustained move above roughly $105-$110 matters more than a one-day spike because it extends the period during which gasoline, diesel and freight costs remain visible to households and firms. A retracement below $95 before the October meeting would relieve some of the pressure. The key weekly observation is that Friday’s 2%-3% oil pullback was enough to help equities rally even though CPI itself was not dovish.

Deep Dive IV: Earnings Still Mattered - Oracle and Adobe

Macro dominated, but the week also delivered a useful AI monetisation check. Oracle reported fiscal first-quarter revenue of $19.34 billion, up 30% year on year, with cloud revenue of $11.61 billion and cloud infrastructure growth of 62%. Remaining performance obligations reached $664 billion after more than $30 billion of new AI cloud contracts, while the company guided to at least $90 billion of full-year revenue. The trade-off is capital intensity: Oracle expects roughly $90 billion to $95 billion of annual capital expenditure.

Adobe also beat third-quarter revenue expectations at $6.76 billion, up about 13% year on year, and said AI-first annual recurring revenue more than doubled. But the shares fell after hours because fourth-quarter revenue guidance of $6.80-$6.85 billion put the midpoint slightly below consensus. The same pattern seen in prior AI reports remains: evidence of demand is necessary, but valuation and forward expectations decide the stock reaction.

Figure 5. AI demand showed up in cloud infrastructure and creative software

Key takeaway: Oracle showed extraordinary infrastructure growth; Adobe showed that AI monetisation is spreading into software. Neither removes the valuation discipline imposed by higher real yields.

Why this matters in a 5% 10-year world

The cost of capital is now colliding directly with the AI capex cycle. Oracle can report 62% infrastructure growth and still face investor questions about financing $90-plus billion of annual capex. Adobe can beat quarterly revenue and still fall if next-quarter guidance misses by a few tens of millions. Higher real yields make the market less willing to capitalize distant cash flows at premium multiples, even when the demand thesis is intact.

What the Market May Be Missing

The inflation shock and the bond shock were not the same thing

Observation. Oil rose almost 9%, headline CPI accelerated and Michigan inflation expectations rose, yet five- and ten-year inflation compensation barely changed on the week.

Why it is non-obvious. The natural narrative is “oil up, inflation up, yields up.” The decomposition says most of the yield move was real. That is consistent with investors expecting the Fed to respond rather than lose control of inflation.

Investment relevance. A real-yield shock is more directly hostile to high-multiple equities than a pure breakeven shock, while inflation-linked bonds offer less protection than they would in a compensation-led selloff.

Friday was a credibility rally, not a dovish rally

Observation. Stocks rose almost 1% even as hike odds moved above 85%.

Why it is non-obvious. Higher rate expectations are usually described as bad for stocks. But a decisive hike can reduce uncertainty and cap the long-end inflation premium. Friday also coincided with a material oil pullback.

Investment relevance. The market may tolerate one hike more easily than an open-ended rise in oil and long yields. The distinction between a one-off adjustment and a hiking cycle will dominate the next meeting.

Portfolio lens

On rates, the week says to distinguish real duration from inflation duration. The five-year nominal move was almost entirely real. Portfolios sensitive to valuation multiples can be exposed even if breakevens are stable.

On energy, $100 oil now matters beyond the energy sector. It changes transport costs, household sentiment, inflation expectations and the policy distribution. A portfolio that treats energy solely as a sector allocation misses the macro transmission.

On equity size, the Russell 2000 fell 2.4% while the S&P 500 lost 0.8%. Higher real yields and tighter policy expectations land harder on smaller companies with shorter financing runways and more variable-rate exposure.

On AI, Oracle and Adobe show that earnings remain strong enough to support the fundamental thesis. The harder question is whether those cash flows are valuable enough at a 10-year Treasury near 5%.

The Week Ahead: 14 to 18 September 2026

The coming week contains the event that this report has been repricing toward: the FOMC decision on Wednesday 16 September. The meeting also includes a Summary of Economic Projections, so the path matters at least as much as the first 25 basis points.

What We Are Watching

The 10-year Treasury yield, 4.96%. A sustained close above 5.05% after the FOMC would say the market sees more than a one-off hike; a move below 4.85% would suggest the event premium is clearing.

The two-year yield, 4.63%. Above 4.75% would imply a renewed cycle is being priced. Below 4.50% after a hike would be a classic one-and-done signal.

Five-year inflation compensation, about 2.40%. A break materially above 2.45% with real yields still rising would mean the market is beginning to doubt policy credibility rather than simply price tighter policy.

WTI at $100.05 and Brent at $104.61. WTI below $95 would remove a major source of near-term inflation anxiety; above $105 would keep household expectations and transportation pass-through in focus.

Michigan one-year inflation expectations at 4.6%. The 25 September final reading is the cleanest confirmation or rejection of the expectation shock.

Russell 2000 versus the S&P 500. Small caps underperformed by roughly 1.6 percentage points this week. A reversal after the Fed would indicate that the market is comfortable with the financing outlook.

Oracle and the broader AI infrastructure complex. If estimate upgrades cannot offset a higher real discount rate, the market is telling us the binding constraint has moved from demand to valuation and financing.

Our thesis in one sentence: rising oil prices helped lift inflation compensation early in the week, while Friday’s payroll surprise pushed up short-term nominal yields and five-year real yields; equity gains remained concentrated in the larger stocks.

Four developments shaped our view.

First, Friday’s jobs report eased concerns about the labour market. August payrolls rose by 162,000 against a Reuters consensus of 56,000. June and July were revised up by 55,000 in total, with July’s 23,000 decline becoming a 21,000 gain. Unemployment held at 4.1%, participation rose 0.2 percentage point to 61.6%, and annual wage growth slowed to 3.1%. September rate-hike odds briefly reached about 65%, then eased to 57% in Reuters’ afternoon snapshot.

Second, most of the weekly bond move preceded payrolls. From 28 August to 4 September, the five-year nominal yield rose 6 basis points while its real yield fell 1, lifting inflation compensation by 7. At ten years, nominal and real yields rose 5 and 1 basis points respectively; at thirty years, they rose 2 and zero. Inflation compensation did more of the work this week, reversing the previous week’s pattern.

Third, hiring still looks cautious. July JOLTS recorded 7.3 million openings and 5.1 million hires; ADP estimated 38,000 additional private-sector jobs in August; and initial claims were 206,000. The ISM Services Employment Index remained below 50 at 47.8, despite activity and orders readings above 60. These measures cover different periods and samples, but none points to a surge in layoffs.

Fourth, Broadcom’s AI sales exceeded its guidance, but the shares fell. Q3 AI semiconductor revenue reached $16.7 billion against a $16.0 billion forecast, and Q4 guidance rose to $21.7 billion. Shares nevertheless lost 2.7% on Thursday after total Q4 revenue guidance of about $34.8 billion fell short of the roughly $35.0 billion consensus cited by Reuters. Strong demand alone was not enough to satisfy investors.

The main challenge to our view is that oil prices can reverse quickly. Inflation compensation also includes risk and liquidity premia, so it cannot be read as a pure inflation forecast. Productivity offers another counterweight: Q2 output per hour rose 1.4% annualised, while unit labour costs increased just 1.2%, or 1.4% over the year. That is little evidence of a wage-cost spiral.

