Can the US build a DJI?

In December 2025, the FCC added all foreign-manufactured drones to its "Covered List," a national security registry also occupied by Huawei and ZTE, which blocks new models from being sold in the US. The case against foreign drone manufacturers has been building since 2017, when the Department of Homeland Security flagged that China's newly passed National Intelligence Law could compel companies like Shenzhen-based DJI to hand over sensitive user data for intelligence. In 2022, the Pentagon added DJI to its list of "Chinese military companies", and when no agency cleared DJI under the FY2025 NDAA, the sales ban officially went into effect.

DJI is by far the biggest name in the drone world, owning ~70% of the global drone market and beloved by hobbyists for its Mini line, which balances great image quality with usability, cost, and mass. Most notably, the mass of each Mini is under 250 grams, the threshold that triggers licensing and registration requirements in most markets. DJI has also demonstrated remarkably fast iteration loops with the Mini line, shipping seven models between 2019 and 2023.

DJI drones are also popular among commercial, government, and defense buyers. In 2025, the California Fire Chiefs Association, which opposes the ban on the grounds that no comparable replacement exists, estimated that over 90% of drones flown by US police and fire departments are DJI.

To find out, we tore down two popular DJI drones — the 2019 Mavic Mini and the 2023 Mini 4 Pro — producing the first per-part bill of materials (BOM) published for either model. We started by taking apart each drone and tracing the source and price of every component. By our estimates, DJI not only sustained exceptionally high hardware margins but also grew them from 72% to 79% between 2019 and 2023. For comparison, Apple, a $4.6 trillion dollar company famous for its supply chain, achieved a 53% gross margin on the iPhone 15 Pro Max.

Our BOM comparison sheds light on how DJI compounded its cost and capability advantage over the intervening four years: what it brought in-house, what it onshored, and what it kept sourcing overseas.

From 2019 to 2023, the share of DJI's BOM cost sourced from China rose 28 percentage points: 22 of those points were driven by DJI's in-housing.

We see this verticalization most clearly in the avionics, where we estimate 26% of total BOM costs for the Mini 4 Pro came from custom electronics, up from 1% for the Mavic Mini. Custom ASIC development typically costs $50 million in R&D, driven by labor, EDA tooling, and mask sets, and only makes economic sense when amortized across millions of units. We suspect DJI began custom silicon development around 2015, when its revenue was estimated to surpass $1 billion.

DJI followed a playbook that Apple ran first: start as a system integrator, buy the best available parts to ship fast and capture share, then move to custom silicon once volume amortizes the design cost, building an integrated product competitors can't buy, at a price they can't match.

DJI's custom silicon first appeared on the 2018 Mavic Air, with the H3 video ASIC. As the pricier line, the Mavic Air generated enough margin per unit to absorb custom silicon's development costs. The Mini reached that threshold in September 2023; we estimate that the Mini 4 Pro carries four custom chips: a vision SoC, a companion Image Signal Processor (ISP), an O4 radio SoC, and a core processing SoC. Across vision, flight control, and radio, DJI in-housed 19% of BOM costs for the 2019 Mavic Mini. Assuming vendors charged a 70-100% markup for the Mavic Mini's chips, DJI would save $7-8 in chip costs per unit of the Mini 4 Pro.

In addition to price, DJI also optimized for mass. Our teardown suggests DJI folded flight control directly into the main SoC, creating a $139 BOM drone that offers cutting-edge flight control, radio, and video processing while staying under 250 grams. Competitors stitch the same functions together from discrete parts, spend thousands on avionics alone, and still end up over the 250 gram regulatory threshold.

Across four key sockets that made up 14% of the Mavic Mini's BOM cost, DJI switched from international to Chinese suppliers by the Mini 4 Pro.

Batteries represent the clearest case, and demonstrate the structural advantages a China-based manufacturer has. DJI's headquarters in Shenzhen's Nanshan district sits inside the industrial belt where China makes most of the key hardware used to build drones globally: gimbal motors with sintered magnets, flight controllers, RF modules, and lithium-polymer battery packs. Layer on the fact that China produces over 80% of the world's lithium batteries, and DJI's onshoring means it leverages both the cheapest and most advanced supply base.

Instead of saving money, DJI paid more to trade up: the Mini 4 Pro's pouch-cell packs from Ampace cost ~$8.60, about two dollars more than the Mavic Mini's Samsung cells. The extra two dollars bought 38% higher energy density (from 176 Wh/kg to 243 Wh/kg) with pack mass falling by 20 grams (from 98 grams to 78 grams) and total energy increasing 10% (from 17.28 Wh to 18.96 Wh).

By shedding 20 grams from the battery pack (8% of the drone's total mass), DJI freed capacity for more sensing and compute hardware. The Mini 4 Pro's board weighs 1.3x its 2019 counterpart, carries 1.5x more chips, and pairs with a camera subsystem 1.6x heavier, together enabling omnidirectional obstacle avoidance and better low-light performance.

Based on our investigation, every Mini from 2019 to 2023 uses a GNSS receiver from Switzerland-based u-blox.

There's no evidence that DJI stuck with u-blox because of a domestic capability gap. Only the cheap alternatives lag: Zhongkewei’s nearly pin-compatible u-blox clones cost roughly $2 less per unit, but they can't track as many satellites, leading to worse positioning accuracy. Unicore (和芯星通, part of the BDStar group) makes multi-band, multi-constellation RTK silicon that achieves parity with u-blox's high-precision line.

