Amazon Acquires Globalstar for $11.6 Billion — The Biggest Direct Challenge to Starlink Yet

Amazon just made the most aggressive infrastructure move of 2026. The company announced the acquisition of satellite operator Globalstar for $11.6 billion in an all-cash-or-stock deal at $90 per share — and the market’s reaction told you everything you needed to know. Analysts said the deal “paid for itself 10x over in one day” in terms of what it unlocks for Amazon’s broader strategy. This is not a speculative bet on the future of space. This is Amazon buying the missing piece it needed to challenge SpaceX’s Starlink head-on.

Globalstar operates 24+ low-Earth-orbit satellites with agreements for 50+ more already in the pipeline. It holds valuable spectrum licenses for direct-to-device connectivity. And it already has a proven commercial relationship with Apple, providing the satellite connectivity that powers iPhone 14+ emergency SOS and roadside assistance features. Amazon is not buying a startup with a pitch deck. It is buying an operational satellite company with real hardware in orbit, real spectrum on the books, and a real customer in Cupertino.

What Amazon Actually Bought

The surface-level read is that Amazon bought a satellite company. The actual read is that Amazon bought three things simultaneously: orbital hardware, licensed spectrum, and a proven direct-to-device capability. Each of those is individually hard to acquire. Together, they are the foundation for a satellite internet business that can compete with Starlink from day one.

Globalstar’s 24+ LEO satellites are already in orbit and operational. The company has agreements in place for 50+ additional satellites, meaning the constellation is designed to scale significantly beyond its current footprint. But the real prize may be the spectrum licenses. In satellite communications, spectrum is the bottleneck — you can build as many satellites as you want, but without licensed frequencies, you cannot legally transmit to consumer devices. Globalstar owns some of the most valuable spectrum for direct-to-device communication, and Amazon just locked all of it up.

The Apple relationship adds a layer that most coverage has underplayed. Globalstar already provides the satellite connectivity that powers iPhone 14 and later emergency SOS and roadside assistance features. That deal continues under Amazon ownership. It means Globalstar’s technology is not theoretical — it is literally running on hundreds of millions of iPhones right now, handling real emergency communications from devices that were never designed with satellite antennas.

Amazon Leo: Direct-to-Device Changes Everything

Amazon Leo is the product this acquisition enables. It is Amazon’s direct-to-device satellite internet service, expected to launch later this year. “Direct-to-device” is the phrase that matters here. It means your phone connects to a satellite overhead without needing a dish, a terminal, or any ground-based equipment. You walk outside, and your phone has internet — anywhere on Earth.

This is fundamentally different from Amazon’s existing Project Kuiper, which is a broadband service requiring a receiver terminal (similar to Starlink’s dish model). Amazon Leo goes further — it aims to put satellite connectivity directly into the phones and devices people already carry. The Globalstar acquisition provides both the satellite constellation and the spectrum licenses required to make that work.

The timing matters. SpaceX’s Starlink has been the dominant player in satellite internet, but its consumer product still requires a dish. T-Mobile and SpaceX have been testing direct-to-cell capability, but the rollout has been gradual. Amazon is entering the direct-to-device race with a proven satellite operator, existing spectrum, and the deepest pockets in tech behind it.

The Full-Stack Infrastructure Play

This is where the deal gets strategically interesting for anyone who thinks about infrastructure. After this acquisition, Amazon owns a combination of assets that no other company on Earth can replicate:

No other company has all four. Google has cloud but no satellites. SpaceX has satellites but no cloud. Apple has devices but neither cloud infrastructure at AWS scale nor its own satellite constellation. Microsoft has Azure but is relying on partnerships for space connectivity. Amazon is the only company assembling the complete vertical stack from data center to orbit to device.

Why This Matters for AI Practitioners

If you are an AI practitioner reading this and wondering why a satellite acquisition matters to you — it matters because connectivity is the constraint that limits where AI can operate. Right now, the most capable AI models run in data centers. Edge AI is growing, but it still depends on network connectivity to sync, update, and transmit results. Autonomous vehicles need constant connectivity for safety-critical decisions. IoT sensor networks need it for real-time data ingestion. Remote industrial operations need it for AI-powered monitoring.

Satellite internet solves the “last mile” problem for AI in a way that terrestrial networks never will. You cannot build cell towers in the middle of the ocean, across the Arctic, or in rural regions where the economics do not support ground infrastructure. But a LEO satellite constellation covers all of it. Amazon is building the connectivity layer that the next generation of AI applications will run on.

Consider the practical implications. AWS already offers machine learning services (SageMaker, Bedrock) and edge compute (Wavelength, Outposts). Now add ubiquitous satellite connectivity. An autonomous drone operating in a remote agricultural region can run local inference on the device and sync with AWS through Amazon Leo — no ground infrastructure required. A fleet of IoT sensors monitoring pipeline integrity across thousands of miles of wilderness can transmit anomaly detection results in real-time through satellite. An emergency response AI system can maintain connectivity even when terrestrial networks are destroyed by the disaster it is responding to.

This is not speculation. This is the logical endpoint of combining the world’s largest cloud with a global satellite network. Amazon just bought the orbital hardware and spectrum to make it happen.

The Starlink Challenge

SpaceX’s Starlink has been the undisputed leader in satellite internet. Over 6,000 satellites in orbit. Millions of active subscribers. A proven direct-to-cell partnership with T-Mobile. The question is whether Amazon can genuinely challenge that lead, or whether this is another Project Kuiper — ambitious on paper, slow in execution.

The Globalstar acquisition changes the calculus. Amazon is not starting from scratch the way it did with Kuiper. It is buying an operational company with hardware in orbit, proven technology on hundreds of millions of iPhones, and the spectrum licenses that would take years to acquire through the FCC. The $11.6 billion price tag suggests Amazon is not treating this as an experiment. It is treating it as core infrastructure.

The competitive dynamics are going to be fascinating. SpaceX has the satellite advantage in sheer numbers. Amazon has the cloud advantage, the enterprise customer base, and now the spectrum for direct-to-device. This is a race where both companies have genuine strengths, and the winner will likely be determined by execution over the next 24 to 36 months.

The AI applications that will define the next decade — autonomous vehicles, industrial IoT, precision agriculture, disaster response, defense systems — all share one common requirement: they need connectivity everywhere, not just where cell towers happen to exist. Amazon just spent $11.6 billion to own that connectivity layer and tie it directly to the world’s largest cloud. That is not a satellite deal. That is an AI infrastructure deal.

Understanding how cloud infrastructure, edge compute, and connectivity converge is exactly the kind of architectural thinking that separates AI practitioners from AI tourists. It is the difference between knowing how to prompt ChatGPT and understanding the systems that make AI work at scale.