Drone delivery caps vs real deliveries: what FAA filings and company counts show

The big numbers in U.S. drone delivery are ceilings, not deliveries. Signed FAA assessments allow up to 400 deliveries a day per Wing nest or Zipline Dallas-Fort Worth site, and up to 1,000 a day per Amazon Texas center, so one signed metro filing can set a ceiling of 17,200 to 60,000 a day. Drafts ask for more, including 65,000 a day per Houston or San Antonio in Zipline's September 2026 Texas filing. Zipline's Dallas-Fort Worth assessment, signed in December 2025, sets a 30,000-a-day maximum but says Zipline expects about 100 deliveries per site per day once the network is built, a quarter of the cap. The same filings say actual flights will vary with demand and weather.

What the companies say they actually fly is far smaller. Zipline says it made more than one million deliveries in Dallas-Fort Worth this year as of November 5, 2026, which works out to a little over 3,200 a day, about 11% of its DFW ceiling. Wing said on October 29 it had already completed one million deliveries in the U.S. this year, and on November 6 it described that as over one million deliveries to homes from over 70 sites, a little over 3,300 a day against a combined per-site ceiling near 28,000 (all Servospace arithmetic on company figures). Amazon has published no running total; it told CNBC in August it makes "thousands daily" and its Prime Air chief says it will pass a 4 million annual run rate by year-end, about 11,000 a day. A recent U.S. site-level company figure, Amazon's hub in Richardson, Texas, filled about 150 orders a day in March against a 1,000-a-day assessment ceiling. Wing's 2023 DFW EA also states flown volume: fewer than 100 deliveries per operating day per nest at three Frisco and Little Elm nests.

None of this means drone delivery is small or stalling. Walmart's own count roughly doubled, from more than one million to more than two million deliveries, between May 29 and November 5, 2026. It means a city, retailer or investor should read an FAA cap as the noise and environmental envelope a company asked to be studied for, then look for flown counts, which come only from company statements, none of them audited or filed with the SEC.

Analysis based on 20 FAA environmental assessments, findings and re-evaluations for Wing, Zipline and Amazon Prime Air posted on the FAA's drone environmental review page, Walmart's May 29 and November 5, 2026 releases, Zipline's November 5 release, Wing's October and November 2026 posts, Amazon's 2025 shareholder letter filed with the SEC, EDGAR full-text searches of Walmart, Alphabet and Amazon filings, and reporting by CNBC, Retail Dive, the Dallas Observer, KBTX and DroneXL, all retrieved November 13, 2026. AI assistance was used in research and drafting, and the piece was checked by a separate AI fact-check and AI editorial review. Servospace did not interview the companies, observe operations or see flight logs. Calculations are Servospace arithmetic on published figures and are labeled as such.

In this analysis

• The short answer
• What kind of evidence each number is
• What the FAA filings allow
• Cap, expected average, and flown: three different numbers
• What the companies say they fly
• Per site and per day
• Deliveries, flights, trips, transits and orders
• What the SEC filings say: almost nothing
• Counterevidence and limits
• What a city, retailer or investor should ask for
• What would change this

The short answer

An FAA environmental assessment (EA) for drone delivery is a review of noise, wildlife, historic sites and other effects, prepared by the operator under FAA supervision, for a proposal to change the operator's Part 135 operations specifications. To keep the review valid for growth, the operator states a maximum: sites, docks or nests, hours, and deliveries per site per day. Going past those numbers needs more review. The maximum is therefore the most the company expects it might need, not a forecast, and several of the documents say so directly.

The flown numbers come from a handful of company statements, each with its own scope: Walmart counts its drone program across providers, Zipline counts one metro, Wing counts U.S. deliveries (Nov 6: to homes), and Amazon gives a run rate it expects to reach. On those statements, each of the three big U.S. operators is flying in the thousands of deliveries a day. Their approved and proposed ceilings in single metros are tens of thousands a day.