Our view would weaken if PPI and CPI are benign, Brent falls below $90, and five-year inflation compensation drops below 2.30% without a growth scare. That would suggest a temporary geopolitical premium rather than a lasting shift in inflation pricing.

Deep Dive I: Payrolls Beat. Hiring Still Looks Selective.

August payrolls beat expectations by a wide margin: 162,000 jobs were added against a Reuters consensus of 56,000. June and July were also revised up by a combined 55,000. July’s initially reported loss of 23,000 became a gain of 21,000, making the summer labour market look firmer than the earlier releases had suggested.

Figure 1. August payrolls repaired the headline

Key takeaway: August job growth, upward revisions and higher participation strengthened the report. Slower wage growth tempered the inflation concern.

Higher participation added some reassurance

The labour force grew by 683,000 and household employment by 569,000. Participation rose from 61.4% to 61.6%, while unemployment held at 4.1%. More people were looking for work and finding it, although household estimates are volatile: the employment increase was below the BLS’s roughly 650,000 significance threshold. Participation also remained 0.5 percentage point below January.

Average hourly earnings rose 0.3% in August and 3.1% over the year, down from July’s 3.2% annual pace. The average workweek increased by 0.1 hour to 34.4 hours. Together, the figures strengthen the case that the economy can withstand tighter policy, while offering little evidence that wage growth is accelerating.

Payrolls and hiring flows tell different parts of the story

These releases are not directly comparable. Payrolls measure net employment changes; JOLTS tracks openings, hires and departures; initial claims count applications for unemployment benefits; and ADP estimates private employment using a separate dataset. BLS private payrolls rose 127,000 in August, compared with ADP’s 38,000. The gap is worth watching, but one month’s disagreement does not establish a trend.

Job gains were also uneven. Restaurants and bars added 59,000 jobs and local government education added 42,000, together representing about 62% of the net increase. Manufacturing gained 16,000 and health care 13,000, while information lost 23,000. Education’s rebound partly reversed July’s decline, making seasonal effects relevant to the interpretation.

What the Fed learned

The report weakened the argument for holding rates solely because employment was faltering. Reuters’ market wrap put September hike odds at about 55% before the release, 65% immediately afterwards and 57% by the afternoon. Those are time-specific market estimates, not a Fed commitment. Next week’s inflation data will help determine whether the stronger jobs report leads to tighter policy.

What would change this view: another strong payroll report alongside wage growth above 3.5% would increase our inflation concern. Claims above 230,000 and renewed downward payroll revisions would point the other way. These are monitoring thresholds, not statistical rules.

Deep Dive II: Inflation Compensation Took the Curve Back

The previous week’s Treasury move was led by real yields, while inflation compensation fell. This week the balance reversed. The five-year nominal yield rose from 4.48% to 4.54%, and the real yield slipped from 2.18% to 2.17%. Subtracting the latter from the former gives an increase in inflation compensation from 2.30% to 2.37%.

Figure 2. Inflation compensation drove the measurable weekly repricing

Key takeaway: Medium- and long-term nominal yields rose with little change in real yields. Most of the increase appeared in inflation compensation.

At ten years, nominal yields rose 5 basis points and real yields 1, leaving a 4-basis-point rise in inflation compensation. At thirty years, real yields were unchanged and nominal yields rose 2. Treasury’s published real par curve starts at five years, so it does not support the same calculation for the two-year maturity.

Inflation compensation is estimated here as the nominal Treasury yield minus the real TIPS yield at the same maturity. It includes inflation-risk and liquidity premia as well as expected inflation. Differences in the instruments and curve construction also mean this is an approximate measure, rather than a directly tradable breakeven.

The timing helps distinguish the two moves

Figure 3. Tuesday and Friday were different rate events

Key takeaway: Five-year inflation compensation rose most on Tuesday. Friday’s jobs report coincided with a rise in the real-yield component.

Five-year inflation compensation rose from 2.30% on 28 August to 2.31% on Monday and 2.37% on Tuesday, then ended Friday at 2.37%. The five-year real yield was unchanged through Tuesday, fell on Thursday as Governor Waller signalled he could support a hold if inflation eased, and recovered 2 basis points on Friday. Friday only partly reversed Thursday’s decline.

Rising oil prices offer a plausible explanation for the earlier move. WTI gained 9.7% over the week to $91.48, while Reuters reported a 7.6% rise in Brent. Renewed U.S.-Iran fighting increased supply concerns as Hormuz shipping remained disrupted. On Tuesday, Brent gained 4.6% and the ten-year Treasury yield reached 4.79%. The timing supports an energy link, although it does not prove causation.

Long-term yields moved less

The thirty-year nominal yield rose only 2 basis points over the week, with no change in its real yield. The two-to-ten-year spread widened from 39 to 41 basis points. That modest steepening is consistent with an inflation premium and some policy repricing; it offers little evidence of a disorderly selloff concentrated in long-dated bonds.

Gold fell on Friday and the dollar initially rose after payrolls, then surrendered part of its gain. Both respond to several forces at once. The Treasury calculation helps identify which yield component moved, but it cannot isolate the cause. Our interpretation rests on the combination of that arithmetic, the release timing and the oil-price move.

Falsification threshold: five-year inflation compensation below 2.30%, alongside WTI below $85, would support a temporary geopolitical explanation. A rise above 2.40% after firm CPI data would strengthen the inflation concern. Neither threshold is conclusive on its own.

Deep Dive III: Hot Services Demand, Cold Services Hiring

Thursday’s ISM Services report showed the tension facing policymakers before payrolls arrived. The headline PMI rose from 54.1 to 55.4, while Business Activity reached 61.7 and New Orders 60.9. Employment remained below 50 at 47.8. Firms reported expanding business alongside falling employment, even though the employment index improved slightly from July.

Figure 4. Services activity expanded while employment contracted

Key takeaway: Strong activity and weak employment can coexist. Productivity, hours or capacity utilisation may help explain the gap, but the survey does not identify the cause.

Price pressures remained high. Services Prices rose from 70.3 to 72.6, matching its highest level since August 2022. Manufacturing Prices stayed at 71.1, while the headline Manufacturing PMI eased from 55.6 to 54.6. These are measures of how widely firms report changes, not the percentage rate of inflation or output growth.

The services survey complicates a simple overheating story. Orders were growing, but employment was still contracting. Firms may be meeting demand with existing capacity or better productivity. They may also be reluctant to hire while costs are uncertain. Rising oil prices could squeeze margins or encourage price increases before they change staffing plans.

Productivity is the counterweight to the inflation story

Revised Q2 nonfarm business productivity rose 1.4% annualised, as output increased 1.7% and hours worked 0.3%. Unit labour costs rose 1.2% annualised and 1.4% over the year. Manufacturing productivity increased 2.4%, while its unit labour costs fell 0.3%. These figures suggest contained labour-cost pressure, although they cover Q2 rather than the latest oil shock.

The Beige Book points to modest growth

The Fed’s Beige Book described modest overall growth and a very small rise in employment. Demand for workers weakened in retail and hospitality, while consumer spending edged up and tourism increased. Price rises were unchanged in pace in eight Districts, slower in three and faster in one. Input costs remained elevated. The picture is one of continued growth with uneven hiring and persistent cost pressure.

