Grubhub’s drone delivery pilot in Green Brook, New Jersey, is the first commercial drone food delivery program in the state — and the first one in the U.S. anchored to a multi-restaurant food hall rather than a single restaurant. The aircraft is a Dexa DE-2020 operating under Dexa’s Part 135 air carrier certificate, the food comes from a Wonder food hall on Route 22, and the pilot runs for three months starting March 18, 2026. Multi-source confirmation: Grubhub’s own announcement, Reuters, Bloomberg, CBS New York, and local outlets from NJ.com to News 12 New Jersey all pin the launch date to March 18.
This is the field report: what flies, what the operation looks like from the customer’s app, how Dexa got the regulatory ground cleared, and where this pilot fits into the broader 2026 US drone-delivery landscape. The lone similar U.S. food delivery pilot we have anchored to a single-operator network is the Matternet NHS London expansion, where the same aircraft-class (multi-rotor, 2-kg payload, tether-or-low-altitude descent) is flying medical samples in a BVLOS corridor under a different Part 135-style operating certificate. The article’s load-bearing angle is the multi-restaurant anchor — the prior U.S. food delivery pilots all picked up from a single brand. Green Brook picks up from a Wonder location that operates 15+ restaurant concepts under one delivery roof.
The 30-second version: what’s actually flying in Green Brook in 2026
As of mid-2026, the Grubhub pilot is a single-site operation: one Wonder food hall, one Dexa DE-2020 fleet, one 2.5-mile radius around the Route 22 location. Customers within that radius can order any menu item from any Wonder restaurant through the standard Grubhub app — the drone option appears for eligible orders, and the delivery is free of any drone-specific surcharge. The order still flows through Grubhub’s marketplace; the drone operation is on the fulfillment side, not the demand side.
Operational envelope, as published by Grubhub and confirmed by Fox Business (reporting on launch day, March 18, 2026, authored by Bonny Chu) and NJ.com:
- Service window: Daily from 10:30 a.m. until sunset. The sunset cutoff is a visual-line-of-sight accommodation — not because the drone stops flying, but because the FAA Part 107 commercial-operations framework still requires day VFR visibility unless the operator holds an explicit night-BVLOS exemption. Dexa holds night authority under its Part 135 certificate but the pilot is not exercising it (see the night-flying regulatory comparison on our night drone operations guide for the full distinction).
- Service radius: 2.5 miles from the Wonder location, measured by road-radius rather than as-the-drone-flies. Inside that radius, the drone can reach any address that has a small clear drop zone (front yard, driveway, sidewalk strip).
- Payload ceiling: 4 lbs. Orders over 4 lbs are not eligible for drone delivery, and the Grubhub app silently routes them to the standard couriers.
- Price ceiling: $40 per order. Above $40, the order is excluded from the drone-eligible pool — the cut-off is a food-safety/time-to-customer consideration, not a regulatory cap.
- Cost to customer: No drone-specific surcharge. Standard Grubhub service fees and tip apply.
The pilot is structured as a three-month test. If the operation proves out — on-time delivery, food temperature at drop, customer retention, drone-side reliability — Grubhub and Dexa have signaled expansion to other Wonder locations through 2026 and 2027. The expansion plans are not contractually committed; the pilot is the gating step.
Why the Wonder food hall changes the unit economics
Every prior U.S. commercial drone food delivery pilot — Wing in Christiansburg, Virginia, Flytrex in Dallas/Fort Worth and the Research Triangle, DoorDash Air in its first in-house markets — has been anchored to a single restaurant or a single small chain. The drone picks up the order from that restaurant’s kitchen and delivers to the customer. The unit economics are tight: the drone is a single-purpose asset, and the order volume is constrained by the restaurant’s own demand ceiling.
Wonder operates differently. A Wonder location is a food hall, not a restaurant — typically 15+ restaurant brands operating under one delivery roof, with a single kitchen doing the final prep and a single pickup counter. A single drone operating out of a Wonder pickup counter can serve orders from any of those 15+ brands. Order volume per drone-hour is the sum of the menu of the hall, not the menu of a single restaurant.
