Wendover’s Drone-Delivery Autopsy Got the Phoenix Math Wrong—and Missed the Business That Followed
Wendover Productions published its investigation into drone delivery in February 2022. Its central argument was persuasive: the grand promise of autonomous aircraft delivering anything to anyone had collided with restricted airspace, difficult landing conditions, consumer price sensitivity and competition from conventional delivery companies.
But the video makes two important mistakes. It uses a questionable Phoenix market calculation to quantify the industry's limitations, and it treats a single delivery base as the natural model for an industry that has since developed around networks of existing retail stores.
By October 2026, Walmart reported one million drone deliveries. Amazon was operating drone delivery in Phoenix. Zipline had completed more than 2.5 million commercial deliveries worldwide.
The original vision has not arrived. But neither has the industry remained where the video left it.
The Phoenix calculation does not establish the market size
The video's most concrete market analysis concerns Phoenix, Arizona.
It identifies a potential operating zone containing 310,987 residents. It then applies two city-wide housing statistics: 63.2% single-family housing and 92% occupancy, arriving at approximately 180,820 possible customers.
The arithmetic is straightforward:
310,987 × 63.2% × 92% = 180,820.
The problem is the denominators.
The 63.2% figure describes housing units, not residents. The 92% figure also describes occupied housing units. The Census Bureau's American Community Survey explicitly separates these measurements into tables for housing structure and occupancy status.
Applying both housing-unit percentages directly to a population count does not establish how many residents occupy suitable delivery addresses. The calculation also applies city-wide housing characteristics to a specially selected service zone, without demonstrating that the zone has the same characteristics.
There is a further problem: the population figure already counts people, not vacant homes. Applying an occupancy percentage to it effectively treats vacant housing as though it were a proportion of the population.
Wendover acknowledges that the methodology is imprecise. That qualification is appropriate, but it does not turn 180,820 into a defensible estimate of potential customers.
The figure does not establish how many households have suitable delivery locations, how many residents would order, how many addresses are legally serviceable or how frequently customers would use the service.
It is an illustrative calculation, not a market study. The distinction matters because the video uses the resulting number to reinforce its conclusion that drone delivery is necessarily restricted to a small and fortunate minority.
Census.gov+2
The single-base assumption became a limitation of the analysis
The video assumes an early delivery service would operate from one central location, with drones returning to recharge after each delivery.
It then uses an approximate six-mile delivery range to construct a viable Phoenix operating zone.
That is a reasonable hypothetical for an early operation. It is not the only way to build a commercial network.
Consider what happened with Walmart and Wing.
Rather than relying on one enormous distribution center to serve an entire metropolitan area, the companies began establishing drone launch facilities at individual retail locations. Existing stores provide inventory, staff and a place to launch aircraft.
Customers order from participating stores. Employees prepare packages, and drones fly directly to eligible addresses.
The FAA's environmental reviews describe Wing's proposed operations involving up to 43 nest locations in Atlanta and 75 in the Dallas-Fort Worth area. These are distributed networks, not one central delivery base.
Federal Aviation Administration+1
This changes the geographic calculation.
Instead of asking how many people can be reached from one charging station, the relevant question becomes how many viable delivery locations can be established across a metropolitan area.
The regulatory restrictions remain. So do range, weather and safety limitations. But adding appropriately located launch facilities changes the potential coverage considerably.
Wendover's Phoenix model measures one possible network design and then uses its limitations to describe the broader opportunity.
That is an unnecessarily narrow way to measure the industry.
Phoenix itself became a commercial drone-delivery market
The choice of Phoenix makes the subsequent developments particularly revealing.
Amazon established Prime Air operations in Tolleson, within the Phoenix metropolitan area.
The FAA's environmental review describes an MK30 operating range of 7.5 miles and a potential operating area of approximately 174 square miles. The approved environmental plan allows up to 469 delivery flights per operating day, or approximately 171,329 annually.
Those figures are authorized operating limits, not evidence that Amazon actually completes that many deliveries. But they demonstrate that a commercial operating area was achievable despite the airport restrictions Wendover highlighted.
Federal Aviation Administration+1
By August 2026, the Los Angeles Times reported that Amazon was again offering Prime Air delivery in Phoenix, with some deliveries completed in under 15 minutes.
