Case Study #2 · Amazon, Inc. · August 2026

Last-Mile Delivery Concentration & Network Resilience

Did Amazon successfully reduce third-party carrier dependency between 2019 and 2024 — or did it simply shift concentration from UPS/USPS to its own DSP network, trading one form of risk for another?

Last-Mile Delivery DSP Network Carrier Concentration Network Resilience FY2019–FY2024 11 Recommendations
Company
Amazon, Inc.
Time Period
FY2019–FY2024
Focus Area
Last-Mile Logistics
Data Sources
SEC 10-K, ShipMatrix
Phases
5 of 5 complete
Phase 1

Company Selection & Scoping

Amazon is the most consequential logistics story of the past decade. Between 2018 and 2024, the company transformed from the largest customer of UPS and FedEx into their most formidable competitor — building a proprietary last-mile delivery network that now handles 28.2% of all US parcels by volume.

Did Amazon meaningfully reduce last-mile carrier concentration between 2019 and 2024, or did it simply shift dependency from UPS/USPS to a new concentrated network of DSPs — trading one form of concentration risk for another?

Why this question matters

Scope

DimensionIn ScopeOut of Scope
GeographyUS domestic last-mile onlyInternational logistics
Time PeriodFY2019–FY2024 (6 years)Pre-2019 historical
SegmentLast-mile (final delivery to customer)Middle-mile, air cargo, freight
Data SourcesPublic 10-K, earnings calls, press releasesInternal Amazon data
BenchmarksUPS, FedEx, USPSNon-US carriers
Phase 2

Data Collection & Research

All data sourced from Amazon SEC 10-K filings (FY2019–FY2024), UPS and FedEx annual reports, USPS filings, Amazon press releases, and verified third-party logistics research including ShipMatrix and the Pitney Bowes Parcel Shipping Index.

Income Statement ($B)

MetricFY2019FY2020FY2021FY2022FY2023FY2024
Total Revenue$280.5$386.1$469.8$513.9$574.8$638.0
Gross Margin %41.0%39.6%42.0%43.8%47.0%48.9%
Shipping Costs ($B)$37.9$61.1$76.7$83.5$89.5$95.9
Shipping as % Revenue13.5%15.8%16.3%16.2%15.6%15.0%
Fulfillment Costs ($B)$40.2$58.5$75.1$84.3$90.6$97.0
Operating Income ($B)$14.5$22.9$24.9-$2.4$36.9$68.6

US Parcel Market Share by Volume (Billions of Packages)

Carrier201920202021202220232024
Amazon Logistics1.7B2.5B3.5B4.8B5.9B6.7B
USPS7.3B7.3B7.6B7.2B6.6B6.9B
UPS5.1B5.3B5.4B5.2B4.6B4.4B
FedEx4.2B4.3B4.1B4.1B3.9B3.6B
Amazon Share %9.1%12.6%16.6%21.9%27.2%28.2%

DSP Network Growth

YearDSP PartnersDriversPackages/DayKey Event
2018~180~5,000~1MProgram launched
2019~800~25,000~4MRapid expansion phase
20201,30085,000~7MCOVID surge; $1B+ in DSP bonuses paid
20212,500150,000~10MDSP surpasses FedEx in volume
20223,000275,000~15MRivian EVs deployed; DSP surpasses UPS in volume
20233,500279,000~20MAmazon overtakes USPS projections accelerated
20244,500390,000~22MAmazon becomes #1 US carrier by volume
Phase 3

Quantitative Analysis & KPI Dashboard

Amazon Logistics Market Share (2024)
28.2%
Up from 9.1% in 2019
Shipping Cost per Package (2024)
$14.30
Down from $22.30 in 2019 (−36%)
Total Shipping Costs FY2024
$95.9B
+153% since FY2019
DSP Driver Count (2024)
390K
Up from 25K in 2019 (+15x)
Network CAGR (2019–2024)
31.5%
vs UPS −2.9%, FedEx −3.0%
UPS Revenue from Amazon (est.)
−48%
$11.2B (2021) → $5.8B (2024)
Chart 01
US Parcel Market Share by Volume (2019–2024)
Amazon's market share gain vs declining incumbents
Chart 02
Shipping Cost per Package (FY2019–FY2024)
36% reduction as internal network scales
Chart 03
DSP Network Growth (2018–2024)
Driver count and package volume exponential growth
Chart 04
Shipping Costs as % of Revenue
Still above FY2019 baseline despite scale gains

