DroneShield: Anatomy of the ASX's Fastest-Growing Short Position

Twelve months ago, DroneShield was an afterthought on the short-selling board. In August 2025 just 2.60% of its register was held short — unremarkable for a small-cap defence technology company, and well below the level at which anyone pays attention.

As at ASIC's most recent report, 14.64% of DroneShield is held short. That is 135,319,820 shares out of 924,090,750 on issue, and it makes DRO the second-most-shorted stock on the ASX, behind only Lotus Resources at 17.19%.

Three weeks earlier, on 31 July, it touched 16.03% — the highest reading in DroneShield's entire history on our records, which run back to 2017. It has been drifting down since.

That is a near-sixfold increase at the peak. This is what it looks like plotted month by month.

What this data is, and when it is from — ASIC publishes aggregated net short positions on a T+4 basis. The most recent reliable figures here run to 17 August 2026. Every percentage below is net short positions as a share of total product on issue — the same basis ASIC reports, and the same one used across this site. Position data tells you what was done, never why; see the closing section.

The build, month by month

Read that column of changes and the position stops looking like a single trade. There are three distinct phases.

Phase one: the step-change (September–December 2025)

The position roughly doubles in September 2025 — 2.60% to 5.14% — dips, then climbs to 8.62% in November and 10.89% by December.

That is a fourfold increase in four months. Short interest does not move like that by accident, and it does not move like that on retail flow. Building to 10% of a register requires borrow availability, and borrow at that scale comes from institutions.

Phase two: the flush (January 2026)

December's 10.89% falls to 8.90% in January — a −1.99pp reduction, the largest of the build until August.

A drop that size is either profit-taking or a squeeze. What the data cannot tell you is which, and the honest answer is that ASIC's aggregate does not distinguish between a bear closing a winning trade and a bear being forced out of a losing one. Both look identical in this column.

Phase three: the grind (February–August 2026)

From 8.90% in January, the position rebuilds steadily — 9.45%, 11.76%, then a plateau through April and May, then 12.04% and a jump to 16.03% in July.

The character here is different from phase one. Phase one was a step-change; phase three is accumulation, with two pauses in it. Someone has been adding patiently for six months.

Where that sits in context

Only one ASX-listed company currently carries a larger short position:

The company keeps unusual company. Most of the names above it and around it are resources (Lotus, Paladin, Boss, Lynas, Genesis) or consumer discretionary businesses with well-worn bear cases (Domino's, Treasury Wine, Flight Centre). DroneShield sits in Capital Goods, and the bear case on a counter-drone and electronic-warfare company is a different animal entirely — it is a bet about contract timing and valuation, not about a commodity price or a same-store sales number.

What the data does not tell you

This is the part most short-interest commentary skips, so it is worth being blunt about.

A large short position is not a prediction. It is a disclosure. ASIC requires it, the shorts comply, and we publish it. It tells you that a meaningful group of institutions has taken a position against the stock and — importantly — that they have kept it on through a near-sixfold build and at least one flush.

It does not tell you why. No part of this dataset contains a thesis. Anyone telling you precisely what the DroneShield bears believe is inferring it, and inference is not data.

Some of it may not be directional at all. Short positions include hedging and arbitrage. A convertible-note holder hedging their exposure appears in exactly the same column as a bear betting on a fall. There is no way to separate them in ASIC's aggregate.

And crowding cuts both ways. A 14.64% short position is simultaneously the bear case's strongest expression and the largest pool of forced buyers that exists if the thesis breaks. That is the mechanical reality behind every short squeeze: the bigger the position, the more buying is required to close it.

What to watch

Three things in this dataset are worth tracking, and all three are free to check:

1. Whether 16.03% holds as the ceiling. The 31 July print is an all-time high, and the position has come off it every week since. A decisive break above says the accumulation phase is not finished.
2. The steady unwind off the high. From 16.03% on 31 July the position has fallen almost every week — 15.65% on 10 August, 14.98% on the 17th, 14.64% on the 18th. That is the first sustained reduction of the entire build. December's then-record 10.89% was given back to 8.90% in January and the trend resumed, so a fall is not by itself the end of it.
3. Whether the resources names above and below it move together. If Lotus, Paladin and Boss all de-short at once and DroneShield does not, the DRO position is idiosyncratic rather than part of a broader risk-off rotation.

You can follow the position on the DroneShield page, see where it sits on the most-shorted board, and compare it against any other ASX stock on our comparison tool.

Not investment advice — This is a description of publicly disclosed short-position data, not a recommendation. A high short interest is not a signal to buy or to sell — it is a fact about positioning, and positioning can stay wrong, or right, for a very long time.

Short-position data sourced from ASIC's aggregated short position reports, updated daily. Figures as at 18 August 2026.