The reported customer pricing was also revealing:
- Prime members: free on orders over $50.
- Prime members: $2.99 on orders under $50.
- Non-members: $4.99.
These prices do not reveal Amazon's actual operating costs, but they demonstrate that the service is being offered as a practical retail option rather than merely an experimental demonstration.
Los Angeles Times
Phoenix has not become a city where every home can receive a drone delivery. Eligible addresses, products, airspace and operating conditions still determine availability.
That is precisely the distinction the original analysis needed to make: a city can be unsuitable for universal drone delivery while remaining commercially useful for a network of carefully selected delivery zones.
One million deliveries changes the argument
Walmart's May 29, 2026 announcement provides particularly useful evidence.
The company reported:
- More than 1 million completed drone deliveries.
- Operations across 66 stores.
- Coverage in four states and five metropolitan markets.
- An average delivery time of 23 minutes.
- Approximately 40% of all one million deliveries completed during FY27's first quarter.
That last number deserves attention.
If 40% of the cumulative total occurred in one quarter, approximately 400,000 deliveries happened during that quarter alone.
This is not a handful of demonstration flights. It is a functioning commercial service with measurable customer activity.
Walmart Newsroom+2
It still does not demonstrate mass-market adoption. Sixty-six participating stores represent a limited footprint, and the service is restricted by eligible addresses and product requirements.
But describing the industry as barely beyond proof of concept no longer accurately describes its leading commercial operations.
The more important development is that drone delivery has found a role alongside conventional retail delivery rather than attempting to replace it.
The consumer survey answered a different question
Wendover relies on a 2016 consumer survey to argue that most people prefer cheaper delivery over paying extra for speed.
The video reports that 70% preferred the cheapest delivery option while approximately 23% were willing to pay more for same-day delivery.
The underlying McKinsey research surveyed more than 4,700 people across China, Germany and the United States. It examined general delivery preferences, not consumer adoption of drone delivery specifically. It also found that only 2% would pay enough extra to make instant delivery commercially viable under the survey's assumptions.
McKinsey
This is relevant evidence of price sensitivity. It is not proof that drone delivery cannot attract repeat customers.
The distinction is important.
A customer might be perfectly happy to wait two days for a phone charger but want a missing dinner ingredient in 15 minutes. These are different purchasing situations.
The commercial opportunity is not necessarily persuading customers to replace ordinary delivery with drones. It may be providing an additional option when conventional delivery is too slow.
There is evidence of this complementary role.
In reporting by the Associated Press in August 2025, DoorDash said that orders requiring human delivery drivers also increased in areas where it offered drone delivery. A restaurant owner in Frisco, Texas, reported a 15% increase in overall DoorDash orders, although he could not determine how much was attributable to customers switching to drones.
That is not proof of a causal increase in sales. But it challenges the assumption that drone deliveries must simply take customers away from conventional couriers.
Los Angeles Times
The two services may generate different kinds of demand.
The industry still has a profitability problem
Here is where I would apply the strongest remaining criticism to the drone-delivery comeback.
One million deliveries demonstrate commercial activity. They do not demonstrate profitability.
Walmart's announcement provides delivery volumes, operating locations and average delivery times. It does not provide the fully allocated cost of completing an order or the profit earned from each delivery.
The distinction is substantial.
In January 2023, McKinsey estimated the direct operating cost of a single-package drone delivery at approximately $13.50 under a model in which one person monitors one drone.
Its analysis found that labor could represent as much as 95% of operating costs.
However, McKinsey also modeled a substantially different scenario: if one operator could supervise 20 drones simultaneously, the estimated delivery cost could fall to approximately $1.50–$2.
The economic outcome therefore depends heavily on automation and the number of aircraft each operator can safely supervise.
McKinsey & Company
An Associated Press report in August 2025 said a Wing pilot could oversee up to 32 drones. That is meaningful progress, although it does not establish the actual cost per successful delivery or prove that every operation can use the same supervision ratio.
Los Angeles Times
The regulatory problem has not disappeared either.