Restaurant Technology News (March 13, 2026) called this out explicitly: “Wonder’s multi-restaurant model takes flight in Grubhub drone delivery pilot.” The framing is that the food-hall model is the unit-economics unlock — the drone’s per-flight operating cost is roughly the same regardless of which restaurant the food came from, so the more restaurants the drone can serve, the higher the throughput per flight. A drone operating out of a single-restaurant kitchen typically completes 6-10 deliveries per hour in dense suburbs; a drone at a Wonder food hall can run a higher effective flight count because the order pool is wider.
This is the part of the story that distinguishes the Green Brook pilot from prior U.S. food delivery pilots. It is not a new drone — Dexa’s DE-2020 has been flying since 2025. It is not a new regulatory framework — the Part 135 air carrier certificate Dexa earned in April 2025 is the same certificate Flytrex and Wing operate under. The novelty is the multi-restaurant anchor. The article’s framing — that this is the third major U.S. commercial food delivery pilot of the 2026 cycle, and the first one to stress-test the multi-restaurant model — is the editorial hook the article is built on.
Inside the operation: route, payload, hours, and the DE-2020 tether system
The DE-2020 is a quadcopter in the 55-lb gross-weight class — large enough to require Part 107 sub-55-lb registration but small enough to qualify under a Part 135 air carrier exemption rather than a full type certificate. The airframe’s special-class airworthiness criteria were published by the FAA in the Federal Register on March 17, 2022 (originally under the Telegrid Technologies name; Telegrid subsequently rebranded to Dexa). The original finding was for a tethered delivery system, and the Green Brook operation uses that tether on every flight.
The DE-2020 deployment model is the defining operational difference from Flytrex and Wing. Where Flytrex and Wing both land at the customer address (or hover low enough to drop from a low altitude), the DE-2020 stays airborne at cruise altitude throughout the delivery. The payload is reeled down on a tether to the customer’s drop zone — front yard, driveway, sidewalk strip — and the aircraft never descends below a few meters AGL. The point of the tether system, per Dexa’s own engineering rationale, is that it eliminates the need for a landing zone at the customer address and removes the risk of the drone tipping, ingesting debris, or being interfered with by pets or children. The tether operation is essentially the same hybrid sit-and-fly profile thatour BVLOS operations guide covers for the broader commercial drone market — the DE-2020 just specializes it for handoff at altitude.
The tether system has a payload penalty. The DE-2020’s effective payload — the food the drone can carry while still maintaining a safe hover under tether — is the 4-lb cap that Grubhub’s app enforces. For a 4-lb order, the drone’s flight time is roughly 12-15 minutes at 2.5-mile radius, factoring in launch, cruise, descent-to-tether-altitude, tether-down, tether-up, and return. Within that envelope, the drone typically completes 4-6 deliveries per hour per aircraft.
The launch and recovery are at the Wonder food hall rooftop — a dedicated drone operations area with weatherproof charging stations, a mission control station, and a fenced loading area. QSR Pro (March 24, 2026) published the only detailed operational view of the rooftop: the Wonder staff brings the order to a window-side pickup counter, the drone crew loads it into the payload bay, the drone launches, the customer gets an app notification with an ETA, and the drone returns to the rooftop for the next flight. The end-to-end order time is 12-20 minutes from order to tether-down, which is competitive with a courier for distances under 2 miles and clearly faster than courier for any address outside the driver’s reasonable driving radius.
The regulatory ground cleared: Dexa’s Part 135 cert and the S-1 airworthiness
The 2026 Green Brook pilot could not have launched without two regulatory milestones Dexa had spent roughly three years clearing. The first was the air carrier certificate. The FAA issued Dexa a Part 135 air carrier certificate in April 2025 — the same regulatory class as a small commercial airline, with operating specifications that permit the DE-2020 to operate for compensation or hire. DroneLife (Miriam McNabb, April 17, 2025) covered the certification as a stand-alone story; at the time, Dexa was the third U.S. drone operator to hold a Part 135 air carrier certificate, after Wing (a subsidiary of Alphabet) and Zipline. The certification is the reason Grubhub’s drone network can legally operate for compensation — without it, every Green Brook flight would be a violation of the FAA’s compensation-or-hire rule. For the broader regulatory anchor, the FAA Part 107 framework walkthrough covers the side-by-side distinctions between Part 107 small-UAS operations and Part 135 air carrier operations.
