Selçuk Bayraktar and the coming autonomy ceiling

Selçuk Bayraktar, chairman and chief technology officer of Turkish drone-maker Baykar, used a 2 July 2026 stage appearance to argue that the human pilot is approaching his expiration date. The framing was deliberately provocative: AI-powered combat aircraft are not the next generation of fighter, he said, but "a different species" — closer kin to Deep Blue than to an F-16. Approximately 15,000 crewed fighters now in service worldwide will "probably" be replaced by unmanned platforms over a span he did not name. The remarks, distributed in clips by the Telegram channel Clash Report on 2 July 2026, are worth parsing carefully because Baykar has both an industrial reason to make them and a battlefield record to anchor them.

His claim deserves a serious test. Baykar's Bayraktar TB2 and Akıncı platforms have logged thousands of hours over Ukraine, the Sahel, and the Horn of Africa, and the company sits inside a small group of exporters whose hardware has actually shaped a modern shooting war. That gives his forecasts more weight than the usual drone-evangelist talk. It also makes them worth pushing back on.

What he actually said

Three claims stand out. First, the categorical one: AI-powered unmanned combat aircraft are not a sixth- or seventh-generation iteration of the crewed fighter. They are a different category of machine. Second, the quantitative one: about 15,000 crewed fighters worldwide are candidates for replacement once the technology matures. Third, the metaphor: comparing the drone-vs-pilot contest to Deep Blue's 1997 victory over Garry Kasparov is not nostalgia. In that framing, the human is the rules-bound opponent the machine has already surpassed in narrow, deterministic domains — and the contest is about extending that reach.

The method matters. Bayraktar was not arguing that pilots will be removed from the cockpit on day one. He was arguing that the constraint will invert: software, autonomy, and sensors will be the scarce inputs, not airframes and trained aviators.

Where the framing holds

Two parts of the argument rest on solid ground. The first is industrial. Production-grade autonomy — the kind that flies a combat mission without a human in the loop — is advancing on a curve that resembles the early commercial-drone curve of 2015-2020: software improvements compounded by cheap sensors, with hardware getting cheaper almost as a byproduct. Baykar's own factory output, and the order books of its few peers, support the claim that unmanned platforms are scaling faster than crewed replacements.

The second is operational. Crewed fighter fleets are aging. The Eurofighter, F-16, and early F-15A fleets are well past their design service lives, and successor programs in the West — Tempest, NGAD, the French SCAF — are running into cost overruns and allied procurement friction. Even a credible unmanned entrant is a competitive offer to air forces staring at decade-long gaps.

Where the framing strains

Two parts do not. The Deep Blue analogy smuggles in a 1990s assumption: that the contested space is closed and rules-defined. Air combat is not chess. Adversaries adapt, missions are networked across sensors and platforms, and the bottleneck is rarely "the pilot cannot compute fast enough." It is the pilot is one node in a system that includes AWACS, datalinks, jamming, decoys, and ground-based air defence. Removing the human from the cockpit does not remove him from the kill chain; it relocates him. Whether the relocated human is faster, slower, or simply more vulnerable is not predetermined.

The "approximately 15,000 fighters" figure also deserves scrutiny. Most air forces cannot afford to replace what they have. Roughly half of those airframes are operated by countries — Pakistan, Egypt, Nigeria, Indonesia, and a long tail of small fleets — whose defence budgets will not stretch to either fifth-generation crewed replacements or peer-class unmanned ones. The bottleneck is money, not technology, and Baykar's argument does not engage with it.

The structural read

What we are watching is the same pattern that compressed airpower into drones over the last decade: a category of weapon becomes cheap, software-defined, and exportable, and the Western lead in crewed platforms turns from an edge into a liability. The 1990s-era assumption that the West could absorb the cost of a $70m fighter and a ten-year training pipeline is no longer tenable for most buyers. Ankara, Guangzhou, and Tehran are not pitching sixth-generation fighters in the marketing materials their procurement officers actually read. They are pitching turnkey autonomy stacks.

If Bayraktar is right — even half-right — the export map for combat aviation is being redrawn faster than the procurement ministries of Washington, London, and Paris would like. NATO's combat-air plan through 2040 still assumes a crewed core. Baykar's customers, with or without that plan, will be fielding something else.

Stakes

For legacy defence primes, the forecast is existential: an unmanned pipeline at a quarter of the unit cost of a fifth-generation fighter implies a fraction of the sustainment revenue. For air forces on the wrong side of that arithmetic, it implies capability gaps. For Ukraine — whose drone programmes have rewritten air combat doctrine in real time — it implies validation. For Israel, whose combat-air philosophy is built on crewed F-35s and a handful of loyal wingmen, the practical question is whether the loyal wingman becomes the airframe.

What remains uncertain is whether the "different species" framing survives contact with the next contested airspace. Autonomy has a long history of working in uncluttered skies and faltering in the cluttered ones — and air defence is, by definition, a cluttering technology. Until an unmanned platform absorbs a contested strike package against an integrated IADS run by a peer opponent, the Deep Blue metaphor is a prediction, not a record.

This publication treats Bayraktar's remarks as an industrial forecast, not an endorsement; the technology is moving, but the operator who depends on the software still carries the risk.

Wire provenance

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

• https://t.me/ClashReport
• https://t.me/ClashReport
• https://t.me/ClashReport
• https://t.me/ClashReport