Seoul's Ministry of SMEs and Startups, Defense Ministry, and the Korea AeroSpace Administration unveiled a 'new security' plan — modeling it on US defense data firm Palantir and its battlefield AI platform Gotham — with a one year rapid procurement lane as the policy's make or…
South Korea wants five defense-AI unicorns by 2030. The number that matters more is one year. That is the target window for the government's new rapid-procurement track, designed to take a civilian AI or advanced-tech proposal from contract to first deployment inside twelve months. Without that clock running on time, the rest of today's "new-security" industrial policy is a press release without an engine.
The Ministry of SMEs and Startups, Ministry of National Defense, and the Korea AeroSpace Administration (KASA) unveiled the "Direction for Nurturing Future New-Security Innovative Companies" at a strategy meeting today at Chungmu Hall inside Cheong Wa Dae, the presidential office compound in Seoul. The plan sets two quantitative goals for 2030: five "new-security innovative companies" valued at 1 trillion won (about $647 million) or more, and 50 firms with annual sales above 100 billion won. The target sectors are seven: drones and robotics, defense AI and semiconductors, defense sensors and advanced materials, space and aerospace, and cybersecurity and quantum communications.
The explicit model is Palantir, the US defense-data and AI firm whose Gotham platform ingests battlefield and intelligence data for the US Department of Defense and its allies. Seoul cited Palantir's roughly $300 billion market capitalization as the benchmark. The framing is explicit: Korea is not partnering with Palantir but modeling a domestic version, a "K-Palantir," on its commercial structure, in which a private software firm sits between raw military data and operational decisions.
The procurement redesign is where the targets either harden or dissolve. The new scheme adds a "rapid procurement" lane promising first deployment of AI and advanced-tech products within one year of contract, against a typical multi-year Korean defense acquisition cycle. It expands a "civilian-proposed military acquisition" track that lets companies pitch operational requirements rather than respond to fixed specs, and creates a non-defense "innovation-promoting contract" with milestone payments and partial liability relief. R&D funding is being reshaped, with up to 10 billion won per company over five years linked to demonstration and purchase.
The capital stack is built around a "Korea-type In-Q-Tel," a state-funded investment vehicle modeled on the CIA's venture arm, which backs US national-security startups. A 1 trillion won mother fund, combined with the defense fund, is meant to scale into a roughly 10 trillion won investment pool.
Behind the capital sits an open-data layer: a "Defense Data Catalog" sharing military datasets, expanded AI training data, "Defense AI Transformation (AX)" hubs, a proposed military-specific AI operating system under development (tentatively designated K-Maven), and dedicated defense AI and "defense world" models. Demonstration infrastructure is also expanding: defense demonstration units to grow to nine by year-end, an exercise called "Republic of Korea Drone Air-Land Warfare," and training ranges open to private firms. The workforce side includes a 500,000-strong "drone warriors" plan and 60,000 commercial training drones.
The feasibility question is what the procurement track runs into. The same Korean defense primes the policy is meant to crowd in (Hanwha Aerospace, Hyundai Rotem, Korea Aerospace Industries, and LIG Nex1) — according to a seoulz.com compilation of DAPA and defense-company investor disclosures — held roughly $69 billion in undelivered backlog as of late 2024, an indicator of how slowly the existing pipeline converts contracts into delivered systems. The Defense Acquisition Program Administration (DAPA) opened a new innovation cluster cycle in early 2026, adding more entrants to a procurement system that has historically favored incumbents. The risk the policy does not name is that "national champion" industrial policy tends to recycle the same four or five primes rather than mint new ones.
What to watch next: whether the prime minister's newly announced "New-Security Innovative Company Nurturing Committee" produces a special act and a National Contract Act amendment that actually shorten acquisition timelines, whether the In-Q-Tel-style fund closes its first checks in 2026, and whether the first wave of "civilian-proposed" contracts hits the one-year deployment mark without slipping into the same multi-year cadence the policy was designed to escape.