Amazon's $11.57 billion move to acquire Globalstar and co-sign a new Apple agreement reshapes who controls the pipe that AI-enabled devices will increasingly depend on.
Driving through Potomac, Maryland, you will lose your signal. Not satellites. No tower on that stretch of River Road has ever penciled out for any carrier. Amazon's announcement this morning is not about fixing that. It is about the geography where no tower will ever get built at all.
Amazon confirmed it will acquire Globalstar, a mobile satellite services operator and pioneer in non-geostationary orbit direct-to-device technology, in a deal valuing the company at approximately $11.57 billion (unaudited, per company announcement). Separately, Amazon and Apple signed an agreement for Amazon's low earth orbit network, now branded Amazon Leo, to power satellite services for iPhone and Apple Watch. That means Emergency SOS, Messages via satellite, Find My, and Roadside Assistance will continue to run, backed by Amazon's growing constellation rather than Globalstar's standalone one.
The deal is expected to close in 2027, subject to regulatory review. No separate Globalstar stockholder vote is required since holders of more than a majority of voting power have already approved via written consent.
Spectrum Is the Asset Amazon Actually Bought
Globalstar's satellites are useful. Its spectrum is the real prize. The acquisition hands Amazon a globally harmonized portfolio covering L/S-band, Band 53/n53, and C-band optionality. Per the SEC filing, combining Globalstar's spectrum with Amazon Leo's capabilities enables higher-capacity, more spectrum-efficient direct-to-device services than legacy direct-to-cell systems. That distinction matters. Direct-to-cell repurposes existing handset radios with modifications at the network layer. Direct-to-device goes further, enabling native satellite communication without cellular tower intermediation. The spectrum licenses Globalstar holds carry global authorizations, which would take years and uncertain regulatory outcomes to replicate from scratch.
Amazon Leo currently has 241 satellites in orbit. Its full authorized constellation is 3,236 satellites across three orbital shells ranging from 590 to 630 kilometers altitude. The Federal Communications Commission originally required approximately 1,600 operational satellites by July 2026. