Nuke-Cut Territory Annexation (v35)

Short answer. Dropping a nuke that slices enemy territory into a piece no longer connected to their main body can hand that cut-off piece to a surrounding enemy, instead of the owner keeping it. It requires an eligible severed inland pocket surrounded by other players and fallout, with an eligible enemy recipient. This behavior is in stable v0.35.0 through v0.35.2; you no longer need a beta build. The older v0.34.25 does not contain this capture path.

The rule in one picture

Think of an enemy’s territory as a single blob of land. The main body is the largest connected cluster by land area, not by border length. Now you drop a nuke. The crater erases a ring of their tiles, and the fallout that follows is an unowned boundary accepted by the annexation check. It is not your land and gives you no border advantage. If the remaining land is split in two, one side stays attached to the main body and the other becomes a free-floating island of their land with no path back to it. That floating piece is what the engine calls the severed cluster, and the new rule decides what happens to it. Before v0.35, the engine’s old enclosure logic rarely treated a nuke-cut arm of a coastal empire as annexable, which is exactly why players complained for a long time that nuking to cut land off “looked random” and “stopped working for coastal empires.” Pull request #5751 reworked the capture path so that a genuinely cut-off, landlocked piece is now claimed by the enemy that surrounds it, instead of being left to the original owner.

The decision this unlocks is a real return-on-investment call, not a passive bonus. A single nuke that would only destroy a few buildings can now also transfer an entire district, sometimes dozens of production tiles, to a third faction or to you. That changes which strikes are worth the gold. A strike aimed at the thickest part of an enemy’s territory may cut nothing at all, while a strike that lands on the narrow isthmus connecting their coast to their inland heart can both crater that connection and detach a chunk that a bordering enemy then swallows. You are no longer just buying damage with the nuke; you are buying a territorial transfer on top of the damage. That is why the feature matters even though it only triggers in a narrow set of board geometries: the geometry is precisely the geometry a thoughtful endgame nuke already tries to hit. This is also why the feature changed how experienced players evaluate a nuclear strike in the first place. The nuke is no longer a pure damage purchase where you subtract the crater cost from the buildings destroyed. It is now a compound investment, where the same gold can simultaneously buy the crater and the transfer, and the transfer is usually the larger of the two. Because the payoff is conditional and geometry-dependent, the difference between a winning strike and a dud is no longer your aim alone but your read of the board geometry. The rest of this guide walks through the exact conditions, the capture logic, the version boundary, and how to read the map so you can set up the cut deliberately instead of discovering it by accident, and it ends with a quick reference table you can run before every strike.

The three conditions that must all hold

The engine does not annex a cut-off piece just because it looks disconnected on the map. Three independent checks must all pass at the same time, and if any one of them fails the piece is left entirely alone and the owner keeps it. The first condition is disconnection from the main body. The cluster must not be connected, through the owner’s own border tiles, to their largest landmass. A piece that still touches the main body by even a single tile is not severed and is ignored by the annex path. This is why a nuke that merely damages the edge of a blob, without fully clearing the connecting isthmus, does nothing at all: the connection survives, so nothing is actually cut off. The second condition is that the piece has no coast and no map edge on its own. If the cut-off piece touches open water or the border of the map by itself, the owner can simply boat out of it or treat it as a self-sustaining enclave, so the engine refuses to annex it. This is the single most common reason a cut appears to “not work” in real games: the detached arm still has its own coastline. The third condition is that the piece borders only other players and fallout, and that at least one enemy and at least one fallout tile surround it. The piece must be fully ringed by fallout, which is an unowned boundary, and by other players’ territory, with no ordinary unclaimed tiles touching it. If any tile of the piece opens onto empty no-man’s-land, the annex is skipped, because there is no surrounding enemy that could claim it.

The practical takeaway is that you need a landlocked pocket of enemy land, fully surrounded by fallout and other players’ borders, that has completely lost its road home. A nuke that carves a ring of fallout around a district while the district’s only land bridge back to the owner sits inside that ring is the textbook setup that satisfies all three conditions at once. Note the order in which players usually fail these checks. First they get the disconnection right but the piece still has a coast. Then they handle the coast but the piece still touches the main body by a single tile. Then they clear both of those but a gap of open land breaks the “borders only players and fallout” rule. Each failure mode is visible on the map before you spend a single unit of gold, which is what makes the whole rule plan-able rather than lucky. The failures section later breaks down each of these with concrete counterplay and the situations where you should avoid the strike entirely, but the core idea is that the rule is precise and deterministic, and a single failed condition means the strike is crater damage and nothing more.

