Delivering Density: Logistics Geography and Population Clusters

On a 28-acre parcel in Cicero, Illinois — a working-class suburb wedged against Chicago's western edge — Amazon opened a 630,000-square-foot delivery station in 2021 that processes roughly 200,000 packages daily. The facility sits less than nine miles from the Loop, close enough that its drivers can reach 90 percent of the metro area's 9.5 million residents within a two-hour window. That proximity is not incidental. It is the entire logic of the operation, and it is reshaping how both Amazon and Walmart think about real estate, zoning, and the geography of American density.

The Last-Mile Arms Race

Last-mile delivery — the final leg of a shipment's journey from regional hub to front door — accounts for an estimated 53 percent of total shipping costs, according to industry research firm Capgemini. That single statistic has driven Amazon and Walmart into a land acquisition frenzy targeting dense urban cores and their immediate suburbs, where population clusters justify the capital expenditure of small-footprint, high-throughput facilities. The two companies are no longer simply competing on price or assortment. They are competing on geography, and the battleground is the urban fringe.

Amazon's internal infrastructure strategy, partially disclosed in SEC filings and lease agreements surfaced through commercial real estate databases, reveals a deliberate tiering: massive fulfillment centers in exurban locations handle inventory storage, while a growing network of delivery stations — typically 100,000 to 200,000 square feet — are planted inside or immediately adjacent to dense population centers. The Cicero facility represents the upper end of that delivery-station footprint, reflecting Chicago's density premium. Records filed with Cook County show Amazon signed a 15-year lease at the site, committing to a facility that required Cook County zoning variances to accommodate its truck court depths on a constrained urban parcel.

Walmart's counter-strategy has taken a different form. Rather than building standalone logistics infrastructure from scratch, the company has been converting underperforming retail square footage — including former Walmart Supercenters — into fulfillment nodes. A 2022 conversion in Grapevine, Texas, repurposed roughly 40,000 square feet of a 185,000-square-foot store into a dedicated e-commerce fulfillment zone, with the remainder continuing to operate as retail. Walmart's chief supply chain officer, Tom Ward, described the approach publicly as "stores as supply chain assets," a framing that obscures a more aggressive territorial logic: Walmart's 4,700 U.S. stores already sit, on average, within 10 miles of 90 percent of the American population.

Zoning as Competitive Weapon

The collision of logistics demand with urban land use codes has produced a new category of municipal conflict. Industrial zoning — the classification that typically governs warehouses and distribution centers — has historically been pushed to city peripheries, buffered from residential neighborhoods by distance and political inertia. Last-mile facilities, by definition, need to be where people live. That tension is playing out in planning commission hearings from Brooklyn to Burbank.

In New York City, Amazon's 2019 acquisition of a 336,000-square-foot former department store on Jamaica Avenue in Queens — purchased for $\mathbf{978}$ million through a broader portfolio deal — required navigating the city's M1 light manufacturing zoning designation. The facility, which opened in 2020 as a last-mile delivery station, drew objections from local council members who argued the truck traffic was incompatible with the surrounding residential fabric. Community Board 12 formally opposed the project, citing inadequate environmental review. Amazon proceeded under existing zoning rights, no variance required — a legal maneuver that effectively foreclosed community input.

Los Angeles presents a contrasting case. The city's industrial land inventory has been shrinking for two decades as developers convert M-zoned parcels to residential and mixed-use under state housing mandates. The result is a genuine scarcity of last-mile-suitable sites within the city's denser westside neighborhoods. When Amazon sought to establish a delivery station near Culver City in 2021, it ultimately settled for a site in Vernon — a 5.2-square-mile industrial city entirely surrounded by Los Angeles — that placed its drivers further from their delivery zones than the company's internal modeling preferred. Vernon's permissive industrial zoning and low land costs made it financially viable; the geography was a compromise.

Walmart has pursued a more politically durable approach in several markets, partnering with municipalities to frame fulfillment conversions as job retention rather than logistics intrusion. In Fort Worth, Texas, city economic development records show Walmart received a $3.2 million tax abatement in 2022 tied to a fulfillment hub that the company projected would retain 350 jobs and add 200 new positions. The abatement was structured under Texas's Chapter 312 property tax exemption framework, which requires a finding of economic benefit — a standard that, critics note, is rarely denied.