Peer Benchmark (2024)

MetricAmazonUPSFedExUSPS
Annual Package Volume6.7B4.4B3.6B6.9B
US Market Share (volume)28.2%18.5%15.1%29.0%
Revenue per Package (est.)$14.30$20.70$17.60$11.50
YoY Volume Growth+7.3%−8.6%−7.7%+4.5%
Network CAGR (2019–2024)+31.5%−2.9%−3.0%−1.1%
Operating Margin10.8%5.8%3.3%N/A
Phase 4

Process Mapping & Structural Diagnosis

Phase 3 established what happened — Amazon's volume grew dramatically and UPS/USPS dependency fell. Phase 4 maps why the network evolved as it did, and where the structural failure points are that create new forms of concentration risk.

Three Structural Failure Points

Failure Point 1 — Vulnerability Lens
The DSP Dependency Paradox
Amazon built its last-mile network to reduce dependency on UPS and FedEx. But in doing so, it created a new form of concentration risk: 4,500+ DSPs that are 100% Amazon-dependent for revenue, financially fragile (average startup capital $10K–$30K), and geographically clustered around existing FC infrastructure. A DSP failure in a high-density metro area can remove last-mile capacity from thousands of daily deliveries with no redundancy.
Failure Point 2 — Vulnerability + Scale Lens
Geographic FC Concentration
Amazon's fulfillment center network is concentrated in a small number of metro areas — Los Angeles, Dallas, Chicago, New York, and Seattle account for a disproportionate share of daily package volume. Unlike UPS and FedEx, which have built multi-decade hub-and-spoke redundancy, Amazon's network is optimised for speed, not resilience. A single severe weather event or infrastructure failure cascades directly into last-mile disruption.
Failure Point 3 — Inefficiency + Scale Lens
The Performance Scoring Feedback Loop
Amazon scores DSPs on delivery success rate, safety, and customer feedback. High-scoring DSPs receive more routes; low-scoring face termination. This creates the same self-reinforcing feedback loop identified in the Nike case study: high performance → more volume → more capacity investment → higher performance. The best-performing DSPs become larger, more geographically concentrated, and more essential — amplifying single-point-of-failure risk.
Core Diagnosis
Amazon didn't diversify its last-mile risk. It internalised it.
The nature of the concentration has changed — from two large public carriers to 4,500 small private DSPs — but the structural risk has not been reduced. In some dimensions (financial resilience of partners, geographic redundancy, transparency) the new structure is significantly worse than the UPS/FedEx model it replaced.

The Concentration Shift

DimensionPre-2018 (UPS/FedEx model)2024 (DSP model)Risk Change
Carrier concentration2 carriers dominate4,500 DSPs — all Amazon-dependentShifted, not reduced
Financial resilienceInvestment-grade public companiesAvg $2–5M revenue, no safety netSignificantly worse
Geographic redundancyMulti-decade hub networksOptimised for speed, not resilienceWorse in disruptions
TransparencyAudited public companiesPrivate, no public disclosureSignificantly worse
Labour riskUnionised UPS drivers (Teamsters)No union, high turnoverDifferent risk profile
Phase 5

Solutions & Recommendations

Amazon's last-mile transformation is genuinely impressive — a 31.5% CAGR in package volume, a 36% reduction in cost per package, and the #1 position in US parcel market share by 2024. But the structural analysis reveals that the diversification narrative obscures a new form of concentration risk. Three pillars of intervention are required.

