In the opening weeks of Russia's full-scale invasion of Ukraine in early 2022, Starlink terminals poured into the country and became a backbone of Ukrainian military and civilian communications. SpaceX's low-Earth-orbit broadband network proved resilient where fixed infrastructure was destroyed or jammed, and it let front-line units, command posts and drone operators stay connected. Starlink was widely and rightly credited with helping Ukraine keep fighting. But the same dependence that made it valuable also concentrated an extraordinary amount of operational leverage in a single private company, and in the hands of one man.
That leverage became concrete in September 2022. According to disclosures from SpaceX personnel and the account in Walter Isaacson's biography of Elon Musk, Musk personally directed SpaceX engineers to withhold Starlink coverage near the coast of Crimea as a Ukrainian attack unfolded. Ukraine had launched uncrewed surface vessels, essentially explosive-laden drone boats, at the Russian Black Sea Fleet anchored at Sevastopol. The drones relied on Starlink connectivity to reach their targets. With coverage denied over the approach, the vessels lost their link, and the attack on the fleet failed. Musk's stated reasoning, as later reported, was fear of being complicit in a major escalation, potentially a Russian nuclear response, if a private American network were used to enable a strike on Sevastopol, which Russia claims as its own territory.
The details were later contested. Musk himself disputed parts of the account, arguing at one point that coverage over Crimea had never been enabled in the first place rather than switched off mid-operation, so that no active denial had occurred. The distinction matters less than it might seem. Whether he turned coverage off or declined to turn it on, the outcome was that a Ukrainian military operation depended on a decision that only he could make, and he made it according to his own reading of escalation risk. A capability the defender treated as available was, in truth, available only at the discretion of a private owner who could withhold it silently and after the fact reframe what had happened. That ambiguity is itself part of the vulnerability. When the switch belongs to someone else, even the question of whether it was thrown may not be yours to answer.
Whether or not one shares his judgement, the structural fact is what matters. A specific military operation, planned by a sovereign government fighting for its survival, was thwarted not by the enemy and not by any government but by a decision taken inside a company by its owner. There was no treaty governing it, no allied command approving it, no legislature debating it. The connectivity a war depended on could be shaped, in real time, by one executive weighing his own read of the geopolitics. This is not a hypothetical about corporate power over conflict. It happened, and it changed the outcome of an attack.
The episode forced into the open a question that governments had been slow to confront. What does it mean to fight a war on communications infrastructure you do not own and cannot command? Ukraine could request coverage, and could rely on goodwill, but it could not compel it. The terms of access sat with SpaceX, and ultimately with Musk, who was also a commercial actor with business interests across many jurisdictions, including in China, and who held his own strong and idiosyncratic views on how the war should end.
The United States government's own response is telling. In 2023 the Pentagon signed a separate arrangement with SpaceX for Starshield, a variant of the network configured for government and military use with control resting on the government side rather than with the company. That is, having watched a private CEO decide the fate of a Ukrainian strike, the American defence establishment moved to secure a version of the capability whose operational switch it, and not SpaceX's owner, would hold. The quiet message of the Starshield contract is that Washington had drawn the same conclusion any careful observer would. Depending on a commercial network run at the discretion of its founder is not the same as possessing a sovereign capability.
For every other state, the Crimea episode is the clearest modern illustration of a specific vulnerability. Buying or receiving connectivity from a dominant commercial provider does not confer control over that connectivity in the moment it matters most. The provider's owner can geofence, throttle or withhold service according to his own risk calculus, and there may be no contractual or legal remedy fast enough to matter in a battle. Sovereignty over wartime communications is not established by having terminals in the field. It is established by controlling the network, or by owning an alternative that no single foreign executive can switch off.