Higher fuel costs. Two different retail tests: Amazon and Ross
By @engtmk · Research as of 11 September 2026
Higher fuel prices can reach retailers through two routes: the cost of moving goods and the amount customers have left to spend. Amazon and Ross Stores face both, but their business models create different opportunities to absorb the pressure.
My working view: Ross offers the clearer test of consumers switching toward lower-priced merchandise; Amazon has more diverse sources of operating profit. Neither advantage establishes which stock is the better investment at today’s price. Ross still needs profitable transactions, not just more visitors. Amazon still needs its delivery network to earn an adequate return, even when other businesses perform well.
August’s inflation release provides the reason to revisit those assumptions. It does not prove that either company’s margins have deteriorated. The most recent reported quarters largely preceded the month’s fuel increase, and some impressive earnings figures contain benefits that should not be projected forward.
The research question is therefore specific: if fuel pressure persists, can each company protect recurring profit without relying on customers accepting higher prices or unusual accounting gains? My posture is to monitor that evidence and normalized valuation before treating either name as protection against a tougher consumer environment.
Cover: AI-generated conceptual illustration of delivery infrastructure, fuel and a clothing store. It does not depict actual Amazon or Ross facilities.
The inflation report: energy explains the divergence
The August US consumer-price report, released on 11 September, showed the following:
| Measure | August change |
|---|---|
| Headline CPI | +0.4% month over month; +3.4% year over year |
| Core CPI, excluding food and energy | +0.3% monthly; +2.4% annually, down from July’s 2.5% |
| Gasoline | +3.9% monthly |
| Gasoline’s contribution | More than one-third of the monthly headline increase |
Monthly changes are seasonally adjusted; annual changes are unadjusted. Source: Bureau of Labor Statistics, August 2026 CPI release, accessed 11 September 2026. This BLS link rolls forward as new releases arrive.
The report describes both a faster monthly increase and a lower annual core rate. Those observations are compatible: a twelve-month comparison reflects the month dropping out as well as the month entering. Lower annual inflation also means slower price growth, not a return to earlier price levels.
For this article, the new evidence is the August outcome and its gasoline contribution. I am not claiming a measured earnings surprise or a specific stock-price response. Establishing either would require a dated expectations baseline and market data.
How the pressure reaches each business
| Channel | Amazon | Ross Stores |
|---|---|---|
| Moving goods | Persistent transport-cost increases can raise delivery and fulfilment expense | Freight and distribution costs can pressure merchandise economics |
| Household budgets | Less discretionary purchasing, different product mix or smaller orders | Customers may seek cheaper alternatives, but also reduce purchases |
| Operating response | Network productivity, order consolidation, mix and pricing | Buying discipline, assortment, inventory control and markdowns |
| Investor test | Recurring geographic-segment profit, with AWS assessed separately | Comparable sales and recurring margin after unusual benefits |
These are analytical pathways, not quantified company sensitivities. Their size depends on contracts, customer behaviour and the companies’ responses.
Consumer gasoline CPI is a warning signal for household budgets. It is not Amazon’s fuel bill, a diesel index, or a forecast of either company’s freight expense. Transportation uses different fuels and contract structures, and a carrier’s surcharge can reach a retailer with a delay.
That distinction prevents false precision. Applying the 3.9% consumer gasoline increase to all shipping costs would produce a misleading earnings estimate.
Illustration from the original post. The two labels identify the emphasis of this comparison; both companies face cost and demand risks.
Amazon: look beneath consolidated earnings
Amazon’s latest release provides a useful starting point:
| Segment | Q2 2026 sales | Operating income | Calculated operating margin |
|---|---|---|---|
| North America | $116.177 billion | $9.123 billion | 7.9% |
| International | $42.197 billion | $1.717 billion | 4.1% |
| AWS | $42.232 billion | $16.621 billion | 39.4% |
The quarter ended 30 June; results were released 30 July. Margins above are operating income divided by sales, rounded. AWS generated approximately 60.5% of the three segments’ combined operating income. Amazon also reported $53.4 billion of pre-tax non-operating other income, primarily from Anthropic investments. Amazon Q2 results.
That earnings mix changes the interpretation of “retail resilience.” Strong consolidated profit can coexist with pressure on commerce. The North America and International segments include services as well as merchandise activity; their margins are not pure delivery or retail merchandise margins.
My analysis would start with those geographic segments, while recognizing that published segment data cannot isolate a fuel effect. AWS supplies a separate earnings engine, but it is not a guaranteed hedge against weak retail demand and has its own competition and investment risks.
