Beijing rehearses the blockade it hasn't yet ordered

At 19:58 UTC on 16 July 2026, the prediction market Polymarket priced the probability of a Chinese blockade of Taiwan by 31 December at 6%. Five minutes earlier, the same platform flagged satellite imagery that, if accurate, depicts the People's Liberation Army building a full-scale replica of Taipei's presidential district to rehearse on. The gap between those two numbers is the story.

Beijing's military signalling has rarely been this literal. A replica capital, in the country that already owns the original, is the kind of prop that does not get built by accident. It is the kind of prop that gets built because someone, somewhere in the PLA's planning chain, wants the troops who will use it to feel what it would be like to operate inside the real thing: the sight lines, the choke points, the surrounding ministries that an occupation force would need to secure within hours of arrival. The drill is not the war, but the drill is now also not hypothetical.

What the imagery actually shows

The pictures circulating on X on 16 July, and tracked through the polymarket information feed, describe a structure in mainland China built to resemble the compound housing Taiwan's presidential office. The reporting does not specify the precise location of the replica, nor which PLA service is operating it, nor which formation is training there. What it does establish is that Beijing's military has chosen, at this moment, to invest in a physical stand-in for the seat of a government it does not recognise, in a city across a strait whose status it has never renounced changing by force.

The significance is the choice of subject. A replica airfield, a replica port, a replica command bunker would all be unremarkable; militaries replicate targets routinely. The choice to replicate the seat of government, the political symbol, not the military centre of gravity, is a signal aimed as much at Taipei and Washington as at the rehearsing troops. It tells the island's leadership that the operation under rehearsal is one in which they, personally, are the target.

The 6% the market is paying

Polymarket's 6% figure, posted on the same day, is the market's collective answer to a different question: how likely is any of this to become kinetic before the calendar turns? Six per cent is not nothing. It is roughly the implied odds of a long-shot political event, the kind of contract that exists because a non-trivial number of traders think it deserves to exist. It is also far below what one might expect if the world's professional risk-takers took a near-term cross-strait war as their base case.

The divergence between visible preparation and traded odds is not a contradiction. It is the gap between capability signalling and intention inference. Beijing can rehearse a blockade in July without ordering one in October, just as Washington can sail a carrier through the Taiwan Strait in any given month without intending to fight a war that month. Replicas are cheap; wars are not. The PLA builds the stand-in. The Politburo decides whether to use it.

The structural read

What the imagery sits inside is a wider pattern of escalation by rehearsal. Across 2025 and into 2026, China's military drills around Taiwan have grown in duration, in the number of services involved, and in the simulated targets selected. Replicating the presidential compound is the next escalation in a sequence that has otherwise been confined to sea and air, and it is also the first that has been conducted entirely on Chinese soil, far from the live fire zones around the island itself. That is a deliberate move. It places the activity beyond the reach of Taiwanese, Japanese, and US maritime surveillance in the way that a live--fire exercise in the Taiwan Strait would not be, and it does so while preserving the propaganda value of "we know what your capital looks like from the inside."

For Taipei, the read is unambiguous: assume the worst, prepare accordingly, and do not mistake the absence of a blockade for the absence of planning. For Washington, the read is more politically loaded, a US administration weighing whether to deploy additional forces to the western Pacific must now reckon with the possibility that the target list is more ambitious than previously assumed. For the rest of the region's capitals, the read is the one they have been quietly rehearsing in their own war games for years: a Taiwan contingency is not a bilateral problem, it is the central organising event of Indo-Pacific security.

What could break the read

Two pieces of news on the same day complicate the picture. A separate polymarket-flagged item at 12:04 UTC reported that Washington is pressing Beijing to release an American seismologist held for nearly two years on espionage allegations. That is a different channel of the bilateral relationship, judicial, consular, scientific, operating entirely outside the military one. The fact that the two conversations are running in parallel says something about how the Trump-era US-China engagement, such as it is, is being compartmentalised: the security track escalates, the consular track grinds.

The other wild card is the replica itself. Satellite imagery is real, but interpretation is not free. The same imagery, in the hands of a less alarmist reader, could read as a long-planned training facility that has only now been noticed, or as an internal PLA propaganda set-piece with no operational value. The market's 6% suggests that the trader class, at least, is not currently convinced that the replica is the precursor to a war this year. That is a defensible position. It is also the position one would be forced to take if one believed that the visible signalling and the political decision are separable, and that Beijing has reasons not to act that have nothing to do with military readiness.

The honest answer is that no one outside Zhongnanhai knows whether the replica is a rehearsal or a museum piece, and the next data points, whether the structure persists through the autumn, whether it appears in subsequent drills, whether Taiwan's intelligence chief publicly names it, will matter more than any satellite release. What is already true, and what the market has not yet fully priced, is that Beijing has decided the cost of being seen to prepare is lower than the cost of being seen not to.

This publication tracks cross-strait signalling through primary imagery and traded odds rather than through either wire alone, on the view that the gap between the two is itself the news.