“Now I am become Death, the destroyer of worlds.” J Robert Oppenheimer’s reflections on watching the Trinity nuclear test in 1945 marked a sea change in global conflict: weapons so overwhelming that serious conflict between two nuclear powers meant risking the very existence of both. Nothing came close.

Drones – so effective in Ukraine – may soon become a new kind of strategic global deterrent. In an optimistic scenario, they will dissuade not just nuclear powers from attacking each other but smaller ones too.

The UK-Ukraine deal announced this week offers a glimpse of what this could look like. It is not simply an agreement to buy or build more drones. Britain is gaining access to Ukraine’s battlefield data and AI expertise, while bringing its own technological and research capabilities to the partnership. In miniature, it is the kind of exchange that a future drone security architecture could be built around.

Such alliances will require enormous resources to both wage and defend against networked unmanned warfare. Only a handful of states will be able to do this at overwhelming scale, combining three elements essential to drone dominance: mass, networking and constant iteration.

First, mass. Huge numbers of drones are required to cover the battlespace – watching and quickly reacting along a border which may be thousands of kilometres long and tens of kilometres wide. Harder still, rapid attrition requires rapid replacement. Ukraine shows the evolution: in 2022 it produced around 3,000 drones; by 2026 it is on course to produce more than five million.

Second, networking. Mass becomes far more powerful when drones are connected and can collectively overwhelm defences. Earlier this year I visited HMS , a British area defence ship. It was impressive: one of the most advanced ships of its type, its radar can track 100 tennis-ball-sized objects to 250 miles. It has tens of missiles to shoot them down. But how would even an advanced warship deal with 200 objects – or 2,000?

And the third element, constant iteration. As I’ve written previously, Ukraine has repeatedly demonstrated the importance of learning from the battlefield to maintain an advantage, with leading figures explicitly drawing on Silicon Valley’s test-and-learn approach.

Doing all three requires deep economic power, industrial capacity, software, communications, intelligence, sensors, electronic warfare and the ability to update them all every few days or weeks. Only a handful of countries can build that at sufficient scale. For everyone else, economies of scale favour pooling resources rather than risking a sub-scale capability that could quickly be overwhelmed.

That naturally leads to “drone umbrellas”: a small number of industrially and technologically powerful countries that can provide the underlying capability for much larger groups of allies, creating competing drone security architectures centred on major powers such as the US, China and potentially Europe.

We can already see the beginnings of this architecture.

In Europe, plans for a “drone wall” from the Baltic states to the Black Sea have been accelerated by drone incursions from Vladimir Putin’s Russia. The European Drone Defence Initiative will bring together radars, sensors, jamming and weapons systems to detect incoming drones – then track and destroy them.

In the Middle East, US Central Command will lead a new drone attack capability. Task Force Falcon Strike will bring together US and regional militaries to deploy unmanned one-way attack systems from “above, on, and below the sea”, alongside counter-drone capabilities.

The recent UK-Ukraine deal goes further still. While Europe and the Middle East are pooling defence systems, the UK and Ukraine are pooling the critical underlying learning and technological ecosystem that drives improvement.

At scale, these security umbrellas could change deterrence itself. Nuclear weapons promise overwhelming force, but typically as a last resort: mutually assured destruction means even nuclear powers can tolerate significant grey-zone, and even conventional, conflict before considering the ultimate deterrent.

But a comprehensive drone network may offer deterrence from the first moments of a conflict. Unlike nuclear umbrellas, which deter through the threat of retaliation, drone umbrellas could deter by making the initial attack futile. I’ve written before about the two essential qualities of deterrence, capability and credibility. Drone umbrellas could offer both.

In the future, an attack risks being both ineffective and cripplingly expensive. An effective drone defence could create a kill-zone tens – potentially hundreds – of kilometres deep, destroying hundreds of millions of pounds of equipment before an attacker even reaches the battlefield.

A rapid, collective response is also far more probable. Drone umbrellas are highly deployable: sensors, software, operators, interceptors and attack drones can be integrated across borders without the huge logistics and supply chain required by a conventional army. Politically, the cost of sacrificing soldiers has become one of the great constraints on Western military action. Unmanned defence potentially removes part of that constraint.

The drone umbrella which creates this deterrence may not look exactly like the nuclear umbrella. Raw economic power will still matter enormously, but small, technologically capable countries may offer disproportionate value. Burnham’s Ukraine deal is a case in point: Britain brings its technological and research strength, while Ukraine brings something that money cannot easily buy – four years of battlefield data and experience through its “Avengers AI lab”.

Drones have already revolutionised warfare. Tomorrow, they could revolutionise military alliances. And 20 years from now, the strategic question may not be whether a country has drones, but whose drone ecosystem protects it.