The operational argument for Amazon is that scale can create options. Better inventory placement, fuller delivery routes and fewer split shipments could offset some transport inflation. Whether they actually do so must be demonstrated in results; these are potential mechanisms, not savings estimates.
The opposing argument is that speed and convenience are expensive promises to maintain when orders become smaller or routes less productive. Raising a fee may recover part of a cost increase, but it can also change demand. Cutting a service level may preserve expense while weakening the customer proposition.
Product mix adds another complication. Revenue supported by lower-value essentials does not necessarily produce the same contribution profit as discretionary purchases. I would ask what happens to profit after the costs needed to fulfil the order, not whether sales alone keep growing.
Ross: more customers are useful only if transactions remain profitable
Ross’s quarter ended 1 August, so it cannot establish resilience to the full August fuel increase. In its 20 August release, Ross reported sales growth of 13% and comparable-store growth of 10%, primarily traffic-driven. Diluted EPS of $2.66 included approximately $0.60 from tariff refunds. Of the reported 610-basis-point operating-margin improvement, 405 basis points came from those refunds; improvement excluding that benefit was 205 basis points. Ross fiscal Q2 results.
There was operating progress beyond the refund. For the earnings-quality framework covering both companies, see Why headline profit is often the wrong number.
The off-price thesis is plausible: when shoppers become more price-conscious, an attractive assortment at a lower ticket can win a purchase that might otherwise go elsewhere. Yet switching stores and spending less can happen simultaneously. A customer may visit more often, buy fewer items, choose a cheaper product or leave without purchasing.
A simple hypothetical shows the test. If completed transactions increase 5% but average transaction value falls 4%, sales increase only 1.05 × 0.96 − 1 = 0.8%. If average transaction value falls 6%, sales decline 1.05 × 0.94 − 1 = −1.3%. These are illustrations, not Ross forecasts. Visitor traffic is not the same as completed transactions: conversion matters too.
Revenue is only one hurdle. The additional sales must carry enough margin to cover buying, freight, store labour and occupancy. A retailer can attract customers through discounts and still earn less if markdowns or operating costs absorb the benefit.
Buying conditions can also cut both ways. Disruption elsewhere may create attractive merchandise opportunities; it may also leave the wrong inventory available, or raise the cost of getting it to the right store. I would not assume that a stressed retail environment automatically improves off-price purchasing economics.
A margin bridge is more useful than a fuel-price forecast
For either company, I would frame recurring operating profit as:
Sales × operating margin, with the margin reflecting merchandise economics, transport, labour, occupancy, mix and efficiency.
An illustrative retailer with $100 of sales and an 8% operating margin earns $8. If sales rise 2% but margin falls to 7%, operating profit becomes $7.14—a decline of 10.75%. This is a generic sensitivity, not a model of Amazon or Ross.
That arithmetic explains why seemingly modest margin changes deserve attention. It also shows why assigning an EPS impact to a fuel move requires more than a headline. We would need the exposed cost base, procurement terms, pass-through, volume response, tax treatment and offsetting productivity.
Until those inputs are established, a precise fuel-to-EPS estimate would disguise uncertainty rather than resolve it.
The Fed creates a separate valuation test
The Federal Reserve’s calendar schedules the next FOMC meeting for 15–16 September, with economic projections. Federal Reserve meeting calendar.
The Fed’s longer-run inflation objective is measured using PCE inflation, not CPI, and policy considers employment alongside price stability. Federal Reserve explanation of its inflation objective. CPI can inform that discussion without determining the decision.
My interpretation is conditional. A temporary energy increase that fades without broader persistence creates a different policy problem from sustained pressure that feeds into other prices or expectations. Weaker household spending can also complicate the response.
For stocks, that creates two distinct risks: an earnings revision if consumers or costs disappoint, and a valuation change if investors demand a higher return. They can occur together, but neither follows mechanically from a single CPI release.
I am not assigning meeting probabilities, forecasting a policy move or claiming current bond-market pricing here. Those conclusions require a separately verified market dataset.
Valuation: normalize the earnings before comparing the shares
The tempting shortcut is to call Ross defensive and Amazon diversified, then infer that either deserves a premium. The more useful question is how much resilience the share price already assumes.
For Ross, I would start with earnings excluding the approximate $0.60 refund benefit rather than treating it as a recurring source of profit. Simple subtraction gives about $2.06 for the quarter; that is an illustrative exclusion of the disclosed benefit, not a complete adjusted-earnings model or an annual run rate. For Amazon, I would separate operating performance from the investment gains discussed above and assess the reinvestment required to sustain growth.