Which enemy actually captures the piece

When the pocket qualifies, start with the players whose actual territory borders it. Candidates friendly to its owner are excluded. The relationship to the player firing the nuke does not add another exclusion: your ally can still qualify if that player is hostile to the owner. Fallout does not count as your territory, so creating a large radioactive boundary gives you no extra votes. A candidate needs real land beside the pocket. Moving a ship nearby or merely placing troops somewhere on the screen does not change the ownership of those neighboring tiles.

The first priority is the largest single ongoing attack against the owner among these bordering candidates. Compare individual attacks, not the sum of all attacks by each player. A neighboring enemy with one larger attack can receive the pocket even when another candidate has more border. An attacker elsewhere on the map does not qualify just by sending a huge force: that player still needs to be one of the pocket’s bordering candidates. This is why checking only the outline before firing can be costly. The fighting already in progress is part of the recipient decision, and its state may change before the transfer is evaluated.

Only when none of those candidates has a qualifying ongoing attack does the boundary comparison decide. The count is shared four-direction contact edges, not a count of distinct pocket tiles. One pocket tile can contribute more than one edge to the same neighbor. In a hypothetical case with no ongoing attacks, fourteen contact edges beat eleven. Do not use that example when an eligible attack exists. Equal highest values retain the candidate encountered first internally; neither a tied attack nor a tied border is a dependable way to promise yourself the land.

For the remaining examples, any conclusion based solely on border counts assumes that no candidate has an ongoing attack against the owner. If that assumption stops being true, repeat the recipient check using attack priority before trusting the example. A strike that meets every geographical condition can still hand the district to a third player. The person paying for the missile and the person collecting the land are separate roles, so include both in your decision. If you cannot confidently identify the recipient, value the shot for its destruction alone.

There is also a settlement distinction. If the pocket is all the owner’s land at the instant it transfers, the recipient conquers that player. This can happen when other territory was taken earlier in the same annexation pass. It does not mean that any lone capital can be cut away from itself, or that conquering one player wins the entire match. First establish an eligible severed pocket, then its recipient, then whether any of the owner’s land remains. Keep these questions separate before counting a possible elimination as your own finishing move.

Version boundary: stable v35 and older v34 games

Nuke-cut annexation is now part of a stable release. v0.35.0 was published on 2026-10-16, and the behavior remains in v0.35.2, published on 2026-10-17. You do not need to choose a beta server to use it in a game running these versions. The practical change is that a deliberate cut can now be evaluated for both crater damage and the land a surrounding enemy may receive. Treat those as separate returns: the damage can still be useful when the transfer fails, while a transfer to the wrong rival can make an otherwise accurate strike a bad investment.

The older v0.34.25 does not contain this particular capture path. That historical boundary still matters when comparing old recordings, older replays, or a custom server deliberately running an earlier build. Do not expect an old game to reproduce the stable v35 result simply because the visible shape looks the same. Equally, do not explain a failed cut in a current v35 game by saying that annexation is beta-only. First establish which version the game uses, then examine the geometry and the beneficiary. A release date identifies when a build became available; it does not establish the build of every custom server or every recorded match.

For a v35 game, start with the connection you intend to remove. A thin-looking strip can still leave a surviving route around the crater, and an isolated-looking district can retain its own coastline. The release does not turn every nuclear strike into automatic annexation. The piece must meet the conditions described above, and you still need to identify the surrounding enemy that receives it. Check that player before counting the expected land toward your own position. If a rival surrounds the district more effectively, the strike can advance their finish rather than yours. Waiting for a better border can be more valuable than firing as soon as you can.

When reviewing a failed attempt, separate version, geometry, and reward. An older build can explain why this capture path is unavailable. In a supported v35 build, inspect the surviving connection, water or map edge, and any unclaimed gap before changing your aim or buying another missile. If the transfer happened but helped someone else, changing versions will not solve the problem: the next decision is about who surrounds the target. This order also makes old videos useful without treating their outcome as a promise for your present match.

Use the official v0.35.2 release notes for the published v35 boundary. They include the v0.35.0 change restoring annexation of territory cut away by a nuke. For actual play, keep the decision simple: supported build, complete cut, eligible pocket, intended recipient, then cost. Stable availability removes the need to seek a playtest; it leaves every geographical and strategic check in place.