The report used information collected on or before 24 August, so it predates this week’s renewed escalation. It helps establish the underlying business conditions but cannot tell us how firms have responded to the latest rise in energy prices.

The non-obvious point: energy can raise inflation risks even when hiring is subdued. The services survey and Q2 labour-cost data are consistent with that possibility, but their different coverage and timing mean they cannot establish how firms are responding today.

Deep Dive IV: Broadcom Validated Capacity, Not the Multiple

Broadcom offered a useful test of whether strong AI infrastructure demand could translate into renewed share-price leadership. Its results supported the demand case: Q3 revenue reached $29.6 billion, up 86% year on year. AI semiconductor revenue was $16.7 billion, up 221% annually and 54% from Q2, beating the company’s $16.0 billion guidance.

Figure 5. Broadcom raised the AI revenue bar again

Key takeaway: AI revenue exceeded guidance, but the company’s overall outlook fell short of market expectations.

Broadcom forecast Q4 AI semiconductor revenue of $21.7 billion, up 236% year on year. Reuters also reported management’s expectation of about $115 billion in fiscal-2027 AI-chip revenue, up from more than $100 billion, and roughly $230 billion in 2028. These remain forecasts. Q3 free cash flow was $13.7 billion, or 46% of revenue.

Shares fell 2.7% on Thursday. Total Q4 revenue guidance of about $34.8 billion was below the $35.03 billion analyst estimate cited by Reuters, despite the strong AI outlook. That supports the view that expectations were demanding. The share-price reaction alone, however, cannot tell us how much reflected valuation, competition or concerns about future execution.

What this does to last week’s AI thesis

The results reinforce the case for strong AI infrastructure demand, but do not establish what investors are willing to pay for it. Broadcom offered accelerating sales and still underperformed on the day. For us, the useful lesson is that AI growth must be assessed against expectations for the whole business, including margins, competition and delivery risk.

The same applies across semiconductors. Revenue can grow rapidly while shares struggle if forecasts already assume that growth, margins disappoint or other divisions weaken. An attractive long-term earnings outlook does not, by itself, establish an attractive near-term entry price.

What would prove capacity leadership is back: sustained outperformance by Broadcom and semiconductors, alongside rising revenue forecasts, would support a recovery. Continued underperformance would leave valuation and execution concerns unresolved.

That is correct as well; it is the difference between the two that is most important.

The impact of one trading session was greater than that of the entire week.

On Thursday, 27 August, the S&P 500 gained 55.29 points. For the entire week, the net increase was 37.39 points, and the other four sessions lost 17.90 points in total.

Thursday’s gain amounted to 148 percent of the week’s total. On the Nasdaq, the figure was even more striking: it had risen by 411.15 points on a single day compared to a weekly gain of 221.97, or 185 percent. On the other four days, 189.18 points were lost.

Most companies, underneath the surface, took the opposite course. The Russell 2000 fell by 1.51%. The proportion of S&P 500 stocks trading above their fifty-day average decreased from 57.76% to 53.47%. Over the course of five sessions, the main index had outperformed its equal-weight counterpart by 93 basis points.

There is a good argument against viewing this as an indication of weakness, and it should be given consideration. If profit growth is actually concentrated among a small number of companies, then a weighted index beating its equal-weight counterpart is only showing that situation, not pointing to a problem. Nvidia’s revenue increased by 106% over the last year, whereas the average company in the index did not.

The risk indicators align with the skeptics’ perspective since volatility had fallen from 15.13 to 14.46, with investment-grade credit spreads narrowing by 2 basis points and high-yield spreads narrowing by 7, both figures being those for Thursday. Generally, credit markets pick up on problems before the stock markets do, but this week they showed no signs of concern. People who are describing this as a risk-off week have not looked at the bonds, the credit markets, the dollar, or volatility.

The difference between us is less than it appears. The fact that there is a breadth gap is not proof of stress; it is instead a sign of a change in the discount rate, and such a change affects two channels simultaneously. When real rates rise, expensive growth companies are compressed due to their valuation duration, since a large part of their value is tied up in the distant future. At the same time, smaller companies are squeezed as a result of higher refinancing costs and their greater reliance on external capital. Though the mechanisms are different, the result is the same, and in both cases the effect is felt most where the index has the fewest holdings.

Nvidia performed better in every aspect, but it proved to be one of the least effective methods of investing in AI to own its stock this week.

Three companies gave their reports after the market closed on Wednesday. Nvidia’s results were almost perfect according to its own standards: it achieved revenue of $96.2 billion, which represented a year-on-year increase of 106% and a quarter-on-quarter rise of 18%, with data center revenue amounting to $89.0 billion and the gross margin remaining at 75.0%. The company stated that its guidance for the October quarter was about $108 billion, compared with consensus estimates of nearly $105 billion, and made it clear that the guidance assumes no data center computing revenue from China.

Nvidia’s share price increased by 1.32 percent that week.

Salesforce rose by 22.39 percent, and CrowdStrike increased by 13.78 percent. Although Marvell surpassed its own guidance and had forecasted better results for the following quarter, its share price fell by 8.61 percent. The semiconductor sector as a whole declined by 2.20 percent.

Microsoft, since it had reported nothing, increased in value by 6.27 percent and was the most successful of the seven largest technology companies.

This is the key point of the week and it doesn’t indicate that investors doubt the demand for AI; rather, it reveals which companies’ balance sheets the market wants to gain from that demand. Over the course of five sessions, funds moved from companies that were developing their AI capabilities to those which were already earning money from them.

The most unusual action of the week was taken by Nvidia; its shares rose by 8.74% on Thursday before falling 4.57% on Friday and thus losing back 57% of the previous day’s gain within a single session.

The hawkish morning priced in less inflation, not more.

It is here that the majority of the reports made the mistake.

On Friday the chairman of the Federal Reserve delivered his first speech at Jackson Hole and the probability that interest rates would be increased in September rose from around 35 per cent to the upper 50s. The headlines emphasized a cautious central bank, higher yields and falling share prices, indicating that there was an inflation problem.

The market indicated the contrary.

For the five-year maturity, which is the shortest at which this analysis is applicable, the nominal yield rose by 5 basis points and the real yield increased by 9; this indicates that the inflation compensation, which is the difference between the nominal and real yields, decreased by 4 basis points. At the ten-year point it decreased by 3, and at thirty years by 1.

Normally, an inflation scare causes the price of expected inflation to rise but this time it declined at all maturities for which it could be measured.

On the same morning, three other prices were agreed upon. The rate of inflation for households over the next year was lowered to 4.0% from an initial estimate of 4.3%, this figure being less than the 4.2% recorded in July. The October contract for Brent dropped by 5.38% during the week. Gold also declined by 2.04%.

When the market increases interest rates while at the same time lowering its expectations of future inflation, it is acting to demonstrate credibility, not out of fear of inflation.

One obvious limitation is that the Treasury does not provide a real yield curve for maturities of less than five years, so we can’t analyze the two-year change in the same way.

