DARPA Just Asked for Proof of a 4:1 Drone Nobody Officially Built. Inside DV022, the $1.5M Direct-to-Phase-II SBIR Closing September 23.

On August 9, 2026, the DARPA Lift Challenge ended at the National Museum of the U.S. Air Force in Dayton with a result that reads, on its face, like a near miss. AVIDrone's Katana — a 29.3-pound aircraft — carried 112.4 pounds around a five-nautical-mile course for a payload-to-weight ratio of 3.84:1. The target was 4:1. Under prize rules that paid the full purse only to teams that hit or exceeded the threshold, AVIDrone collected $1.25 million instead of $2.5 million. Second place (MTech Operations, 3.64:1) took $750,000. Third (Xtreme Aerial Concepts, 3.45:1) took $500,000.

Seventeen days later, DARPA opened a Direct-to-Phase-II SBIR topic that requires proposers to submit empirical proof they have achieved a 4:1 payload-to-weight ratio in order to bid.

That is not a contradiction. It is the most informative thing DARPA has said about heavy-lift UAS all year, and reading it correctly determines whether your company should spend the next month writing a proposal or should stand down and wait for Release 6.

What DV022 actually funds

DPA26BZ05-DV022 — Commercialization of Ultra-High Payload-to-Weight UAS Subsystems, Tactical Technology Office. It arrived as part of DARPA's FY26 SBIR Release 5, which pre-released August 5, opened August 26, and closes September 23, 2026 at 12:00 PM ET. Technical questions stop being answered September 16 — DARPA does not respond to inquiries inside the final seven calendar days.

The money:

The word doing the heaviest lifting in the title is Commercialization. This is not a research topic. DARPA is not asking anyone to discover whether 4:1 is possible — the Lift Challenge answered that question in August, and answered it more emphatically than the official leaderboard suggests. This topic is asking whether a demonstrated capability can be turned into a manufacturable, certifiable, mass-producible subsystem.

The two tracks are the Challenge's design prizes, promoted

DV022 splits into two tracks, and a company may submit one proposal per track but cannot win both.

Track A — Revolutionary Aerodynamic Design. Ultra-lightweight structures and novel airframe/rotor configurations, plus the commercialization scaffolding: communications, autonomy, safety systems, and the start of a certification campaign. Track A carries a hard 1,320-pound maximum gross weight — a number chosen to sit inside FAA Part 108 rather than for any aerodynamic reason.

Track B — Revolutionary Powertrain Design. Transitioning high-efficiency, high-power-density propulsion from one-off prototypes into mass-produced, certified modules that work across multiple UAS platforms.

Now compare that to the Lift Challenge's two subjective awards, each worth $500,000:

• Most Revolutionary Aerodynamic Design — University of Maryland's Autonomous Micro Air Vehicle Team, for a tri-rotor that transitioned horizontal propellers into forward flight.
• Most Revolutionary Powertrain Design — Portal Aircraft Company of Mansfield, Texas, for turbine exhaust-driven rotors that eliminated mechanical linkages.

The SBIR tracks are not merely similar to the Challenge categories. They are the same two categories, with the prize money replaced by a development contract. DARPA ran a public competition to find out which two technical avenues were most likely to break the payload barrier, got its answer in a week of live flight trials, and immediately stood up a funding line down each avenue. That is the fastest concept-to-solicitation loop the agency has run in the UAS space in years, and it is worth understanding as a pattern: DARPA prize challenges are increasingly the front end of a SBIR pipeline, not standalone publicity events. Director Stephen Winchell said as much in the award announcement, noting the agency would support maturation through SBIR topics and investor introductions via its Commercial Strategy Office.

The entry bar is the whole story

Here is where most companies reading this should stop and do honest arithmetic.

DV022 is Direct to Phase II only. There is no Phase I. To be responsive, proposers must document existing proof of concept — "empirical data, flight logs, or prototype test results" — establishing the 4:1 threshold. DARPA states plainly that Lift Challenge participation satisfies this requirement, while allowing that data from "similar rigorous testing environments" also counts.

Now look at the Challenge funnel:

Nine. Out of 489 applicants, nine aircraft produced an officially scored run, and the best of those was 3.84:1. On the face of it, the population of companies holding clean 4:1 evidence is close to empty.

Except for one entry on the results sheet. DefendTex, of Moab, Utah, won Most Promising Technology for a string-supported structure that achieved 9.63:1 — more than double the target — on a run that did not complete official scoring. That single unofficial number reframes everything. It says the 4:1 barrier is not a materials-science wall; it is an integration-and-reliability wall. Teams did not fail to lift. Teams failed to lift and fly the full course and drop the load and fly the extra mile and have all of that captured under scoring conditions.

That distinction is precisely what DV022's milestone schedule is built to attack.

The gates: variance, not peak performance

Read the three readiness tests in order and the topic's real thesis becomes obvious.

Month 9. Track A must deliver three identical structural components with less than 10% variance in failure loads at the 4:1 ratio. Track B must deliver three mass-production-ready units with under 5% performance variance on a dynamometer.