The comparison needs different valuation approaches. Amazon’s commerce and AWS operations have different growth, margin and capital needs. Ross requires assumptions about comparable sales, new stores, merchandise margins and ongoing investment. A single headline P/E can obscure those differences.
As a general illustration, earnings growth of 10% alongside a valuation-multiple decline of 15% implies a share-price change of 1.10 × 0.85 − 1 = −6.5%, before dividends. A business can defend profit and still disappoint investors who paid for a stronger outcome.
This article does not contain verified contemporaneous share prices or consensus estimates, so it does not establish which stock is cheaper or offer a price target. The next valuation step is to compare a dated price with recurring earnings and cash flow under a conservative consumer scenario.
Three scenarios worth monitoring
These are conditional scenarios, not assigned probabilities.
| Scenario | Amazon test | Ross test |
|---|---|---|
| Fuel pressure fades | Do productivity gains remain visible once the cost headwind eases? | Does customer momentum persist as budget pressure moderates? |
| Fuel stays high; demand holds | Can service quality and efficiency protect geographic-segment profit? | Can transactions and merchandise margin offset freight and basket pressure? |
| Fuel stays high; households retrench | Do smaller orders and weaker discretionary mix erode operating leverage? | Does switching toward cheaper goods outweigh reduced purchases and markdown risk? |
The strongest counterargument to the bearish consumer case is adaptation. Households can change travel, purchasing patterns or savings rather than reduce every retail category equally. Companies can adjust assortment, procurement and operations. The CPI release does not specify how much adjustment will occur or who captures the resulting spending.
That is why the comparison should stay tied to company evidence.
What would strengthen or weaken the view?
| Monitor | Evidence of resilience | Evidence requiring reassessment |
|---|---|---|
| Amazon geographic segments | Recurring profit holds as sales and delivery activity evolve | Profit weakens despite sales growth |
| Amazon group earnings quality | Operating results support the investment case independently of gains | Consolidated EPS obscures deteriorating operating performance |
| Ross customer demand | Comparable sales supported by profitable transactions | Visits fail to convert, or baskets shrink enough to overwhelm transaction growth |
| Ross recurring margin | Progress persists after the refund benefit rolls out | Markdown or operating-cost pressure reverses underlying improvement |
| Inventory and cash conversion | Stock and spending remain aligned with demand | Inventory or working capital absorbs cash while sales slow |
| Inflation and policy | Energy pressure fades without wider persistence | Broader price pressure or changing rate expectations challenge valuation assumptions |
I would use the next company reporting cycles to update the operating thesis and the September Fed meeting to update the policy context. One release should not be expected to resolve both.
For a new investment, the posture is watchlist pending normalized valuation. For existing exposure, the task is to re-underwrite the relevant earnings drivers rather than trade on the fuel headline alone. This article does not prescribe position sizes or assume current portfolio weights.
My conclusion
Ross has the clearer commercial proposition for shoppers seeking lower-priced merchandise. Amazon has more diverse operating profit, including a large contribution from AWS. Those are different strengths, not interchangeable forms of consumer protection.
My preference is to demand proof of recurring profit. At Ross, that means transactions and margin beyond the tariff-refund benefit. At Amazon, it means understanding commerce performance beneath consolidated earnings and investment gains.
The August report sharpens those tests. It does not settle them.
A tougher household budget can redirect spending, reduce spending and raise delivery costs at the same time. The better investment will be the business whose recurring cash returns hold up better than its valuation requires.
Related research: why higher oil prices can also pressure technology valuations.
Sources and freshness
As of: 11 September 2026, after the August CPI release and before the September FOMC meeting.
Sources used: BLS CPI, the Federal Reserve’s calendar and inflation-objective explanation, Amazon’s 30 July earnings release and Ross’s 20 August earnings release, linked near the relevant facts.
Evidence posture: The inflation figures and reported company metrics are supported by primary sources. Mechanisms and comparative judgments are analysis. Numerical sensitivities are explicitly hypothetical; calculated margins and the Ross refund subtraction are identified as calculations.
Scope limitations: The company results are historical baselines, not measurements of August’s full fuel impact. Current market prices, consensus forecasts, fuel-contract sensitivities and customer-level transaction data are not established here. No fair-value ranking, policy probability or fuel-specific earnings forecast is claimed.
My views, not personalized investment advice. Investments involve risk, including loss of capital. For the current public portfolio record, see @engtmk on eToro.