What it changes

The most important point for investors is the discount rate, not the company’s earnings. When the primary risk is higher real interest rates rather than inflation, gold, commodities, and inflation-linked bonds offer less protection against this type of shock than is generally expected. Moreover, at the present time, it is less risky to hold long-term bonds than to hold long-duration stocks. Investors who currently base their AI investments on rising valuations rather than higher earnings estimates face real rate risk, which has become both more expensive and more apparent this week.

Before they are reported elsewhere, here are two points to make clear.

The payrolls figure hasn’t actually been reduced. The statistics bureau has issued a preliminary estimate stating that employment in March 2026 will later be lowered by 79,000 and private employment by 178,000. Since both amounts are less than the average annual revision of 0.2% over the last ten years, this is a normal adjustment. The changes will not be reflected in the published payrolls until February 2027. People who claim the payrolls have just been cut by 79,000 are describing something that hasn’t happened.

Second, the majority of the retail margin increases this week were the result of tariff refunds, not because of improved business performance. For instance, 680 of the 850 basis points of margin gained by Dollar Tree, about 81 of the 127 gained by Dollar General, and 37.5 of the 60 gained by Best Buy were attributable to refunds. This shows that between 62.5% and 80% of the reported improvement came from one-off refunds. These are not subsidies, they won’t occur again, and next year’s comparison will be more difficult.

What would prove us wrong

If the two-year yield falls below 4.24 percent, yet the equal weight index continues to lag behind the cap-weighted index, then our explanation based on real rates for the breadth gap ceases to be valid, and some other factor must be involved. We will admit that if this situation arises.

With regard to inflation, if the five-year compensation rate goes above 2.40% while the two-year yield continues to rise, this would indicate that the market is anticipating an inflation issue which the central bank is having difficulty in controlling.

If the next large-cap report causes the market to rise next week, then this was a one-week positioning move and we should drop it.

The complete weekly report – this includes the cross-asset scoreboard, the daily breakdowns, a detailed analysis of the earnings for the week’s top performer, and a table listing the strongest evidence against our thesis – is available at moatpeak.com.

This content is for educational purposes only and does not constitute personal investment advice. MB “MoatPeak Group”.

The fact that last week demonstrated that favorable economic figures can still be bad news, even when the long-term real interest rates remain unchanged.

The U.S. Composite PMI reached 56.0, the highest figure it had been since over four years earlier. Yet all of the major U.S. stock indexes fell, the Nasdaq declining by 2.05 percent. The drop was not due to slowing growth but rather to the costs associated with growth.

Now investors are taking each positive data point and turning it into two questions: does it serve to keep inflation high? and who will have to borrow in order to finance the next phase of investment? The first of these questions applies to the energy sector; the second applies to AI.

We do not consider the latest increase in Treasury yields to have represented a permanent change in real interest rates. During that week, inflation expectations for the five-year term rose by 10 basis points, although the 30-year real yield remained unchanged. The main cause of this was higher energy prices, together with investors pulling out of overpopulated trades. Underlying all of this, however, there is a more significant change taking place: AI companies are now financing their infrastructure through borrowed funds rather than drawing on their own cash.

It by no means signifies that the demand for AI is disappearing; it only means that the problems facing the sector will be different.

The latest change in the rates was due to inflation, not because of the length of time that money is invested.

Nominal yields consist of two components: the real yield, which represents the cost of waiting, and a payment for inflation, which covers the expected decrease in buying power. Although this distinction may appear to be technical, it is important when the two parts change for different reasons.

Last week, the five-year inflation compensation increased by 10 basis points, the 30-year real yield ended the week unchanged, Brent went up by 6.6 percent, and U.S. retail diesel rose by 19.7 cents. By the end of the week the market was demanding a higher price for near-term inflation risk but had not assigned a permanently higher real price to capital.

The situation was far from smooth. On Monday, August 17, the 30-year real yield reached 3.06%, whereupon it exerted pressure on stock prices before dropping again. Although Treasury buybacks helped to improve liquidity for longer-term bonds and provided a certain amount of short-term relief, they are unable to address budget deficits, inflation uncertainty, or the continuing requirement for capital.

It is important to realize that a sharp fall in valuations resulting from short-term inflation is quite different from a gradual increase in real interest rates. The two should not be confused. Yet inflation should not be ignored because it can cause central banks to remain cautious, reduce the amount that households are able to afford, and make it more difficult for investments to be profitable, even if long-term real rates remain low.

Why momentum cracked while speculative innovation rallied

The stocks performed contrary to expectations. Although steady long-term real yields usually help companies that depend on future earnings, this time, technology stocks fell by 3.53%, and Utilities declined by 3.48%.

A more appropriate example can be found within the market. The MTUM Momentum ETF dropped by 3.80%, the ARKK Innovation Fund increased by 6.30%, and Bitcoin surged by 23.3%. We believe that investors were disposing of the popular winning stocks rather than spreading their avoidance of risk across all areas. The rise in the most speculative assets was most likely due to short covering.

The semiconductor stocks also displayed the same kind of division. The SOX index fell by 5.45%, yet the gap between the best and worst performers was almost 20 percentage points. Marvell increased by 6.77% because of its own strengths, whereas Arm and Intel declined by 12.93% and 12.13%, respectively. Nvidia had dropped 4.64% prior to the release of its earnings report.

The market wasn’t just choosing to take on more or less risk; rather, the company’s narrative, its financial condition, the way it was valued and its position before the earnings announcement carried more weight than the general index. Therefore, the large fall in the headline index did not give the full picture when compared with the differences between the individual stocks.

Credit could turn out to be the new barrier to the growth of AI.

The most significant change could be taking place outside the stock market. Tech companies have raised around $220 billion for data centers that are planned for 2026. Nebius has issued a $5 billion convertible bond. Nvidia has provided backing for $105 billion in lease obligations for SB Energy. The spread on Amazon’s $25 billion bond was nearly double what it was last year.

Instead of using their own cash, companies are now obtaining money by borrowing from the credit markets. For many years, investors were unsure whether spending on AI would generate sufficient revenue. At this stage, the issue is who will assume the risks involved in borrowing, using leverage, and refinancing in order to set up this infrastructure.

We don’t see this as proof that the AI buildout is coming to an end since demand can still be strong even if financing conditions worsen. The difference lies in the fact that cheap credit has now become another assumption incorporated into the AI thesis.

AI infrastructure is now having to compete with the government for long-term financing, which can result in wider spreads and lower valuations even if demand remains strong. The important issue is not merely the amount of growth in computing and data centers, but whether each new project is able to cover its cost of capital.

For investors, it is important to tell the difference between companies that can pay for investments themselves and those that need to keep borrowing; between those who own key assets and those who borrow to use them; and between steady cash flows and riskier bets. The next downturn in AI may show up in credit markets before it affects revenue.

Retail exhibited a less substantial instance of the same trend.

The figures from the retail sector presented a similar analytical issue in a more limited form: the overall figures appeared stronger than the underlying economic data. Each of the seven major retailers examined reported results that included one-time IEEPA tariff refunds.

Walmart received $2.9 billion and used the money to reduce its prices; its adjusted operating income increased by 28.8%, although the growth would have been 17.4% if the refund had not been included. Target performed even better, with store traffic rising by 3.6% and earnings per share increasing by 20% without the refund. The situation with TJX was more complicated since Marmaxx sales only increased by 1%, rather than the 2.7% that had been expected.