The second milestone was the FAA’s nationwide environmental finding issued in July 2026. The Flight Brief (G. Reverdiau, March 24, 2026, with the FAA’s July 2026 NEPA finding cited as the most recent state of play) summarized the move: the FAA published a Final Programmatic Environmental Assessment and FONSI/ROD covering small UAS commercial operations nationwide, with the explicit effect of overriding the state-level objections that 16 state attorneys general had filed against the proposed small-UAS rule. For operators like Dexa, this shortens the market-expansion timeline — instead of clearing a separate environmental review for each new city, the operator can rely on the nationwide programmatic finding. The Green Brook pilot launches under that umbrella.
For comparison, Flytrex holds a separate Part 135 air carrier certificate and uses its own operations specifications (issued in conjunction with the FAA’s August 2025 BVLOS approval). Wing operates under Alphabet’s Part 135 certificate and uses the company’s own BVLOS-enabled operations specifications. Each of these operators has a different path to the same legal state: an FAA-recognized commercial operating certificate with aircraft-specific operations specifications. Green Brook is the third commercial food delivery pilot in the U.S. operating under this regulatory frame; the structural variation at Green Brook is the multi-restaurant anchor, not the regulatory mechanism.
Why this isn’t a Part 108 commercial BVLOS playbook yet
The U.S. drone-delivery community spent 2024 and 2025 waiting for the FAA to publish a Part 108 rule — the proposed framework for routine commercial small-UAS operations, including BVLOS, package delivery, and operations over people. The rule remains a proposal as of mid-2026. The FAA’s Part 135 air carrier regime is the only fully-formed regulatory path for commercial drone delivery; until Part 108 is finalized, every Green Brook flight, every Flytrex flight, and every Wing flight is operating under the older 14 CFR Part 135 framework, with aircraft-specific operations specifications layered on top. For the international comparison (EASA Open, CAAC, etc.), our FAA Part 107 vs CAAC vs EASA guide covers the three regional regimes side by side.
What the absence of Part 108 means in practice is that the commercial drone delivery segment is still constrained by the air carrier certificate application timeline. Each new city requires the operator to amend its operations specifications, which the FAA approves on a case-by-case basis. For Dexa, that means the Green Brook site is the only authorized delivery site as of mid-2026; expanding to a second Wonder location would require an OpSpecs amendment, which can take 60-180 days. The programmatic environmental finding helps, but the operations specifications are still operator-and-aircraft-specific.
There is a related, separate question about whether the FAA’s proposed Part 108 will eventually replace Part 135 for routine small-UAS commercial delivery. If Part 108 finalizes in 2026 or 2027, the operators expect to migrate their operations specifications from Part 135 to Part 108; the operational fact (a drone picking up food from a restaurant and delivering it to a customer) is the same. The migration is a paperwork exercise, not a redesign. The carriers’ investments — fleet, command-and-control infrastructure, route surveys — are portable across both regulatory frames.
For operators watching the pilot, the practical takeaway is that the Green Brook test is the third validated point in a still-thin U.S. commercial food delivery data set. The first two data points were Wing in Christiansburg and Flytrex in Dallas/Fort Worth / Research Triangle; Green Brook is the third. What the pilot tests is whether the multi-restaurant anchor increases the through-put-per-flight enough to justify the per-site operating cost of the drone fleet.
What’s not in the pilot — the limits regulators and Grubhub drew on purpose
There are several constraints the pilot is operating under that are not always obvious from the marketing-side coverage. The first is the geographic radius. 2.5 miles is a small operational area — every eligible customer is within a 5-minute drone flight of the Wonder location, which keeps the tether down-time and the risk surface low. The radius is not the regulatory maximum; it is the operating radius Dexa and Grubhub agreed to for the pilot. Expanding to a 5-mile radius would require a new operations specifications amendment with the FAA. For a useful comparison on what BVLOS rules actually permit vs what the Green Brook pilot is exercising, the FAA’s Remote ID enforcement reality guide walks through the boundary cases.
The second constraint is the payload ceiling. 4 lbs is well below the DE-2020’s physical cargo capacity — Dexa has demonstrated heavier loads in test environments. The 4-lb cap is the Grubhub-app-side filter, not the FAA-side limit. The cap reflects the food-safety expectation that a 4-lb order is a typical single-meal-and-drink delivery, and that a heavier order is more likely to be a multi-meal or catering order that requires a thermal bag or special handling. The drone’s payload bay does not have a refrigerated compartment; the food is in the same insulated bag Grubhub’s couriers use.