Reading the map before you commit

Turning the three conditions into a pre-strike checklist makes the difference between an accidental bonus and a reliable one, and each check is a simple yes or no read you can do before spending gold. First, find the isthmus: the narrowest band of the owner’s land that connects the target district back to their main body. Check the owner’s actual surviving land after the expected blast; fallout is an unowned boundary accepted by this check, not your land or an extra friendly radius. It gives you no contact edges. Do not infer a severed connection from a radioactive color alone. If a land route survives the strike, disconnection fails and this path does not transfer the pocket. You need the actual connection removed, rather than a larger-looking visual effect, so skip that target and look elsewhere. Second, check the coast on the cut-off side. Trace the outline of the district you are trying to detach and confirm it has no water and no map border on its own. If the district you want to gift to a rival has its own shoreline, it is unannexable no matter how cleanly you cut it, and you are far better off targeting the district’s landlocked inner half instead of the coastal shell. Third, verify the surround. The detached piece must be ringed by fallout and other players’ land, with an eligible enemy candidate; not every player neighbor has to be hostile. There must be zero ordinary unclaimed tiles touching it. Any gap of open no-man’s-land breaks the “borders only players and fallout” rule and kills the annex, so the cut needs to sit deep inside enemy territory, not on the fringe of unclaimed land where a stray gap is likely.

Do all three checks before you spend a single unit of gold, because a single no means the strike is crater damage only and the transfer is impossible. Then, and only then, do the fourth check, which is the recipient check that decides who benefits. Check ongoing attacks first, then contact edges if no candidate is attacking, and confirm that the recipient is you rather than a rival. If a third faction would collect the land, the strike is a gift to them and you should either reposition to gain border first or pass entirely. The two scenario sections that follow apply these exact reads to two concrete board states, so you can see the thresholds, the tile counts, and the positions and costs that separate a useful transfer from a poor investment. The skill is: read the isthmus, read the coast, read the surround, count the border, and only then decide whether the crater damage is worth the territorial transfer, with the version boundary as the zeroth check that gates everything else in this entire guide.

Scenario A: cutting a coastal arm off a mid-game empire

Set the board for a concrete mid-game case. Three factions on a standard island map. You are at about sixty percent population. The target enemy owns a long coastal strip on the east side that connects to their inland capital by a two-tile-wide isthmus. Their east strip has its own coastline, which is the trap that catches most players. If you nuke the isthmus and detach the whole east strip, the strip still touches the sea, so the “no coast” condition fails and the owner keeps every tile of it. That is the mistake to avoid. Instead, look at the inland pocket behind the isthmus. It is a cluster of roughly eighteen farmland tiles that the east strip protects, and it is fully landlocked. It is bordered on its north and south by your own territory and on its west by the isthmus itself. A MIRV that bursts directly on that isthmus produces a crater and a fallout ring that is wide enough to clear the two-tile connection. After the detonation the inland pocket is disconnected from the main body, has no coast of its own, and is ringed by unowned fallout and the territory of the factions around it, so the three conditions all hold at once.

Now count the actual value. Eighteen tiles do not imply nine gold per tick: there is no fixed half-gold-per-tile income rule. These are hypothetical map dimensions, not an income formula or a guarantee that a MIRV’s impacts will clear the connection. Value the position, access routes, and any buildings that actually survive separately from the crater damage. Assume for this example that no bordering candidate has an ongoing attack. If you have the most contact edges, you receive the pocket. If the northern candidate has more, that player receives it. If a qualifying attack starts, attack priority replaces this border-only comparison. The decision is therefore conditional. Even after confirming the recipient, compare the position and cost before firing; receiving land alone does not prove a net win. If you cannot confirm the recipient, the strike buys the north player an eighteen-tile district, and your only sane options are to reposition so that you border the pocket first, or to pass on the strike this round and wait for a better geometry. This is the whole skill compressed into one example: read the isthmus, read the coast, read the surround, count the tiles, and only then decide whether the crater damage is worth the transfer. The numbers, not the drama of the crater, are what make the decision.

Scenario B: a severed pocket becomes the last land

Consider a hypothetical endgame with you, a rival, and a weakened target. Before the annexation pass, the target still has a main body and a separate eligible inland pocket of twenty-five tiles. The pocket has lost its connection, touches neither water nor map edge, and is surrounded by other players and fallout without an ordinary unclaimed gap. During that same pass, the main body is taken through enclosure. When the severed pocket is evaluated for transfer, it has therefore become the target’s last remaining territory. The order is essential: it was identified relative to another main body, rather than being a lone capital somehow disconnected from itself.