People are still spending, but the strength is focused on certain areas. The mass-market and discount types of retail are faring better than those in the big-ticket and confidence-dependent sectors. Moreover, within the world of discount retail, the category is not a single entity.

The lesson involves more than just retail. In fact, temporary benefits, simpler financing, or short-term assistance can make the results appear better without, in fact, improving the business. Investors should ignore the headlines and concentrate on the quality of cash flow.

A better dashboard for the next phase

In the old days, people mostly wanted to know whether growth was accelerating and whether the Federal Reserve was easing its policy. That is no longer sufficient. We think the following four questions are now more important:

Is the rate at which inflation compensation is increasing exceeding that of real yields?

What AI investments are based on debt or on contingent support and which of them are financed internally?

Whether or not the market breadth is actually improving, or whether capital is merely rotating among a small number of sectors?

Are the results obtained by consumers due to traffic, the size of the tickets, or to temporary fiscal advantages?

What would change our mind

A long-term decline in crude oil prices and in the amount of inflation compensation would undermine the energy aspect of this argument. If substantial issuance of AI debt continues without there being wider spreads or a reduction in market access, the financing risk would still seem to be under control.

The other risk is just as significant. If long-term real yields rise continuously rather than just temporarily, the market would be experiencing a structural real-rate shock, not merely a temporary inflation spike. In that case, the valuation pressure would be wider and more long-lasting than our present understanding.

The market is not at this moment rejecting growth; rather, it is now wondering who is going to finance it. This is a more difficult question, but also one that is of greater significance. Since it is becoming more difficult to obtain capital, a company’s financial strength and cash flow may be just as important as its growth story.

MoatPeak conclusion: The market is not turning away from growth, but is questioning who will pay for it. The next big issue for AI may show up in credit markets before it affects revenue.

MoatPeak Team

First, the drop was in petrol prices, but those have already gone back up.

A 2.9% drop in gasoline prices helped bring down July’s inflation number. This was the most encouraging detail in the report.

But in the week after the data period ended, Brent crude oil jumped 5.95% to $88.52 because of new tensions near the Strait of Hormuz.

So the lower inflation everyone cheered was for a month that’s already behind us, and the key input cost has already risen again. If Brent stays at this level, it will push up inflation in September and October instead of helping.

This isn’t a prediction – it’s just pointing out which months these numbers actually reflect.

Second, short-term rates eased, but long-term rates rose.

Over the same week:

The two-year Treasury yield dropped by 2 basis points to 4.17%. Meanwhile, the thirty-year yield rose 6 basis points to 5.25%. The latest thirty-year auction cleared at 5.216%, the highest since 2001.

The yield curve steepened after the soft inflation report, which is the opposite of what you’d expect if prices were the main story.

This isn’t just happening in the US. German ten-year yields went up from 3.13% to 3.20%, and UK ten-year yields rose from 4.93% to 5.05% in the same period. When long-term yields rise in three countries despite lower inflation at home, the cause must be global.

What does it mean for you as an investor? The discount rate applied to anything with a value ten years out has gone up, not down. That is a headwind for long-duration growth assets, for property, and for utilities, regardless of what the Federal Reserve does in September.

Third, in the AI sector, the market shifted what it values.

This is the part that will matter most, and it happened in a single week of price changes.

Suppliers saw big gains: SanDisk rose 35.4%, Super Micro 28.0%, Samsung 18.8%, Western Digital 17.2%, SK Hynix 15.7%, and Lumentum 13.6%.

Assemblers, meanwhile, fell: Applied Materials fell 5.9% and Cisco 8.0%.

Cisco’s revenue held up, but its adjusted gross margin dropped from 68.4% to 66.3%. The company blamed higher memory costs and the lower margins of large AI cluster projects.

Applied Materials had a record quarter but still fell nearly 6%, partly due to a $253 million export compliance charge and ongoing exposure to China, which now accounts for 28% of its revenue.

Taken together, these changes show a shift. For two years, investors in AI focused on revenue growth. Last week, they started rewarding profit margins instead, moving value from assemblers to suppliers.

We call this the margin tax. Memory suppliers are now setting the terms for everyone further down the supply chain, and you can see the impact clearly in company results.

There is a valid exception: Lumentum rose 13.6% despite reporting a $7.2 billion GAAP loss. That loss was a non-cash charge from retiring convertible debt, and the company’s core business was strong. So the market isn’t just reacting to headlines. It’s looking past accounting noise and focusing on real profit trends, which is more nuanced than it seems.

Now to the consumer, a part of the story many prefer to ignore.

Retail sales fell 0.6%, contrary to expectations. The control group also fell by about half a percent. Consumer sentiment fell to 51.0 from 55.2, with declines across all groups. Existing home sales slipped 1.7% as the ten-year yield hit 4.68%.

On the other hand, initial jobless claims were 209,000, showing the labor market is still holding up.

This creates an uneasy mix of weak demand but few layoffs. In this situation, the idea that ‘bad news is good news’ no longer works, because if demand drops too much, lower earnings will hurt more than any help from a Federal Reserve pause.

We’re keeping an eye on the following three points:

First, if the thirty-year yield goes above 5.35%, that would signal a real shift in the cost of long-term borrowing, not just a temporary move.

Second, whether Nvidia confirms the ‘memory tax’ on August 26. Cisco blamed memory costs for its margin drop, and Nvidia is the biggest buyer of those components. If Nvidia faces the same issue, it points to a wider supply chain problem, not just a Cisco issue.

Third, whether Walmart’s results confirm the retail control group data. A 0.4% drop could be just noise or the beginning of a trend, and Walmart, the largest retailer, reports on the 20th.

What would prove us wrong? Brent fell below $83.55 and the thirty-year yield dropped under 5.19%. Both are clear and easy to check.

You can find our full scenario framework, probability weights, and detailed supply chain analysis at moatpeak.com.

Educational research only. Not personalized investment advice. MB “MoatPeak Group”.

Everyone is giving the same explanation. On Friday morning, weak labor data came out. Payrolls dropped by 23,000. Fewer jobs make it less likely the central bank will raise interest rates, and lower chances of a rate hike are good for stocks. Bad news turns into good news, and the market closes at a record.

It sounds like a neat story, but the timing does not add up.

Most of the week – 92% – was already over before the news came out.

Let’s look at what really happened, day by day.

Monday and Tuesday added 246.80 points, which was 92% of the week’s total – before any labor data was released. Friday, the day getting all the credit, only added 47.68 points – about 18% of the week’s gain.

So what drove Monday and Tuesday? Two things, and neither was about interest rates. First, hopes for a shipping deal in the Strait of Hormuz pushed oil down more than 8% for the week. Second, earnings: Palantir reported after the close on Monday and its stock jumped nearly 40% over the week, thanks to a 63% free cash flow margin.

This is important, and not just a technical detail.

If you think the rally happened because of rate relief, you might expect more rate relief to keep the rally going, so you would hold your position until Wednesday’s inflation numbers. But if the rally was really about a shipping lane and two earnings reports, you are actually betting on a diplomatic negotiation that no one can predict, in a strait that has been mostly closed since February.

These are two very different investments. They just happened to give the same result this time.

We date a move from its loudest moment, not its largest one.