The third constraint is the time-of-day limit. The sunset cutoff is a visual-line-of-sight accommodation. Dexa holds night operations authority under its Part 135 certificate, but the operational practice for the pilot is to keep every flight in day VFR. The FAA’s Part 107 small-UAS rule prohibits night operations without explicit anti-collision lighting and a night BVLOS waiver; Part 135 air carriers can apply for night authority, and Dexa has the certificate, but the pilot is not exercising that authority. The pilot is the conservative mode.
What this means for operators who are evaluating the pilot’s results is that the unit economics may shift as the operation expands. The 2.5-mile radius caps the eligible-customer pool; the 4-lb payload cap reduces the per-order value; the sunset cutoff reduces the daily operating hours. Each of these is a controlled constraint, not a regulatory maximum, and an expanded second-site pilot would likely relax one or more of them. The first three months are the conservative-operating-mode test; the next site is the variable relaxation test.
FAQ
The questions that operators, restaurant-tech integrators, and commercial pilots are most likely to ask about the Green Brook pilot, with the clearest available answers from the public sources as of August 2026.
Is Grubhub drone delivery live in New Jersey now?
Yes, at one location — the Wonder food hall in Green Brook, NJ (Somerset County). Service runs daily from 10:30 a.m. to sunset, within a 2.5-mile radius, for orders under $40 and under 4 lbs. There is no charge beyond standard Grubhub service fees.
Who actually flies the drone — Grubhub or Dexa?
Dexa flies the drone. The DE-2020 is a Dexa-owned Part 135 air carrier aircraft under Dexa’s air carrier certificate; Grubhub is the marketplace/demand side, and the order is routed through the standard Grubhub app.
What is the anchor restaurant — and why does it matter?
The anchor is a Wonder location, not a single restaurant. Wonder operates 15+ restaurant brands under one delivery roof, so a single drone pickup can be any menu item from any resident restaurant. This is the multi-restaurant model that distinguishes this pilot from prior single-restaurant drone deliveries.
Why isn’t drone food delivery just allowed under Part 107?
Part 107 small-UAS rules explicitly prohibit carrying property for compensation or hire. Commercial food delivery today requires a Part 135 air carrier certificate with individual exemptions and operating specifications. The FAA’s proposed Part 108 framework is still a proposal as of 2026 — it has not been finalized.
Will this expand to other Grubhub markets?
The pilot is structured as a three-month test (launched March 18, 2026). Expansion to other Grubhub markets depends on three factors: Dexa’s ability to clear additional geographic approvals under its Part 135 certificate, the FAA’s nationwide environmental finding (July 2026) which shortens the timeline, and Wonder’s footprint (Wonder is the anchor partner, not Grubhub’s existing restaurant base).
Sources
- Grubhub newsroom — operational envelope, daily hours, no surcharge, launch date
- Reuters — wire-level multi-source confirmation of the launch date and operator partnership
- Bloomberg — three-month trial framing, business-side context (Natalie Lung, named journalist)
- CBS New York — day-before-launch explainer, March 17, 2026
- Fox Business — launch-day coverage, Bonny Chu, named journalist
- NBC New York — “Meals from the sky” framing, March 19, 2026
- Los Angeles Times — wire-level confirmation of the trial launch
- NJ.com — Somerset County local coverage, March 12, 2026
- News 12 New Jersey — launch-day local coverage
- Restaurant Technology News — Wonder multi-restaurant anchor context
- QSR Pro — QSR-logistics operational view, March 24, 2026
- DroneLife (Miriam McNabb) — Dexa’s April 2025 Part 135 certification + S-1 airworthiness
- The Flight Brief (G. Reverdiau) — FAA Part 108 still proposed, Part 135 is the commercial path
- Wonder newsroom — the multi-restaurant anchor partner’s own announcement
- FAA Part 107 waivers — the regulatory anchor for BVLOS compensation-or-hire rules
- Federal Register — special-class airworthiness criteria for the DE-2020 (originally Telegrid DE2020)