Now assume neither bordering candidate has an ongoing attack against the target. You share fourteen four-direction contact edges with the pocket and the rival shares eleven. These are hypothetical contact counts, not fourteen different pocket tiles versus eleven different tiles. Under this explicit no-attack assumption, your larger count makes you the recipient. If the numbers are reversed, the rival receives it. If the rival instead has the largest qualifying ongoing attack at evaluation time, the rival takes priority despite your fourteen edges. Recheck the battle state before treating a promising boundary as a guaranteed reward.

Because the transfer now covers all of the target’s remaining land, the recipient conquers the target rather than merely taking one district from a still-living player. That is a player elimination, not a promise of victory in the whole match. The rival in this example still exists after the target is gone, and the rest of the board can remain dangerous. If the main body had not already been lost, the same pocket transfer would leave the target alive elsewhere. Likewise, another surviving outpost would prevent this transfer from being the last-land case. Do not decide that a visible capital is the entire player without checking.

For your strike decision, compare the board after both possible recipients. If you receive the pocket, does the new border expose your core to the surviving rival? If the rival receives it, does removing the target release pressure on that rival’s other front? Destruction can still be useful, but do not pay for it under the assumption that the conquest belongs to you. When the recipient is uncertain, postponing the strike can preserve a shared front and give you time to change the actual territory boundary. Extra nearby troops without a land transfer do not increase your contact count.

The useful lesson is the sequence, not the twenty-five-tile figure: establish a qualifying pocket, account for other land lost in the same pass, identify the recipient by attack priority or its border fallback, and only then assess elimination. A more dramatic crater cannot replace any of those checks. If even one premise differs in your match, recompute the outcome rather than copying the hypothetical conquest.

Failures and counterplay

Most failed annexes trace back to one of four failure modes, and each one is visible on the map before you spend a single unit of gold, so the goal is to catch the failure in the planning phase rather than after the strike. The first failure mode is that the connection survives. Your crater plus fallout did not fully clear the isthmus, so a single tile still links the piece back to the main body and the disconnection check fails. The counter is to aim the burst at the narrowest point of the isthmus and confirm that the actual land connection will be removed, without treating fallout as your own land or an extra friendly radius. Otherwise, look for a different target. The second failure mode is that the piece keeps a coast. You cut the district but it retains its own shoreline, so the no-coast condition fails and the owner keeps the land, still able to boat out of it. The counter is to target the landlocked inner half of the district rather than the coastal shell, or to accept that this particular district is unannexable and pick a different pocket that has no coast of its own.

The third failure mode is that open land breaks the surround. A gap of no-man’s-land touches the piece, so it does not border only players and fallout, and the annex is skipped entirely. The counter is to only cut districts that sit deep inside enemy territory, where the surround is complete and there is no fringe of unclaimed land that could leave a gap. A piece that sits on the edge of open land is a guaranteed dud. The fourth, and most expensive, failure mode is that you gift the land to the wrong player. All three conditions hold, the transfer fires, but the capturing player is the enemy selected by attack priority or the no-attack border fallback, so the land can go to your rival. The counter is to inspect both ongoing attacks and real contact edges before you commit. If you cannot establish that you are the recipient, either change the actual territory boundary or pass on the strike this round.

There is also a counterplay angle for the owner of the territory being cut, and it is worth knowing because it tells you when a player can safely refuse to set up the annex against them. An owner defends by never leaving a landlocked pocket fully surrounded. Keep at least one coastal exit on every district, which makes the no-coast condition fail for any attacker. Keep the isthmus to your main body wider than the expected strike can actually clear. Avoid an eligible pocket becoming your last land after other regions are lost. Conversely, if you are the owner and you are about to lose a district, the best response is usually to open a coast or a road back to your main body rather than to try to fight the crater, because a surviving connection blocks this severed-pocket path. Other attacks and enclosure can still threaten the land. A third faction can change its real territory boundary before the strike; merely moving units does not increase contact counts. A qualifying ongoing attack also takes priority over the border fallback. The annex is therefore a contest involving actual land and attacks, not an automatic reward for the person who fires. Recheck both when the board changes instead of relying on an earlier outline.