Once you notice this pattern, you start seeing it everywhere. It becomes a quick test: change the starting point and see if the story still makes sense.

Here are three more examples from the same week where the story does not hold up.

Test one: Most of last week’s so-called disasters happened in just one bad afternoon.

Three companies were widely reported as having collapsed after their earnings reports.

AMD “fell 7%”. That happened on Wednesday. But over the whole week, AMD actually rose 1.51%. The number people remember is from the reaction day, and it is misleading.

Datadog “fell 19%”. That was on Thursday. For the week, it was down 12.70%. Still a big drop, but about a third less than the headline number.

Honeywell Aerospace “fell 23%”. That was on Thursday. Over the week, it was down 18.49%. It is important to note this is HONA, the aerospace business that started trading separately on June 29. The parent company actually rose 1.30% that week. So, anyone who heard “Honeywell fell 23%” and owns Honeywell shares was thinking of the wrong company.

None of these numbers are made up. Each one is a real close-to-close move. They are just measured over a time frame picked for drama, not by the calendar.

Test two: The AI trade did not actually split by region.

The most repeated market observation of the week was that artificial intelligence has stopped being a global trade and become a Western one. The evidence looked overwhelming. The American semiconductor index rose 9.25%. Samsung fell 12.00%. SK Hynix fell 17.23%.

That is a 26-point gap in just five sessions. Case closed, or so it seemed.

But if you shift the starting point by just one day, from Friday, July 31 to Thursday, July 30, the picture changes.

The American semiconductor index: +9.32%. Samsung: +11.59%. SK Hynix: +7.56%.

Everything went up.

That is because Friday, July 31, saw the biggest single-day gain in Korean market history, up 17.91%. SK Hynix jumped 29.95% that day, and Samsung rose 26.81%. If you start counting from Saturday morning, you are only measuring the pullback from a record high, not the record itself.

The regional divergence was real for exactly one choice of start date.

Test three: Let’s apply this to our own position.

It is always easier to spot mistakes in others, so let’s run the same test on one of our own positions.

We have often said this market is dangerously narrow, and last week seemed to prove it. The index rose 3.58%. The equal-weighted version, where every company counts the same, rose 2.36%. The median stock was up 1.99%. Nvidia alone made up about a quarter of the index’s total gain.

By any measure, that is a narrow week.

But if you look back three months instead, the picture changes.

During that time, the equal-weighted index is up 8.05%, while the main index is up 5.26%. On average, smaller companies have outperformed the giants by almost three percentage points.

Both facts are true. Only one of them is a trend, and it is not the one that suits our argument. To be honest, last week was narrow, but the past three months were not. One week of narrow trading is something to watch, not something to draw big conclusions from. If it keeps happening, we will point it out again. We are not going to claim that five days make a new trend.egime.

The real new development was hidden inside a falling share price.

Amid all this, one truly new fact came out last week. It was buried in one of the misquoted numbers, which is probably why few people noticed.

On its earnings call, Datadog said its biggest customer – described only as a leading artificial intelligence company – had signed a nine-figure renewal, but with a usage reduction starting in the third quarter. The finance chief said this reduction is already included in their guidance.

Think about that for a moment. A major AI company reviewed its monitoring costs and decided to cut back.

For two years, everyone assumed AI workloads would only grow, and that selling to AI companies meant a one-way market. Now, a big customer is publicly choosing a different path. Not because the technology failed, but because it was too expensive and someone found a way to optimize costs.

To be fair, Datadog’s revenue actually grew faster, reaching 36% growth for the quarter. The company also said it now has more than 750 AI-native customers, and that group is still growing. The share price dropped because of the guidance and the risk of relying too much on one customer, not because the business had already declined.

But the direction of travel is the point. When capacity is scarce, you pay whatever it costs. When it stops being scarce, procurement gets involved. This is what the beginning of that looks like.

Two things that do not depend on where you start the clock

The first is about the central bank. On July 29, four days before the week started, the rate-setting committee voted 9 to 3 to keep rates at 3.50% to 3.75%. All three dissenters wanted to raise rates.

Sit with that. The market spent last week celebrating that a September rate rise had become less likely. Three of the twelve voters had wanted one immediately, days earlier. The odds of a September hike did fall hard over the week, from roughly two-thirds on Monday to about 40% on Friday. But the conversation in that room is still about whether to go up. Nobody is discussing cuts.

The second point is about workers’ share of national output. In the second quarter, it was 52.9% – the lowest since records began in 1947. Real hourly pay also fell at a 3.1% annual rate during the same period.

This fact does not depend on when you start measuring, because it is not about change. It is a level, and it is the lowest ever recorded.

Companies are protecting their profit margins, and they are doing it at the expense of household income. This helps profits – until it starts to affect what households can spend. Retail sales numbers come out on Friday.

What we are keeping an eye on this week

Taiwan Semiconductor reports July revenue on Monday, giving us the first monthly look at the AI supply chain since Asian chip stocks dropped. U.S. inflation data comes out Wednesday morning, and that is the main event: if it is above 3.6%, talk of rate hikes will return. Retail sales are out on Friday and will show if the record-low labor share is affecting spending yet.

And a suggestion you can use on any market story you read this week, ours included. Before you accept an explanation for a price move, check when the move happened and when the explanation happened. If the explanation arrived second, it is a story about the story.

Our comprehensive weekly research, including what we think holds Tuesday’s highs and the level that would change our mind, is at moatpeak.com.

Educational research only. Not personalized investment advice. MB “MoatPeak Group”.

The strange part is that this was not a story about who is spending and who is not. They are all spending. It was a story about who can show the money coming back.

The question the market started asking

For the last two years, a company could announce an enormous artificial intelligence budget and watch its share price rise. Ambition was the product. The bigger the number, the more the market liked it.

That stopped last week. The question changed from how much are you spending to what is it giving you back, and can we see that in the cash, this quarter, not in a promise about 2030.

Microsoft could answer. Azure, its cloud business, grew 43 percent. Amazon could answer too, with its cloud arm growing at its fastest pace in eighteen quarters while keeping its margins healthy. In both cases the money went in one end and visibly came out the other.

Meta could not answer as cleanly. Its revenue grew 28 percent, which in any normal year would be a triumph. But the cash it had left over after paying for everything, including its building programme, fell by more than 90 percent from a year earlier, down to a number smaller than the dividend it pays its shareholders. Revenue was not the problem. What the spending did to the cash was the problem.

Apple fell for a different reason. It had a record quarter and still dropped, because of what it said about the months ahead, including higher memory prices. Even a record is not enough if the next chapter looks harder.

The mechanism, in plain terms

It helps to know why heavy spending hits the accounts the way it does.

When a company builds data centres, it pays cash now. In the profit statement, though, that cost is spread over many years, a little at a time. That spreading is called depreciation. So in the early years of a big building programme, the reported profit can look healthy while the actual cash is draining away.

That gap is fine when the buildings quickly produce revenue. It becomes uncomfortable when the spending keeps growing and the revenue arrives more slowly, because the company has to find the money somewhere. It either uses its own cash, or it borrows, or it sells new shares. Each of those costs something, and each one is more expensive now that interest rates are higher for longer.