What kind of evidence each number is

What the FAA filings allow

The FAA posts each operator's EA on its drone environmental review page. The per-site ceiling is 400 a day for Wing nests and Zipline's older-style sites. It is 1,000 a day for the Zipline "Chargers" in its Texas and Arizona drafts and for most Amazon drone delivery centers (PADDCs); Amazon's Tolleson, Arizona center was assessed at 469. What grows from filing to filing is the number of sites and the size of each metro's envelope.

Zipline's newer filings change the unit. In the Texas metros and Arizona drafts, a "Charger" can hold up to 72 docks and fly up to 1,000 deliveries a day, and packages are picked up from separate "Dropboxes" at merchants. That design, not a change in demand evidence, is why the Texas draft reaches 220,000 a day across five metros. Our news story on the Arizona draft covers its public comment deadline and the noise analysis.

Cap, expected average, and flown: three different numbers

The clearest evidence that a cap is not a plan sits inside the EAs themselves.

• Zipline DFW. The final EA says "Initially, Zipline expects to fly considerably less than 100 deliveries per day from each site and then gradually increase to up to 400 deliveries per day at high volume sites as consumer demand rises." Its phasing table runs from 5 sites and 20 daily flights per site in month 1 to 75 sites at month 18, with the per-site figure given as "400 (expected 100 per site)". On those numbers the expected network at full build is 7,500 a day, a quarter of the 30,000 ceiling (Servospace arithmetic). The Seattle draft repeats the "expected 100 average per site" wording.
• Wing Houston and Atlanta. Both 2026 final EAs say: "Initially, Wing expects to fly considerably less than 400 deliveries per day from each nest and then gradually increase to 400 deliveries per day as consumer demand rises."
• Wing DFW, 2023. This is the only EA in the set that states flown volume. It says Wing's three operating nests in Frisco and Little Elm "have consisted of fewer than 100 deliveries per operating day per nest", while the same document raised the per-nest ceiling from 100 to 400.
• Amazon. The Texas EA describes 1,000 flights a day per center for 365 days as the analyzed maximum. At Richardson, one of the 22 Texas centers, Amazon's Sam Bailey told the City Council on March 9, 2026 that the center was "currently fulfilling only about 150 orders a day", according to the Dallas Observer, which also reported 13,000 orders since the hub opened. A local blogger's transcription of the meeting quotes the Amazon representative saying the site flies "between 150 to 250 flights per day" and that 1,000 is "the absolute maximum, which this site would not be able to hit" given its size and flight paths. Servospace has not heard the recording, so the quote is attributed to that transcription.

Caps also ratchet ahead of demand. Zipline's Pea Ridge, Arkansas operation went from 20 daily flights in its 2022 EA to 100, then to up to 400 delivery trips a day and 24-hour operation in a February 2026 re-evaluation. Wing's DFW per-nest ceiling went from 100 (2022) to 400 (2023), and its nest ceiling from 25 to 75 (2025). Each step is permission to grow, not a record that growth happened.

What the companies say they fly

These counts cannot be added. Walmart's two million spans every provider since 2021. Zipline's DFW million includes restaurants and other merchants, not only Walmart; Zipline's release says "Hundreds of businesses are now growing their revenue through Zipline orders." Wing's million covers all its U.S. partners. Wing's and Zipline's 2026 counts together already exceed two million, the same as Walmart's all-time total since 2021. That suggests a large non-Walmart share, but no company breaks it out. Walmart's May "200,000+ drone deliveries in Texas" is a floor and does not show how much of Zipline's DFW volume came from Walmart.

Per site and per day

Dividing these claims by days and sites gives rough averages, not rates. Every line below is Servospace arithmetic on company figures, and each claim is a floor ("more than").

Zipline's own boilerplate in Walmart's November 5 release says it "makes a delivery somewhere in the world every 15 seconds", which is 5,760 a day worldwide, including its African medical networks (Servospace arithmetic). That is consistent with a DFW average above 3,200 a day only if DFW is a large share of Zipline's global volume, which Zipline does not confirm or deny.

Deliveries, flights, trips, transits and orders

The filings and claims do not use one unit, and the differences matter for noise and for utilization.

The same Richardson meeting produced "about 150 orders a day" in one account and "150 to 250 flights per day" in another. Both can be true if some orders take more than one flight or if flights include repositioning. Neither is a filed number.

What the SEC filings say: almost nothing

All three companies behind the volume claims are public or owned by public companies, yet none reports drone delivery volume to the SEC. EDGAR full-text searches of filings from June 2025 to November 13, 2026 found:

• Walmart: no mention of "drone" or "Zipline" in its filings. The one million and two million figures come from its newsroom and an earnings call, not a 10-Q or 10-K.
• Alphabet: "Wing" appears as an exhibit name, a "Form of Wing Bet Performance Unit Agreement" filed with its first-quarter 2026 10-Q, plus proxy material. No delivery or revenue figure for Wing.
• Amazon: Prime Air appears in its 2025 shareholder letter, filed April 9, 2026, which says Prime Air "plans to serve communities with 30 million customers by year-end, and expects to deliver half a billion packages by the end of this decade". That is a goal and does not say whether half a billion is a yearly or cumulative figure.

These are absence findings for the filings searched as of November 13, 2026, not proof that no volume data exists. Zipline is private and files no reports. For comparison, the sidewalk robot sector has one SEC filer whose numbers can be checked; our Serve Robotics evidence guide shows what that looks like.

Counterevidence and limits

• The case for scale. Walmart's count roughly doubled in about five months, Wing and Zipline each passed a million deliveries in 2026 alone, and Walmart plans more than 270 Wing locations and 300 more Zipline sites. Volume is growing fast from a small base; low utilization of a cap is normal early in a network built for growth.
• Caps can bind. In Darlington, England, Amazon's trial was capped at 10 flights an hour and 100 deliveries per weekday, as DroneXL reported. That cap was far above use: Amazon reported more than 500 flights and deliveries from February to September, about three a weekday by DroneXL's arithmetic. Our Darlington story has the details. A UK cap is not an FAA EA, and the two systems should not be compared directly.
• Approved sites close. Amazon's College Station, Texas site received a final FAA supplemental EA and finding in September 2024 covering the MK30 at about 470 flights a day, yet Amazon ended deliveries there on August 31, 2025, with its lease expiring, KBTX reported. Amazon also briefly suspended its Tolleson, Arizona operation after two drones hit a crane boom in 2025, TechCrunch reported.
• Low averages can still be loud. Richardson residents complained of noise from a site Amazon says fills about 150 orders a day, with most of its traffic directed over one neighborhood by a hospital no-fly zone, according to the Dallas Observer. A site at a fraction of its cap can still be the dominant noise source for the homes under its main path.
• Operating counts in EAs are snapshots. Wing's 2025 DFW supplemental EA says in one passage that Wing operates from 16 nests and in the next that the 75 would include "the 15 currently operating nests". The document does not resolve the difference.
• What this piece cannot see. Servospace did not review FAA Part 135 operations specifications themselves, which are not public in full, or city permits outside Richardson. Company counts are unaudited, use their own definitions, and are dated differently. Per-day averages spread volume evenly across days and sites when it is not.

What a city, retailer or investor should ask for

What would change this

• Walmart's third-quarter fiscal 2027 results, November 19, 2026. Watch for an updated drone count, store count, or any split between Wing and Zipline.
• Amazon's year-end figure. If Amazon confirms it passed a 4 million annual run rate, the Amazon row above should be redone with its stated date and scope.
• Zipline's Texas and Arizona decisions. A signed FONSI would turn the 220,000 and 130,000 daily requests into ceilings; it would still say nothing about flown volume.
• Any site-level disclosure. A city meeting, permit record or partner filing that gives per-site deliveries would be the first check on these averages outside company releases.

Servospace will update this guide after Walmart's November 19 results and when a new EA is signed. Related: Wing's Fort Worth manufacturing plan, Zipline's $1.5 billion raise, our drones topic page, and for Walmart's warehouse automation, what Symbotic's filings show.