Month 18. Track A must fly an integrated UAS at or under 1,320 pounds, with full communications, autonomy, and safety systems, in an untethered hover at 4:1 payload for five continuous minutes. Track B must complete accelerated lifecycle and environmental stress testing to MIL-STD-810, plus a continuous-operation test on a heavy-lift mission profile.

Month 24. Track A delivers a complete technical data package and formally initiates certification under AS9100 and FAA airworthiness standards. Track B delivers a production-focused data package mapped to FAA and MIL-STD requirements.

Notice what is absent: nowhere does DARPA ask anyone to beat 4:1. Not once. The month-9 gate is a repeatability test — build three, prove they behave the same. The month-18 gate is a duration test — five minutes of untethered hover, which is unglamorous next to a five-mile race but far harder to fake. The month-24 gate is a paperwork test in the most consequential sense: AS9100 quality-system certification and an opened FAA airworthiness file.

DARPA already knows the physics works. What it does not have is a supplier that can build the same thing twice.

Who should bid, and who is kidding themselves

Bid if you have scored flight data at or above 4:1. That is a small set — realistically the Lift Challenge finalists, DefendTex given the 9.63:1 run, and a handful of firms with proprietary test programs. If you are in it, DV022 is close to a sole-purpose contract vehicle built around your existing hardware, and the two-track split means an airframe house and a powertrain house are not competing head-to-head.

Bid if you own the subsystem, not the aircraft. The title says subsystems. Track B in particular is a propulsion-module play — you do not need a complete flying vehicle to win it, you need a powertrain that survives MIL-STD-810 and comes off a line with under 5% unit-to-unit spread. Motor, ESC, and turbine-hybrid specialists with drivetrain test data should look hard here even without a Lift Challenge entry.

Do not bid if your 4:1 evidence is a simulation, a component-level bench result, or a tethered lift. DARPA named the acceptable evidence types — empirical data, flight logs, prototype test results — and a Direct-to-Phase-II topic with a hard feasibility gate is the single easiest place for a reviewer to disqualify a proposal without reading the technical volume. Over-claiming here does not buy you a partial score; it buys you a non-responsive determination.

The compliance overhead is not boilerplate

Three requirements will disqualify unprepared teams regardless of technical merit.

ITAR/EAR restricted. Export controls apply to the full effort. Foreign national involvement must be disclosed with country of origin, visa status, and the specific tasks each person performs. For a drone startup staffed largely by international graduate students — a common profile in this sector — this is a real structural constraint that needs answering before you write, not after.

CMMC Level 1. Cybersecurity Maturity Model Certification is required. Level 1 is the entry tier and achievable through self-assessment, but "achievable" still means an actual assessment against seventeen practices with a signed affirmation in SPRS. If you have never touched CMMC, start today, not in week three.

Venture-backed eligibility. Companies majority-owned by venture capital, hedge funds, or private equity may apply, but only if registered in the SBA Company Registry Database. That registration is not instantaneous. Given how much of the heavy-lift drone sector took institutional money in 2024–2026, this catches more teams than it should.

The proposal itself is a 20-page technical white paper plus a 15-slide deck, across seven required volumes including the Company Commercialization Report, Fraud Waste and Abuse training certification, and the Foreign Affiliations webform.

Why the certification language matters more than the money

$1.5 million is a modest DARPA award. The reason DV022 is strategically significant sits in the month-24 deliverable: formal certification initiation under AS9100 and FAA airworthiness standards, with Track A's weight ceiling explicitly set at 1,320 pounds to align with Part 108.

Part 108 is the beyond-visual-line-of-sight rule framework that determines whether heavy-lift drones can operate commercially at scale in U.S. airspace. By writing the weight limit to that number and funding the start of an airworthiness file, DARPA is not buying a defense prototype. It is buying a certifiable commercial product that also does tactical resupply, ammunition delivery, and casualty evacuation — the Phase III applications the topic names alongside construction-materials delivery, disaster-response logistics, agriculture, and cargo transport.

Program manager Phillip Smith framed the underlying bet directly: vertical lift is "the next great technological inflection point that is economical for widespread adoption." Today's specialized helicopters run roughly 1:1 payload-to-weight. Small multirotors are usually worse. A fielded 4:1 system does not improve military logistics incrementally — it changes what "resupply" costs by an order of magnitude, and it does the same for every civilian sector that currently moves heavy things short distances with a truck and a road.

DARPA spent $6.5 million and a week in Dayton establishing that the engineering is real. DV022 spends $1.5 million per awardee establishing whether anyone can build it repeatably enough to certify. That is a narrower question, a harder one, and the only one that still stands between a demonstration and a product.

Proposals close September 23 at 12:00 PM ET. Technical questions close September 16. If you are holding 4:1 data, the clock on the most target-rich SBIR topic in the UAS sector is measured in days.

Want federal SBIR topics matched against your company's actual test data and technical profile the week they pre-release? Granted tracks solicitations across DARPA, DOD, DOE, NSF, and every other agency — and surfaces the ones where your capabilities clear the entry bar, not just the ones that sound relevant.