That is the real shift. The market has started to treat artificial intelligence not as a story about growth, but as a construction project that has to be financed. And it is asking each company the same simple question a bank would ask: can you pay for this out of your own pocket, or do you need to keep coming back for more.

Where it gets uncomfortable

There is a second layer, and it is the one we spent the most time on last week.

Some of the profits reported this season were not really cash profits at all. Alphabet reported $9.11 of earnings per share. When we stripped out a one-off accounting revaluation of a stake it holds in another company, our adjusted figure was closer to $2.85. Tesla showed a similar pattern. Both companies reported records. Both actually burned cash in the quarter.

Nothing improper is happening here. The accounting is legitimate. But an accounting gain from re-marking the value of something you own is not the same as a customer paying you money, and the two behave very differently when conditions get harder. In a year when everyone is spending heavily on the same thing at the same time, the difference between profit on paper and money in the bank is exactly where the risk hides.

That is why the market’s reaction was not really about one bad quarter at Meta. It was investors deciding, all at once, to look at a different number than the one in the headline.

The strongest case against us

We should be honest about how this argument could be wrong, because the counter-case is genuinely strong.

The demand is not fake. Cloud revenue is accelerating at the biggest providers, not slowing. If you believe artificial intelligence is a decade-long buildout, then a company spending heavily today and showing weak cash flow is not failing. It is investing, and the market is being short-sighted in exactly the way it usually is.

There is history behind that view. Plenty of companies have been punished for spending on infrastructure that turned out to be the foundation of everything they became. A quarter of negative cash flow, in a business that grew revenue 28 percent, may look absurd as a reason to sell in three years’ time.

We take that seriously. Our argument is not that spending is bad. It is narrower: the market has changed the terms on which it will finance that spending, and companies that need outside money to keep going now pay a real price for it.

What would tell us we are wrong

We would rather name this in advance than explain it away later.

If, over the next quarter or two, the heavy spenders show their cash flow recovering while revenue growth holds, then the market’s punishment was a timing problem and not a verdict, and we would say so. Equally, if investors go back to rewarding the biggest budgets regardless of cash conversion, then the shift we are describing did not really happen, and we will mark that against ourselves.

Those are the things we will be watching, and we will report them either way.

The quiet part underneath

One last thing worth noticing. While three enormous companies added 1.5 trillion dollars, the average stock barely moved. The equal-weight version of the index, which counts every company the same rather than letting the giants dominate, rose less than one percent. Smaller companies went nowhere.

The index looked healthy last week. Most companies in it were not really part of the story. That is worth remembering whenever someone tells you the market had a good week.

If you would like the full weekly, with the numbers behind all of this and the rest of the week’s research, it is on the platform at moatpeak.com.

Stay calm, and look for the moat. The MoatPeak Team

MoatPeak publishes general, non-personalized investment research and scenario analysis. It does not consider any reader’s individual circumstances, objectives or risk tolerance. All investing involves risk.

Start with the setup, in one picture. For the week into July 10, the mega-caps advanced while the broad market contracted underneath them. The Nasdaq and the semiconductors rose, but the Russell 2000, the Dow and the equal-weight S&P all fell. That is index growth without participation, and a market held up by a handful of names is a market with no cushion.

The second half of the setup was valuation. When a market is priced for flawless execution, it has no risk budget left for a shock. We flagged that any earnings disappointment, or a fresh jolt of imported oil inflation, would land hard precisely because the price left no room to absorb it. That was the fragility. This week supplied the shock.

Thread one: the crash had an address. When the selling came, everyone reached for the same phrase, the AI bubble is popping. The flow data disagreed. Foreigners pulled a record 137 billion dollars out of Asian tech in the first half of this year, the fastest pace in sixteen years, and the epicenter was Korea, not the S&P. This was a rebalancing out of the most crowded winners, not a fresh verdict on whether AI demand is real. A crowded trade does not need bad news to fall. It just needs everyone to stand up at once.

Thread two: a shortage is not a glut. The trigger that week was a phone maker raising its prices and blaming a memory shortage. The market sold every chip stock on demand and margin fears. Read that back slowly. A shortage is the opposite of a glut. When a buyer that large cannot get enough memory and has to pay up, that is scarcity, and scarcity is pricing power for the supplier, not evidence the customer is leaving. Our weekly work already had the concentration mapped, with the memory names doing the heavy lifting.

Thread three: the wrong thermometer. Into all of this, a soft consumer-price print sparked a relief rally in rates and long-duration equities. The same month, US import prices ran plus 7.1 percent from a year earlier, the hottest since 2022, and plus 4.2 percent even after stripping fuel out entirely. The consumer price index makes the headline. Import prices and the Fed’s preferred gauge quietly set the policy path and squeeze corporate margins. The market cheered the thermometer that flatters, not the one that binds.

Thread four: the un-crowded trade is now crowded. The reflex response to the unwind was the rotation, own the un-crowded, not-AI stuff, energy, defensives, nuclear. That was a good idea a year ago because almost nobody was doing it. Now every research desk prints it, which means the un-crowded trade is quietly becoming the next crowd. Contrarian is a moving target, not a destination you arrive at and park.

The through-line is one discipline. Separate the mechanism from the mood, because why a market moved tells you more about what comes next than how far it moved. A narrow rally, a rebalancing rather than a rout, a shortage read as a glut, an inflation gauge chosen for comfort, and a contrarian trade going mainstream. None of these are visible if you only watch the index close.

That is the whole method, dated and in writing, every week. The full weekly deep dives, the scenario odds and the exact conditions that would prove each call wrong are at moatpeak.com.

Educational research only. Not personalised investment advice. MB “MoatPeak Group”.

A rally this narrow is not a sign of strength. It is a concentration of risk. When a handful of names are doing all the lifting, the whole index quietly becomes a bet on those few names staying perfect. That is a fragile place to stand, because it removes your margin for error. If even one of those leaders stumbles, there is no broad market underneath to catch the fall. The index looks resilient. The structure holding it up has narrowed to a ledge.

You got a preview of what that fragility looks like last week, and it came from an unlikely place: a blowout earnings report. Samsung posted an operating profit nearly nineteen times larger than a year ago, a genuine record, powered by exactly the AI-memory demand everyone is betting on. The stock fell almost 7 percent. Not because the number was bad, but because the number was already in the price. After the shares had run up around 150 percent, a perfect quarter was simply the cost of admission, not a reason to go higher. That is what “priced to perfection” means in practice, and it is the trap sitting under this entire earnings season. When expectations are this high, even excellence can disappoint.

There is a deeper shift underneath all of this. For a couple of years, bad economic news was good news for stocks, because it meant the Fed would ride to the rescue with rate cuts. That reflex is breaking. The Fed’s latest minutes did not just walk back the promise of cuts, they revealed a committee where some members are now arguing for hikes. And this is happening at the exact moment when even mainstream economists are pointing at the AI buildout itself as a new source of inflation, through the enormous electricity demand it puts on the grid. Sit with that. The boom everyone is celebrating may be part of the reason the rescue everyone is counting on is getting harder to deliver, right when the market leans on it hardest.