The FAA published its proposed Beyond Visual Line of Sight (BVLOS) framework in August 2025. The proposal was intended to establish a more predictable pathway for routine operations without requiring pilots to maintain visual contact with every aircraft.
The FAA reopened the comment period in January 2026, and the rule remained in the final-rule process in October 2026. The Office of Information and Regulatory Affairs continued recording review meetings as late as October 29.
Individual operators can obtain specific authorizations, but the broader regulatory framework has not yet reached its intended final form.
GovInfo+2
The industry is expanding before the regulatory and economic questions have been fully settled.
That makes the million-delivery milestone impressive, but insufficient to establish a sustainable business model.
The real development is a different kind of delivery network
The original promise was a drone replacing the delivery van, travelling directly from a distribution center to practically any home.
That model encountered serious problems.
The emerging commercial model is different. Walmart supplies the retail network. Wing and Zipline supply the aircraft and delivery infrastructure. Existing stores become launch points. Customers receive small, urgent orders without requiring a conventional courier to make every journey.
Medical logistics provides another valuable market, particularly where conventional road transport is slow or unreliable.
This is a more specialized proposition than the 2013 vision, but it is also a more realistic one.
The unresolved question is whether operators can reduce the complete cost of each successful delivery enough to make the service profitable without relying indefinitely on promotional pricing, subsidies or strategic retail spending.
That is the number I would demand from Amazon, Walmart, Wing and Zipline before declaring drone delivery a commercially proven replacement for conventional last-mile transportation.
Verdict
Wendover was right to reject the fantasy of anything, anywhere, anytime delivery. Restricted airspace, unsuitable addresses, safety requirements and difficult economics remain genuine obstacles.
But the Phoenix calculation was never strong enough to quantify the market, and the single-base model was never a permanent limit on how the industry could organize itself.
By 2026, drone delivery had become a genuine commercial service in selected markets, with substantial delivery volumes and expanding retail partnerships.
The important development is not that drones have replaced delivery drivers. They have found a narrower job that conventional delivery networks do not always perform as quickly.
Wendover explained why the original dream failed. It did not adequately establish the limits of the business that replaced it.
And until the companies disclose their actual cost per completed delivery, neither the industry's triumphalists nor its critics have earned the right to declare the final economic verdict.
References
1. Drone Delivery Was Supposed to be the Future. What Went Wrong? — Wendover Productions, February 8, 2022. Original video. The original argument, Phoenix calculation and consumer-demand analysis examined here.
2. Units in Structure (ACS B25024) and Occupancy Status (ACS B25002) — U.S. Census Bureau, American Community Survey. B25024 · B25002. Establishes that the Phoenix housing percentages describe housing units rather than residents.
3. How Customer Demands Are Reshaping Last-Mile Delivery — McKinsey & Company, October 19, 2016. Original research. Provides the consumer survey underlying the video's delivery-price argument.
4. Public Involvement and Environmental Review for Drone Operations — Federal Aviation Administration. FAA records. Documents the Tolleson operating area and FAA-reviewed commercial drone-network proposals.
5. Walmart Celebrates 1 Million Drone Deliveries — Walmart, May 29, 2026. Company announcement. Provides the delivery milestone, participating stores, average delivery time and recent volume growth.
6. Amazon Is Bringing Drone Delivery Back to California — Los Angeles Times, August 29, 2026. Original report. Reports Amazon's expanding Prime Air operations, including actual service in the Phoenix area and customer pricing.
7. Drones Take to the Sky, Potentially Disrupting Last-Mile Delivery — McKinsey & Company, January 3, 2023. Economic analysis. Explains the labor assumptions behind the $13.50 delivery-cost model and the potential savings from multiple-drone supervision.
8. Delivery Drones May Soon Take Off in the U.S. Here's Why — Dee-Ann Durbin, Associated Press, August 12, 2025, republished by the Los Angeles Times. Original report. Provides independent reporting on Wing's operating model, human oversight, payload restrictions and DoorDash's experience.
9. Normalizing Unmanned Aircraft Systems Beyond Visual Line of Sight Operations — FAA, proposed rule published August 7, 2025; subsequent regulatory review through October 2026. Federal Register proposal · Official regulatory review record. Establishes the regulatory pathway and the continuing delay in finalizing routine BVLOS operations.