Seoul puts North Korea's nuclear stockpile at up to 120 warheads, well above the figure cited by Trump

South Korea's intelligence assessment, as reported by Reuters on 20 August 2026, puts the upper bound of North Korea's nuclear stockpile at roughly 120 warheads, a range that sits well above the publicly cited figure offered by President Trump and lands on the same day that Pyongyang fired ten short-range missiles towards its eastern waters. The salvo, reported by South Korean officials and relayed by the Telegram channel RNIntel, gives the numerical argument a kinetic backdrop.

The numbers are not academic. They sit at the centre of an intensifying debate inside the US-Republic of Korea alliance over how much latitude each side has to describe the threat, and how much of the deterrent burden Washington is willing to underwrite. Reuters's reporting frames the Seoul estimate at up to 120 warheads, far above the figure publicly cited by the US president. The mismatch is now on the public record in two different capitals at the same moment that Pyongyang is putting metal in the water.

The stockpile, on paper

Reuters, reporting on 20 August 2026, frames 120 warheads as the upper end of a range associated with the South Korean estimate. The cited Reuters item does not specify, within the material available to this article, the lower end of the range, the delivery systems that host the warheads, or whether the briefing addressed fissile-material production rates, tactical-nuclear posture, or the readiness of mobile launchers. Reuters also does not, in the cited item, name the specific South Korean agency that delivered the assessment to lawmakers. The 120 figure therefore travels as a headline number ahead of its methodology, and what it does is re-anchor the conversation: if the upper bound is 120, policy planning built around a smaller number is incoherent by definition.

The cited Reuters headline also notes that the Seoul estimate is far above Trump's figure, without specifying that figure in the item this article is able to source. The Trump administration figure is therefore reported here only as something the cited Reuters item references as the lower end of an ally-versus-ally gap, not as a number this article can independently verify against the source material provided. Monexus analysis: when two allies publish materially different counts for the same arsenal on the same day, the gap itself is the news, regardless of which estimate proves closer to the ground truth. The political signal is what travels.

Missiles into the sea, signals into the air

Concurrent with the stockpile reporting, South Korean officials reported ten short-range missiles launched towards the eastern waters on 20 August 2026, according to RNIntel's Telegram relay. RNIntel's reporting does not specify, within the cited item, the launch site, the airframe, or the splashdown points. The brief Telegram post says only that the launches were reported by South Korean officials and characterises them as short-range missiles; it does not on its own confirm the airframe class, the timing, or the impact zones. Independent reporting from other outlets has supplied additional location and timing detail; the cited RNIntel item does not contain those specifics on its own, and this article does not import them.

The pairing is the story. The legislative briefing carries the bomb count; the launches telegraph what Pyongyang is willing to demonstrate when the count is being argued over. The optics inside the Blue House and the Pentagon are not lost on either side.

What the drills cut, and what they don't

The stockpile question is not separate from the exercise programme. CNBC reported on 20 August 2026 that the Trump administration is scaling back the scale of Joint Combined Exercise Program (JCEP) drills with South Korea, a move that the outlet's analysis frames as raising questions about the credibility of Washington's commitment to its alliances. The drills are the visible operational manifestation of the US nuclear umbrella. Cutting them at a moment when Seoul's own intelligence assessment is publicly revaluing the threat upwards is not a neutral scheduling choice. It changes the cost-benefit calculation in Pyongyang's missile programme by shortening the distance between a North Korean test and a US response.

Monexus analysis: deterrence is a function of capability and resolve, both of which are priced in real time by the adversary. When the visible US footprint at the alliance's most exposed seam shrinks, the price Pyongyang attaches to a salvo falls. The Seoul estimate's upper bound becomes the variable the alliance is being asked to defend with a smaller drills footprint. That is the credibility question CNBC is naming on the page.

Stakes, and what the next 72 hours look like

The operational question for the next several days is whether the cut in JCEP scope is a budgeting decision, a political signal to Pyongyang in advance of a possible summit, or a down-payment on a broader rebalancing that the administration has not publicly named. The sources cited in this article do not specify the administration's motive. The Seoul-Washington gap on the warhead count, if it persists into the working week, will force a choice between publicly reconciling the figures or letting the disagreement harden into a known unknown.

The Korean Peninsula is absorbing two pressures at once. Monexus analysis: the public-facing posture of each capital shapes the domestic politics of the other, and the gap travels faster than the underlying intelligence does. The upper bound of 120 will now travel as a number Seoul can point to, and a number Washington will have to either ratify or rebut. The two-track messaging, durable for a quarter, runs into the wall of an actual missile salvo faster than the alliance has adjusted for in the recent past.

The available source items do not specify the lower bound of the Seoul estimate, the precise JCEP reduction figure, the Trump administration's publicly cited warhead number, or the timeline for any further exercise cuts. They do specify that the warhead count and the drill tasking are now on the same page in the same week, and that the alliance is being asked to hold both at once.

Desk note: Monexus treats the Seoul-Washington gap as the story, not the upper bound in isolation. The wire lead in Western outlets is the missile salvo; the structural fact is the gap between the two numerical claims about the same arsenal, on the same day, while the exercise that operationalises the alliance is being cut.

Wire provenance

This editorial synthesis draws on the following public wire/social posts:

• https://reut.rs/4zs2MmF
• https://x.com/Reuters/status/2090383366717591878
• https://t.me/rnintel/65667
• https://www.cnbc.com/2026/08/20/trump-south-korea-military-drills-cuts-.html