Then, over the weekend, the picture got harder. The long-running crisis in the Strait of Hormuz escalated again. After Iran fired on a commercial ship and declared the strait closed, the United States struck back, and a waterway that carries roughly a fifth of the world’s oil is contested once more. Brent crude is back above 76 dollars. We are not going to pretend to forecast a war, and we will not treat a serious conflict as a trading game. But the market mechanism is straightforward and worth understanding: higher energy prices act as a tax on everyone and as fuel for exactly the inflation that is keeping the Fed hawkish. And they press hardest on the parts of the market that were already weakest, the smaller and more indebted companies furthest from the AI story.

Put the pieces together and you get the real picture of this market. A narrow, expensive rally, leaning on a few perfect companies, into a Fed that has stopped promising rescue, with a fresh energy shock pushing inflation the wrong way. None of that is a prediction of a crash. It is a description of fragility. The index can absolutely grind higher from here. But it is doing so on a narrower base, with a thinner safety net and a higher cost of being wrong than the calm headline number suggests. The useful move right now is not to turn bullish or bearish. It is to know which of these forces your own portfolio is quietly betting on.

That is the work we did this week. Our latest weekly note lays out the full framework: how the AI resilience and the macro fragility fit together, the specific scenarios we think the next few weeks run through, and the exact signals that would prove the thesis right or wrong. If you want the whole map instead of the headline, it is at moatpeak.com.

The free letter is where we think out loud. The scenarios, the levels, and the daily research are the desk’s job at moatpeak.com.

Educational research only. Not investment advice. MoatPeak Group, UAB.

A record high is supposed to mean confidence — lots of companies doing well, money flowing in across the board. That isn’t what this was. Underneath the record, the money wasn’t spreading out; it was huddling up. Investors sold the exciting, economically-sensitive things — chips, smaller companies — and crowded into a handful of giant “safe” names and the dullest, steadiest corners of the market: utilities, household staples, health care. That’s not an everything-rally. It’s the portfolio version of moving to higher ground. A record made by everyone crowding through the same few doors isn’t strength. It’s fear wearing a party hat.

Two things give the game away. The first: while the Dow set records, the index of small companies actually fell. Small caps are the part of the market tied most tightly to the ordinary domestic economy, and they didn’t get the invitation to the party. The money that did move went hiding — into the sectors you buy when you want shelter, not the ones you buy when you’re optimistic. When the leaders of a “record” week are the electric utility and the toothpaste maker, the market is telling you something it isn’t saying out loud.

The second tell is quieter and harder to wave away. A weak jobs report — and this week’s was weak, barely half the hiring forecasters expected — normally pushes longer-term interest rates down, because slower growth means cheaper money ahead. This time the long end refused to fall, and the yield curve steepened instead. In plain terms: the bond market looked at the very same weak jobs number the stock market was cheering and drew the opposite conclusion. Not “the Fed will ride to the rescue,” but “growth is cooling and the costs aren’t.” When stocks and bonds read one fact in opposite directions, bonds are usually the one worth believing.

Underneath all of it, the story that has carried this whole market took its first real hit. For two years the case for anything with “AI” attached rested on a single word: scarcity. Computing power was supposedly endlessly scarce, so whoever made the chips could name their price. Then this week Meta — one of the very biggest buyers of those chips — signalled it might start selling its spare computing capacity back to everyone else. A shortage story dies the instant its biggest hoarder turns into a seller. That is what the chip crash was really pricing: not a bad few days, but the first crack in “shortage forever.”

It’s also worth knowing how the calm surface got manufactured, because it makes the calm harder to trust. A couple of the market’s giants quietly papered over the damage — a jump in Apple on product-pipeline buzz did a lot of the index’s heavy lifting, holding the average up while the typical stock sank beneath it. And the “good news” wasn’t always what it looked like: one carmaker reported record deliveries and the stock fell anyway, because the market finally looked past the volume to the margin it cost to get there. A record you have to squint to believe is a record worth squinting at.

So the useful way to read a week like this one isn’t “Dow record, all clear.” It’s almost the reverse. Put a single question to any strong tape: is the strength broad — plenty of ordinary companies rising, real confidence — or is it narrow, a few safe giants holding the average up while everything else quietly leaks away? This week it was narrow, and narrow strength is the fragile kind. None of that is a reason to bolt for the door. It’s a reason to notice which market you’re actually looking at: the one on the scoreboard, or the one underneath it. The levels and the odds are ours to trade. The shape of the week is yours to keep.

Free weekly letter and explainers like this; the daily research, levels and the Daily Compass are at moatpeak.com.

Educational research only — not investment advice. MoatPeak Group, UAB.

The first force was relief. A tentative framework to wind down the conflict with Iran, brokered out of Islamabad, sent crude tumbling — down roughly a fifth from its highs for the year — as the Strait of Hormuz looked set to reopen. Cheaper energy is the kind of thing that lifts the mood: lower input costs, softer headline inflation, a reason to buy. For a few sessions it did exactly that.

The second force was heavier, and it’s the one that lingers. The Fed held its first meeting under a new chair, and the tone was firmly hawkish: officials nudged up their own rate projections and, tellingly, dropped the soft forward guidance markets had leaned on for years. The message under the message was that this Fed cares more about pinning down inflation than about keeping the market comfortable. (We dug into that shift — a Fed that’s stepped back from telling you its next move — in a recent note.)

Put the two together and you get the week’s real story. Stocks can still rise, but the new Fed has quietly lowered a ceiling over how far valuations can stretch. For years a big chunk of returns came not from companies earning more, but from investors happily paying more for each dollar of those earnings — and that willingness ran on the expectation of ever-cheaper money. A hawkish Fed takes that lever away. From here, gains have to be earned the hard way, through real profit growth, rather than borrowed from a friendlier discount rate. That’s the ceiling. The index can keep climbing under it; it just has to do more of the work itself. (It’s the same gravity we explored in a recent Decoded piece: when rates rise, every dollar of future profit is worth a little less today.)

You could see the new regime in the week’s sharpest moves. One of the world’s largest consulting firms reported a perfectly good quarter and lost nearly a fifth of its value in a day, because the market stopped paying for the story and started pricing the threat underneath it. The dollar pushed to its highest in a year, which quietly tightens conditions everywhere and means a slice of any “record” is really just the measuring stick moving — gold slipped even as stocks held. None of these were one-offs. They’re what a market under a ceiling looks like.

And the cheerful half of the week sits on a shaky leg. The oil relief depends on a truce that, by every account, is preliminary and fragile — a pause, not a settlement. So part of last week’s calm is borrowed, and worth holding loosely. When a market quietly un-prices a live risk, it doesn’t become safer; it just becomes cheaper to be caught out by.

So the useful way to read a week like that isn’t “stocks up, all clear.” It’s that the engine changed. Cheaper money is no longer the wind at your back, the easy gains from rising multiples have a lid on them, and some of the calm is on loan from a fragile truce. None of that is a reason to panic. It’s a reason to value businesses on what they actually earn, and to notice when the index’s strength is the real thing versus the borrowed kind. The week’s specifics — the levels, the odds, and what we’re actually doing about them — are the desk’s work at moatpeak.com. The shape of the week is the part worth carrying into this one.

Free weekly letter and explainers like this; the daily research, levels and the Daily Compass are at moatpeak.com.

Educational research only — not investment advice. MoatPeak Group, UAB.