Turkey's AI-Powered Missiles: Precision and Peril

AI arrives on the missile battlefield

On 18 February 2026 Roketsan CEO Murat Ikinci told a packed hall at Boğaziçi University that turkish defense leverages enhance missile technologies with artificial intelligence to make weapons "smarter" — able to track, identify and discriminate targets with higher confidence. His remarks came amid visible momentum in Turkey's defence sector: Roketsan reported double‑digit export growth and is developing everything from ballistic and cruise missiles to the multilayered Steel Dome air‑defence architecture. Ikinci's claim reflects a wider trend: AI is now embedded across sensor suites, seekers and command networks, not only in research labs but in deployed systems being offered to international customers.

turkish defense leverages enhance missile guidance with AI

Roketsan and partner firms describe AI as a force multiplier for guidance: machine‑learning models clean noisy sensor returns, fuse electro‑optical, infrared and radar inputs, and provide robust target‑classification scores that can feed a missile's terminal seeker. In practice this means missiles equipped with onboard algorithms can better reject decoys, prioritise moving vs static signatures, and accept midcourse updates from a networked command node. Turkish systems makers also stress the economic logic: software updates and improved models can raise a legacy missile's effectiveness without a full hardware redesign, shortening development cycles and supporting export competitiveness.

That software‑first approach is already visible in integrated projects such as the national Steel Dome architecture, where companies like ASELSAN are adding electronic‑warfare and high‑power microwave layers to complement kinetic interceptors. These non‑kinetic tools — and the AI that coordinates them — are intended to shift cost curves when countering cheap drone swarms or low‑cost loitering munitions discussed in recent defence analyses. For buyers this mix of sensors, compute and weapons creates flexible, layered packages rather than single‑purpose interceptors.

turkish defense leverages enhance targeting, AI and the kill-chain

Artificial intelligence shortens the sensor‑to‑shooter loop in multiple ways. At brigade and theatre level, AI systems sift high volumes of imagery and cue human analysts or automated targeting modules; at missile level, onboard models perform rapid image recognition and aimpoint refinement in the last seconds. Studies of recent conflicts show that cheap drones and loitering munitions have altered the "kill‑chain": ubiquitous ISR and fast effects compress decision timelines and force defenders to automate aspects of engagement. Turkey's industry is adapting to that environment by integrating target‑classification networks with missile guidance and national command nodes.

But the same CEPA and NATO analyses that celebrate speed also warn of limits: AI classification still has non‑trivial error rates, sensors degrade in high‑EM‑noise environments, and adversaries can use deception or adversarial inputs to mislead models. That argues for architectures that keep humans in critical decision loops for high‑consequence strikes, and for interoperability layers so allied forces can share cues and avoid fratricide — a hard practical problem when export, national policy and standards diverge.

Steel Dome, Ejderha and the new non‑kinetic layer

Turkey is not only fitting AI into warheads and seekers; it is combining electronic attack, directed energy prototypes and command‑and‑control automation. ASELSAN's Ejderha and other microwave‑based counters are being positioned as short‑range, low‑collateral solutions against swarms, while systems branded under the Steel Dome project aim to network sensors, shooters and electronic‑warfare nodes under an AI‑assisted management fabric. Proponents argue this reduces expensive interceptor expenditure against cheap threats and creates corridors for friendly unmanned systems to operate.

Operational experience — and public demonstrations — remain limited, and defence analysts emphasise that non‑kinetic tools have performance envelopes highly dependent on environment and range. Yet combining multiple layers with AI orchestration is a pragmatic path for countries that must protect dense urban centres and critical infrastructure while keeping logistics manageable.

Benefits: accuracy, resilience and export opportunity

Strategically, an indigenous software stack and data‑processing capability reduce dependence on foreign suppliers. The success of projects such as TÜRKSAT 6A and a growing constellation of domestic smallsat firms shows a broader ambition: integrate space, ISR and weapons development into a cohesive national capability that can be marketed to partners — particularly in regions where Western suppliers are constrained by export controls.

Risks: ethics, legal frameworks and adversarial warfare

AI‑enabled missiles raise well‑known ethical and legal questions. Where an autonomous system can identify and engage a target without timely human oversight, concerns about accountability over lethal force escalate. International law requires meaningful human control over targeting decisions; many governments and analysts therefore urge architectures that ensure a human on‑the‑loop for lethal engagements. Turkey's public statements emphasise discrimination and precision, but detailed rules of engagement, audit logs and fail‑safe designs are rarely publicised for obvious security reasons.

What NATO and Europe must reckon with

European projects such as the European Sky Shield Initiative aim to harmonise missile defence across many states, yet political friction persists over procurement choices, national industrial priorities and dependence on non‑European technologies. Turkey's entry into pan‑European plans offers operational advantages — geographic coverage, indigenous medium‑range systems and industrial capacity — but also raises interoperability questions. Allies will have to reconcile differences in doctrine, data formats and rules of engagement, and decide how much trust to place in partner software and shared networks.

CEPA and allied technical studies recommend NATO invest in federated data infrastructure, validated AI toolchains, and joint testbeds to harden algorithms against spoofing and to certify human‑in‑the‑loop safeguards. Those measures, combined with political agreements over procurement and information sharing, will determine whether mixed NATO‑Turkish architectures improve collective resilience or simply add complexity.

Export, strategy and the blurred border with space launch

Turkey's defence momentum is entwined with its wider aerospace ambitions: satellite production, a planned equatorial launch site and dual‑use rocket expertise. That convergence matters because an orbital launcher and a long‑range missile share core technologies. For partners, the diplomatic calculus becomes more complex: cooperation may accelerate capability development, but it also requires transparency to avoid unintended escalation or proliferation concerns.

For Ankara, the commercial prize is clear — a software‑centric missile and air‑defence offering sells better in many international markets — but this commercial imperative collides with NATO's need for standardisation and allied states' political sensitivities about technology transfer and export control regimes.

Where this is headed

Turkish defence firms are putting AI into the field at a time when sensors, compute and networks are cheapening fast. That combination accelerates the pace of capability change and forces allies to act on doctrine, training and legal frameworks as much as on hardware. Practical steps include: agreed NATO standards for data and target metadata, certification regimes for AI modules in weapons systems, robust human‑on‑the‑loop safeguards, and multinational testbeds for resilience against EW and adversarial tactics.

If those steps are taken, AI can genuinely improve precision and reduce collateral harm. If they are not, speed and autonomy risk producing brittle systems that fail under electronic attack or produce tragic mistakes. The choice facing NATO and partners is therefore not whether AI will be used in missiles and air defence — it already is — but how governance, interoperability and collective resilience will shape the technology's operational effect.

Sources

• Center for European Policy Analysis (CEPA) comprehensive report "An Urgent Matter of Drones"
• NATO – Alliance Ground Surveillance (AGS) and Alliance Persistent Surveillance from Space (APSS) initiatives
• TÜBİTAK Space Technologies Research Institute (TÜBİTAK UZAY)
• Turkish Space Agency (TUA) technical and programmatic materials