Population Density as Underwriting Criterion

Both companies now treat population density data with the rigor previously reserved for financial underwriting. Amazon's real estate team, according to former employees who have spoken to trade publications including The Information and Supply Chain Dive, uses proprietary delivery-density modeling that scores candidate sites by the number of deliverable addresses within a 90-minute drive time, weighted by current and projected order frequency. Sites that clear a minimum threshold — reportedly around 40,000 deliverable addresses within a defined radius — advance to lease negotiation. Those that don't are rejected regardless of rent economics.

This density-first underwriting has produced a pronounced geographic sorting. Amazon's delivery station network is heavily concentrated in the Northeast Corridor, Greater Los Angeles, the Chicago metro, and the Dallas-Fort Worth Metroplex — the four U.S. markets that combine high population density with high e-commerce penetration. Secondary markets with lower density but strong growth trajectories, including Phoenix and Nashville, have attracted investment at a slower pace, with Amazon typically deploying smaller 80,000-to-100,000-square-foot facilities rather than the full-scale delivery stations that anchor its dense-market strategy.

Walmart's density calculus is structurally different because it starts from an existing physical footprint rather than a greenfield site selection process. The company's internal analytics team, operating under the banner of Walmart GoLocal — its third-party delivery service launched in 2021 — has been mapping delivery density against existing store locations to identify which stores can function as viable last-mile nodes without additional capital investment. GoLocal's pricing, which starts at roughly $7.95 per delivery for third-party clients, is only economically defensible in markets where order density is high enough to allow drivers to complete multiple deliveries per hour. That requirement effectively limits GoLocal's competitive viability to urban and dense suburban markets, a constraint Walmart has not publicly acknowledged.

Infrastructure Friction and the Urban Bargain

The physical infrastructure required to support last-mile density — loading docks, truck courts, van staging areas, charging infrastructure for electric delivery vehicles — creates measurable friction with the urban environments these facilities are designed to serve. In Brooklyn's Sunset Park neighborhood, a 2022 proposal by a Brookfield Asset Management-affiliated entity to develop a 1.2-million-square-foot last-mile logistics complex on the Industry City waterfront drew sustained opposition from local elected officials, including then-Assemblymember Marcela Mitaynes, who argued the project would generate 1,400 additional daily truck trips through residential streets. The project was restructured following political pressure, with the logistics component reduced and a larger share of the square footage reallocated to light manufacturing and creative office uses.

The electric vehicle transition, which both Amazon and Walmart have publicly committed to — Amazon through its 100,000-vehicle order from Rivian, Walmart through a partnership with electric van startup Canoo — does not resolve the spatial conflict. Electric delivery vans still require staging areas, charging infrastructure, and driver facilities. A 100-van Amazon delivery station requires roughly 2.5 to 3 acres of impervious surface for vehicle staging alone, according to site plans filed with municipal planning departments in multiple jurisdictions. That footprint is incompatible with the sub-acre parcels that dominate dense urban cores, which is precisely why facilities like the Cicero station occupy suburban-fringe locations that offer land area unavailable inside city limits.

The deeper tension is fiscal. Cities that attract last-mile facilities gain commercial property tax revenue and, nominally, jobs — though the employment density of a delivery station, typically 200 to 400 workers per shift, is low relative to the square footage and traffic generation. Cities that lose industrial land to last-mile conversion sacrifice the optionality of future manufacturing or mixed-use development for a logistics use that may become obsolete as autonomous delivery technology matures. Several planning directors in major metros, speaking on background to Cornice, described the same dilemma: the short-term revenue is real, the long-term land-use cost is speculative, and the political pressure to approve is intense.

What neither Amazon nor Walmart has resolved is the fundamental incompatibility between last-mile logistics economics and the texture of dense urban neighborhoods. Drone delivery, autonomous sidewalk robots, and micro-fulfillment embedded in residential buildings have all been piloted — Amazon's Scout program in Snohomish County, Washington; Walmart's drone delivery tests in Frisco, Texas — but none has scaled to a point where it displaces the van-and-driver model that still accounts for the overwhelming majority of residential deliveries. Until that changes, the competition for last-mile supremacy will continue to be decided not by algorithm or brand loyalty, but by who controls the right parcel of land, in the right zoning district, at the right distance from the right concentration of people willing to pay for next-day delivery.