← Cross-cutsCross-cut · On-demand delivery

The 1%-margin logistics body

Four Teardown studies cover on-demand delivery — Uber, DoorDash, Meituan and Instacart. Read together, they share a structure that took a decade and tens of billions in losses to prove out: the delivery itself barely makes money, and the real profit is a high-margin advertising business riding on top of the logistics. The studies also share the same three threats to that hard-won model.

The delivery doesn’t pay; the ads do

The defining number across this cluster is how thin the core logistics margin is. DoorDash’s contribution profit is just 5.1% of order value; Uber’s delivery take rate is ~19%; the actual moving-of-food earns almost nothing. What turned these from cash-incinerators into profitable businesses is the ad layer: Instacart earns its profit from a $1.07B advertising line at ~80% margin bolted onto a thin transaction take, DoorDash has crossed a $1B ad run-rate, and Meituan’s highest-margin line is merchant advertising. The cross-cut’s core lesson mirrors the marketplaces one: delivery is the customer-acquisition body, and advertising is the organ that actually pays — which is why the platforms with scale (DoorDash, Uber, Instacart) reached profitability and the one fighting a subsidy war (Meituan) fell back into loss.

Density is the moat — and it’s local

What protects each platform is delivery density: enough orders in a geography to keep drivers busy and economics positive. That moat is real but stubbornly local, which is why the cluster is a set of regional champions rather than one global winner — DoorDash owns ~60.7% of US food delivery, Meituan ~55–58% of China, Uber spans 70+ countries on mobility, Instacart owns US grocery. The studies show density cuts both ways: it makes the leader hard to dislodge at home (DoorDash’s US lead, Meituan’s rider network), but it makes expansion expensive and unproven — DoorDash’s ~$12B international bet (Wolt, Deliveroo) is the contested question its own study flags, because density doesn’t travel.

Three shared threats to the hard-won profit

Having finally reached profit, all four face attacks on it — and they sort the cluster. Autonomy threatens Uber most directly: AV owners running their own apps could skip the marketplace and compress its ~30% mobility take. A subsidy price war is the live danger for Meituan, where JD and Alibaba poured a combined >RMB 100B into the core and pushed it back into loss — the clearest proof in the cluster that this profit must be continually re-defended. And Big Retail’s first-party scale squeezes Instacart from above (Walmart, Amazon) while DoorDash and Uber squeeze it from the side. Same business model, three different incoming threats — and each study’s central question is essentially “does the profit survive my specific attacker?”

The valuation tell: priced like software, built like logistics

The sharpest disagreement the cross-section exposes is valuation. DoorDash trades at ~67× earnings and Instacart is priced off its ad platform — multiples that assume these are software businesses. But their contribution economics (~1–5% of order value) are those of a logistics operator. Uber, the most profitable and diversified, trades at only ~15× FCF — the market rewarding proven cash over growth narrative. The studies split precisely on whether the ad-and-data layer justifies a software multiple on a thin-margin physical-delivery body, or whether the physical economics eventually reassert themselves. It is the same question investors got wrong for a decade by funding the logistics; the bet now is on the ads.

Where they agree — and where they split

All four agree the model finally works at scale, that advertising is the profit engine, and that local density is the moat. They split on what that profit is worth and how defensible it is. The bull reading: these are durable, ad-supported local-commerce platforms with proven GAAP profit (Uber, DoorDash, Instacart). The bear reading: a price war can erase it overnight (Meituan), an AV shift can disintermediate it (Uber), or Big Retail can out-scale it (Instacart) — and a ~67× multiple leaves no room for any of those. The demand for delivery isn’t the question anymore. Whether the ad layer is a durable moat or a thin skim on a commoditizing logistics business is.

The cluster at a glance

CompanyScaleModel & takeProfitabilityBiggest threat
UberNYSE:UBER$193.5B bookings · 13.6B trips~30% mobility / ~19% delivery takeGAAP-profitable · $9.8B FCFRobotaxi / AV disintermediation
DoorDashNASDAQ: DASH$103.0B GOV · ~60.7% US food~13% take · 5.1% of GOV contribution$935M GAAP NI · $1B+ ads~$12B Wolt/Deliveroo intl bet · labor
MeituanHKEX:3690RMB 364.9B rev · ~55–58% shareDelivery toll + merchant adsFY net loss RMB 23.4B (price war)JD/Alibaba >RMB 100B subsidy war
InstacartNASDAQ: CART$37.2B GTV · 1,800 banners~7.2% take · $1.07B ~80%-margin ads$447M GAAP NI · 29% EBITDA marginWalmart/Amazon + DoorDash/Uber squeeze

Figures as of each study’s stated date (2026-06); GOV/GMV/bookings are on different bases and not strictly comparable. See each teardown for sourcing and the full weighing.

The four studies — and the question each turns on

Uber Technologies, Inc.NYSE:UBERDoes the arrival of autonomous vehicles make Uber the indispensable demand aggregator or an avoidable middleman?Read the full weighing →DoorDash, Inc.NASDAQ: DASHDoes a thin-margin (~1%-of-GOV) delivery logistics business that trades at a software-like ~67x multiple have a durable moat and durable profits?Read the full weighing →Meituan (美团)HKEX:3690Is Meituan's local-services dominance a durable, profitable moat — or a position that must be re-purchased whenever a deep-pocketed rival decides to spend?Read the full weighing →Instacart (Maplebear Inc.)NASDAQ: CARTIs Instacart a defensible retail-media/software platform, or a fragile ad business strapped to a thin, contested delivery marketplace that AI could disintermediate?Read the full weighing →

This is the kind of reading the Desk does for you

A cross-cut takes four teardowns and one business model and asks where the money really is. The Desk does the same for the companies you actually own — your thesis, the rivals that move it, and the tripwires that would change your mind.

See the Desk →