It's possible that switching costs were too high. Modern hardware is software-defined, so replacing a chip isn't just a hardware swap. DJI's flight stack uses u-blox's proprietary UBX binary protocol: switching receivers would have required re-integrating and re-validating the software behind the Kalman filter, which sits at the core of flight control and is critical to both performance and safety.

The more likely explanation is that the switching cost of the ecosystem was too high. Over two decades, u-blox has built an ecosystem of support services and tools around its chips: AssistNow speeds up satellite lock, PointPerfect delivers global correction data for precision positioning, and u-blox offers a library of reference designs so engineers can build around the chip without starting from scratch.

US-based Skydio, considered DJI's closest competitor until it exited the consumer market in 2023, built a more intelligent drone with better autonomy, but it didn't sell. While DJI's 2019 Mavic Mini lacks proactive obstacle avoidance, the 2019 Skydio 2 can route itself through dense obstructions. It has six 4K cameras that feed an NVIDIA Jetson TX2, enabling it to track up to ten objects moving 36 mph at once. It can also autonomously plan a path around obstructions, rather than just detecting them and braking.

The Skydio 2+ arrived in January 2022 with modest improvements, a denser battery pack, and a higher price starting at $1,099, before Skydio left the consumer market in August 2023. A month later, the Mini 4 Pro closed the gap on omnidirectional camera coverage, transmission range, and flight time, as well as real-time route planning around obstacles, using four fisheye vision sensors versus the Skydio 2+'s six navigation cameras.

DJI fell short of Skydio on fully autonomous flight and subject tracking, but not because of mass constraints: DJI's Mavic 3 carried eight vision sensors at 895 grams, and most reviewers still place the Skydio 2 ahead on tracking. Consumers weren't willing to pay for the difference, so Skydio narrowed its focus to enterprise buyers who would.

Skydio proved that a US company can not only build drones at parity with China, but surpass it on performance. What Skydio couldn't do was match DJI on price and mass while turning a profit. Even while charging a premium, Skydio reportedly lost money on every consumer sale. It sourced from similar suppliers, but on worse terms, with a third of DJI's iteration speed.

Distance explains much of it. Both companies bought key components from the same industrial belt: Skydio's battery cells came from Dongguan, roughly 50 kilometers up the Pearl River Delta from DJI's headquarters in Nanshan. But DJI sits inside that belt, and Skydio sits 7,000 miles away, so components that reach DJI in one to two days take weeks to reach Skydio. This gap matters most in the early design-build-test loop, where iteration speed is rate-limited by the slowest part delivery. PCB revisions came back in hours for DJI, versus three to six weeks for Skydio. As a result, a Shenzhen-based team can run two to three iteration cycles in the time a US team like Skydio runs one.

DJI is fighting the ban on multiple fronts: a January 2026 petition arguing the FCC violated due process by skipping a standalone security audit, and appeals in two federal circuits. DJI has also ramped up its federal lobbying, spending $2.2 million in 2025 alone to argue that American consumers depend on DJI for cheap, essential drone technology. In April 2026, the Pentagon filed comments opposing DJI's petition, citing an interagency finding that DJI poses an "unacceptable risk to the national security of the United States".

Even as the ban pushes some buyers, especially government agencies, toward domestic manufacturers, no clear winner has emerged. US state and federal agencies have started adopting Skydio, but the 1.2x to 2.2x price premium poses challenges for local governments. The 2019 Huawei ban is a useful reference point: banning a Chinese manufacturer doesn't guarantee domestic dominance, it just routes demand to whoever's most competitive. In that scenario, foreign incumbents Samsung, Ericsson, and Nokia took over 5G, since no competitive American RAN provider existed.

We are working on a follow-up investigating the drone manufacturing and talent ecosystem in the US to answer the question of whether the US can build a DJI. If your work sits at the frontier of autonomy, power, or motors, we would love to talk. You can reach out to us at transmissions@arenaphysica.com.

A few perspectives worth reading today:

• "The U.S. military must employ a 'Hedge Strategy'... This 'high/low' approach enables the U.S. military to complement its arsenal of (high) expensive defense platforms of ships, tanks and planes with new capabilities such as small, inexpensive, AI-enabled and upgradeable unmanned systems (low)." Mike Brown, General Partner at Shield Capital and former Director at the DIU

• "'What Shenzhen knows' is not written down in a book whose covers can be sliced off and ingested into a model, it's a few million [manufacturing] stories stored in people, and a pace of knowledge-rich adjustment that keeps cost down and yield high. The only way to get it back is work with people who have brought up lines, to reshore and grow the ecosystem of people who have this knowledge, and to capture that knowledge as we build. Lots of people have clever ideas about the data capture/API piece, but we think the business of manufacturing may need equal portions pain tolerance, customer commitment, and cleverness on talent. America needs to solve the other half of reshoring." Sarah Guo, Founder & General Partner of Conviction Capital

• The team at Neros believes they can become the first American company to produce one million drones per year: “We’re ramping up as fast as we can to try to move to a world where, down to the unit level, somebody can, almost like Amazon, order ten Archers, 1,000 Archers, or 10,000 Archers, and we’ve got them ready to go. That’s sort of Nirvana.” Felix Pomerantz, Head of Finance at Neros