Pillar A — DSP Network Resilience

#RecommendationMechanismTimelinePriority
A1DSP Minimum Revenue Guarantee
100% Amazon-dependent DSPs are financially fragile — route reassignment can trigger insolvency within weeks
Guarantee minimum 80% of prior year volume to DSPs with 3+ years tenure. Build into DSP contracts.6–18 monthsCritical
A2DSP Financial Resilience Scoring
Amazon has no visibility into DSP financial health before capacity risk materialises
Require quarterly financial statements. Score on liquidity, leverage, revenue concentration. Flag DSPs with <3 months cash runway.6–12 monthsCritical
A3Third-Party Carrier Redundancy Floor
Amazon has reduced UPS/FedEx dependency so aggressively it now lacks backup capacity
Maintain minimum 10% of volume through UPS or FedEx in every metro area to ensure backup during DSP failures.12–24 monthsHigh
A4DSP Consolidation Cap
Top DSPs are growing larger, creating hidden concentration within the DSP network itself
Cap any single DSP at <2% of metro area volume. Actively develop new DSP entrants to prevent oligopoly.18–36 monthsHigh

Pillar B — Geographic Network Resilience

#RecommendationMechanismTimelinePriority
B1Geo-Redundancy Stress Testing
No public framework exists for testing last-mile resilience by metro area
Model 30% capacity loss for top 20 US metros. Identify metros with <48hr backup capacity and prioritise investment.6–12 monthsCritical
B2Strategic Inventory Pre-Positioning
Same-day/next-day SLAs leave no buffer when FC capacity is lost
Pre-position 7-day inventory buffers for top-100 SKUs in secondary FC locations within 50 miles of primary hubs.12–18 monthsHigh
B3Rural Last-Mile Diversification
DSP network is urban-heavy; rural delivery relies on financially stressed USPS
Develop rural DSP program with modified economics (lower volume requirements, higher per-package rates) to build rural coverage.24–48 monthsMedium

Pillar C — Transparency & Systemic Risk Governance

#RecommendationMechanismTimelinePriority
C1Annual DSP Network Health Report
Amazon discloses nothing about DSP financial health, concentration, or labour turnover
Publish annual DSP metrics: active partners, geographic distribution, average revenue, driver retention, safety incidents. Align with GRI 308.6–12 monthsHigh
C2DSP Labour Standards & Living Wage Floor
DSP driver turnover exceeds 100% annually in some markets due to below-market wages
Set minimum DSP driver compensation ($22/hour minimum nationally). Tie DSP contract renewal to compliance.12–24 monthsHigh
C3Real-Time Last-Mile Risk Dashboard
No real-time monitoring system for last-mile disruption risk by geography
Deploy dashboard tracking DSP capacity by metro, weather exposure, driver availability, UPS/USPS backup capacity. Update in real time.12–18 monthsHigh
C4Force Majeure Escalation Protocol
No published disruption response framework — response to DSP failures is reactive
Publish tiered Force Majeure protocol: Level 1 (single DSP failure), Level 2 (metro-area loss), Level 3 (multi-region event). Test annually.12–24 monthsHigh

Risk Matrix

#RiskLikelihoodImpactScoreRatingMitigation
R1DSP mass insolvency in major metro area4520CriticalA1, A2, A3
R2Extreme weather knocking out primary FC hub3515CriticalB1, B2
R3DSP driver strikes across multiple markets4416CriticalA3, C2
R4DSP feedback loop creating hidden concentration5315HighA4, C1
R5USPS capacity degradation eliminating rural backup3412HighA3, B3
R6Regulatory intervention in DSP labour classification3412HighC2, C4
R7FC geographic concentration amplifying single events339HighB1, B2
R8Competitor logistics networks growing236MediumB3

Investment Summary

CategoryLow EstimateHigh EstimatePrimary Recs
DSP Stability Programs$200M$400MA1, A2, A4
Geographic Redundancy$500M$1,200MB1, B2
Rural DSP Development$300M$600MB3
Governance & Systems$50M$120MC1–C4
TOTAL$1.05B$2.32B4-year horizon, FY2025–FY2028
Linkwell Verdict
The numbers are real. The narrative is incomplete.
Amazon's last-mile transformation delivered genuine scale efficiency — 31.5% CAGR, 36% cost-per-package reduction, and #1 US parcel market share. But the structural analysis reveals a critical gap: Amazon did not eliminate carrier concentration, it internalised it. The recommended investment of $1.05–2.32B over four years is the price of building a last-mile network that is not just fast, but structurally resilient. Break-even point: one avoided major metro disruption (~$2–5B revenue at risk per week).