Amazon's next warehouses will be small, close in, and hard to build

Amazon's same-day delivery network runs through roughly 85 facilities; Project Mercury would take it past 1,000 by 2031, and the number matters less than the map. Business Insider, which reviewed internal planning documents, reported that Amazon wants a network within 10 straight-line miles of 80% of its Prime customers, a demand signal for small, close-in warehouses, the format the current construction pipeline serves least well, and Bisnow carried the findings.

The target Business Insider described, a network inside a 10-mile straight-line radius of 80% of Prime customers, is the loosest version of a coverage metric, ignoring the roads and drive times that decide whether a site actually performs as last mile; the effective ring is tighter than the radius suggests. Amazon's own documents call the program a "transformational initiative to reshape the Same-Day Delivery network," and even on the looser measure, the scarce input is the parcel: entitled, small, and close to customers.

The money attached to the plan is large but explicitly contingent: Business Insider reported that Amazon slated $6.8 billion to increase same-day capacity in 2026 and 2027, inside an effort the outlet described as costing billions and aimed at rivaling Walmart's same-day delivery. Internal financial estimates put the return at $7.1 billion in economic value over the next decade, with positive cash flow by 2030, and the assumption underneath that figure is stated plainly: faster delivery makes customers shop more. Same-day is expected to account for roughly 28% of Amazon orders in 2029 and half its volume by 2035, and grocery, a business generating more than $150 billion a year for the retailer, is among the inventory the plan would place closer to customers.

Strip out the buildings and what remains is a retail thesis: the property demand is levered to whether Prime customers actually order more often, a merchandising bet that real estate sponsors are implicitly taking when they buy land against the program.

The building format is where the real estate math turns; Business Insider previously reported a plan called Orbital, 100,000-square-foot facilities at about $48 million each, sized to fill 75,000 same-day orders a day. That works out to roughly $480 a square foot for the shell, before land, and inside a 10-mile radius the dirt costs more than it does on the exurban acreage where the current pipeline was trained. Developers have spent a decade getting good at the other format, the big box delivered cheaply on land an hour from the population. Mercury's demand lands on the format they are worst at delivering.

Amazon's own recent record argues for reading the program carefully: from 2019 to 2021 the retailer added 453 million square feet of warehouse space, then in 2023 canceled development projects and put nearly 100 facilities up for sublease, a contraction that punished anyone who had underwritten the earlier pace. Greg Boler, founder of the Atlanta industrial developer KMT Partners, told Bisnow that Amazon's movement in either direction — creating demand or pulling back — is a "heavy indicator" for industrial development demand. The current numbers support the upside case: MWPVL reported Amazon is expected to add 51 million square feet to its U.S. warehouse footprint this year, up from 39 million square feet in 2025.

Amazon is not the only bidder for proximity: Walmart has been tying up industrial capacity from the other side of the market: as this publication reported, it took a 16-year lease on an entire Medley warehouse, with extension options that could hold it in place through 2057, and it is building automated projects itself rather than leasing them from speculative developers, as its $1.3 billion Georgia project shows. Two retailers converging on the same finite set of infill parcels is a different demand curve for the land than either one alone.

Amazon's spokesperson told Business Insider the internal Project Mercury projections were preliminary, subject to significant revision, and should not be treated as finalized plans — a caveat worth taping to the front of any site-level underwriting built on the 1,000-facility figure. The coverage does not say which markets Amazon is targeting, whether it would own or lease the buildings, or how many sites are already under contract, gaps that separate a demand signal from a pipeline and are the details a lender would need before financing a building sized to a plan that may not survive its own revisions.

The pattern around Amazon's other infrastructure builds runs through this one in a different key. The Shreveport campus came with a $400 million water system because the binding constraint on that project was physical rather than financial, and same-day delivery poses the same species of problem: the input that cannot be financed or accelerated is entitled land near customers. The sponsors positioned to earn the Mercury trade already control sites inside the ring; the ability to put up square feet quickly is a different skill, and the two are rarely held by the same firm.

That leaves developers with a question narrow enough to underwrite this quarter: at roughly $480 a square foot for the shell plus land on top, does a 100,000-square-foot same-day facility beat what the same parcel earns in another use? The answer will arrive parcel by parcel, well before anyone can say whether Project Mercury reached 1,000 buildings.