Connectivity Infrastructure Is Not a Neutral Service

June 11, 2026

satellite-internet geopolitics starlink infrastructure-policy tech-governance

When SpaceX deactivated Russian military terminals in February 2026, a firmware whitelist did what sanctions committees and foreign policy briefings could not — and no elected body authorized it.

That’s worth sitting with for a moment. A private company, operating satellites in low Earth orbit, made a real-time decision that degraded the command-and-control capability of a nation-state’s military by roughly 90%. Russian units that had been relying on smuggled Starlink terminals — acquired through Central Asian black markets and deployed along the Donetsk frontline — suddenly went dark. They reverted to VHF radios and fiber-optic improvisation. The whitelist was technically elegant, operationally decisive, and politically unaccountable in any meaningful democratic sense.

I find myself thinking about this not because SpaceX made the wrong call — they probably made the right one — but because the framing we’ve used to understand these companies for the past two decades is no longer coherent.

The Neutral Pipe Fiction

The legal and regulatory architecture governing connectivity infrastructure was built on a fiction: that the companies owning it are neutral conduits. Common carrier doctrine. Dumb pipes. The infrastructure is the road; the content is the traffic; the company is just the asphalt.

That framing was always strained. It became unsustainable when the infrastructure started being controlled by software that could make granular decisions about who gets connectivity based on GPS coordinates, device registration, and travel speed. SpaceX didn’t just build a pipe. They built a pipe with a whitelist, geofencing that cuts off terminals moving faster than 90 km/h (to prevent integration into attack drones), and over-the-air firmware updates that can swap out GPS positioning systems to counter electronic warfare. That’s not asphalt. That’s a weapons system with a terms-of-service agreement.

The consequences of maintaining the fiction are real. When Elon Musk declined to activate Starlink over Crimea in 2022 — effectively vetoing a Ukrainian submarine drone operation against the Russian Black Sea Fleet — there was no oversight body, no appeals process, no democratic check. A CEO made a battlefield decision for a sovereign nation. We don’t have a framework for that because we’ve been pretending the framework isn’t necessary.

The Accountability Vacuum

What makes this genuinely dangerous isn’t any specific decision — it’s the absence of any consistent structure for making those decisions at all.

Iran’s 2026 internet blackout offers the mirror image. When the regime took national connectivity to roughly 1% of normal levels, the approximately 50,000 smuggled Starlink terminals inside the country became the primary communications lifeline for dissidents. SpaceX engineers responded to state-sponsored GPS jamming by pushing firmware updates that let terminals orient themselves using Ku-band satellite signals directly — bypassing terrestrial GPS entirely. That’s SpaceX doing something that looks a lot like supporting a population resisting its government. Again: no elected body authorized it. Again: probably the right call.

But “probably the right call” is not a governance framework. It’s a prayer that the private actor making unilateral geopolitical decisions also happens to share your values about which populations deserve connectivity and which militaries deserve disruption.

The problem compounds when you extend it beyond satellite infrastructure. China’s NDRC blocked Meta’s $2 billion acquisition of Manus AI in April 2026 by invoking a “look-through” doctrine — asserting that regardless of where Manus had incorporated (Singapore, via a holding company called Butterfly Effect Pte.), the underlying technology originated in China, was trained on Chinese resources, and therefore remained under Chinese sovereign jurisdiction. The founders couldn’t leave the country. The deal was unwound by one-line directive.

That’s a state asserting that intellectual property carries a national origin that corporate restructuring cannot launder away. Meanwhile, a week later, the NeurIPS Foundation was briefly enforcing sanctions compliance rules that would have barred researchers from Huawei and several Chinese universities from submitting papers — until the Chinese academic apparatus threatened a boycott severe enough that the foundation reversed course within four days.

The Infrastructure Is the Policy

What all of these episodes share is a collapse of the separation between infrastructure and political action. The infrastructure is the policy. The whitelist is the sanctions enforcement. The firmware update is the foreign policy. The acquisition block is the export control.

I’ve been building software for long enough to understand why this happened gradually and somewhat accidentally. You build a system that needs to handle fraud, or export law, or spectrum licensing, and you add controls. Those controls accumulate. Before long you have a capability that can be exercised at geopolitical scale, even if that was never the intent. The engineering decisions and the political decisions become the same decisions, and the people making them are accountable to shareholders and their own judgment, not to the populations affected.

The fiction of neutrality lets us avoid the uncomfortable question: if connectivity infrastructure is now a geopolitical actor, who does it answer to? Not in the abstract — concretely. What body can override a SpaceX decision to whitelist or blacklist a territory? What review process exists when a firmware update becomes the decisive factor in a military campaign?

We don’t have good answers. We barely have the questions yet. The infrastructure is already acting; the accountability structures are still in the fiction.

Sources

We Sold China the Keys to Our Supply Chain, Then Called It a Threat

April 24, 2026

supply-chain geopolitics rare-earths corporate-incentives manufacturing