Mode and map adjustments

The annex behaves differently across the board states players actually meet, so the same three-condition read has to be adjusted to each one before you trust it. On standard island maps, which are the natural home of this feature, coasts are everywhere, which means most cut-off pieces have their own shoreline and fail the no-coast condition. That is why the annexes that do fire on standard maps are the rarer, higher-value landlocked pockets, and exactly why they reward a deliberate setup rather than an accidental bonus. On standard maps, prioritize isthmus targets deep inland where the surround is complete, and expect the capturing-player race to usually involve a third faction you have to account for before you fire. On maps or modes that include a large central lake or an inland sea, water can separate territory without creating an eligible inland pocket. There is no guaranteed increase in annexations simply because a map has more water. The no-coast check remains essential: a piece with its own lake shore is unannexable even though it looks completely isolated, so on those maps you should target the pocket that the lake does not touch.

In high-nuke or MIRV-heavy modes, evaluate the actual expected impact locations rather than treating a MIRV as a guaranteed wider single crater. Multiple explosions can remove different connections; they can also leave a route intact or destroy a pocket you intended to receive. If your mode allows MIRV, repeat the complete connection check for the expected pattern and retain uncertainty where you cannot predict the surviving land. Cost can be justified by destruction, territorial position, or both; annexation is not the only possible return. In endgames, identify whether the eligible pocket really becomes the owner’s last land during settlement, instead of assuming a lone capital always triggers conquest. A transfer can eliminate the target while leaving other rivals alive. Check attack priority before the border fallback, and decide whether removing that shared front helps you more than the rival. Across every mode the underlying read is identical: surviving connection, coast, surround, recipient, and remaining land. The frequency and the stakes of each condition shift with the map. Check whether receiving the pocket exposes a new front, and whether keeping the target alive still ties down a rival’s troops. Those questions remain relevant even when the annexation itself is certain. The numbers in the two scenario sections above should always be recomputed for the specific board you are actually on, not copied from a different map shape.

Quick reference table

Run this table top to bottom before every strike, and do not skip rows. If the version row is wrong you stop immediately, because this path is absent in older v0.34.25 builds and the rest of the checklist is wasted effort. If the disconnection row is wrong the owner keeps every tile, so the whole annex is impossible no matter how clean the crater looks. If the no-coast row is wrong the piece is unannexable by definition, and the only remedy is a different target. If the surround row is wrong the engine skips the annex outright, and there is no in-game way to close the gap after the strike. If the capturing-player row is wrong you have chosen the right geometry but the wrong beneficiary, which is the most expensive kind of mistake this rule can produce, because you spend the full gold and a rival collects the land. The only-land row is the one that decides whether the strike is a tile transfer or a full conquest, and it is the single highest-stakes row on the table because a wrong answer can end the entire game in the wrong direction.

The table separates eligibility, recipient, and settlement. Only the eligibility checks determine whether this annex can happen; the last-land row changes settlement. Check version, disconnection, coast, then surround, then capturing player, then only-land, because each earlier row is cheaper to check and eliminates more candidates than the ones after it. Checking the version before you do any tile math saves the entire rest of the exercise when you are on an unsupported older build. Checking disconnection before coast saves the coast check when the piece is not actually severed. The full derivation of each row is in the sections above, and the two scenario walks show the tile and gold numbers that turn a passing row into a real decision on your specific board. The table is deliberately short, because the goal is to run it in under a minute at the moment of decision, not to re-derive the rule under time pressure.

Check	What to confirm on the map	If it fails
Version	Verified stable v0.35.0–v0.35.2; v0.34.25 lacks this path	Assess the actual build; do not assume beta-only
Disconnection	Crater plus fallout fully clears the isthmus to the main body	Owner keeps the piece; re-aim at the narrowest isthmus
No own coast	Detached piece touches no water and no map edge	Unannexable; target the landlocked inner half
Complete surround	Other players and fallout surround it; no ordinary unclaimed gap; an eligible enemy exists	No eligible annex through this path
Capturing player	Largest single ongoing attack among bordering enemies; otherwise most contact edges	Recipient may be a rival; reassess
Last-land settlement	Is this all the owner’s remaining land at transfer time?	No: ordinary land transfer; yes: recipient conquers the target

For the general enclosure rules and their older v0.34.25 context, see Annexation and Enclosed Territory. For the gold math behind deciding whether the crater damage is worth a strike at all, see Nuke Gold Calculator. For how fallout tiles behave in the gap you are cutting, see No Man’s Land and Fallout. For multiple-warhead strikes and their limits, see MIRV.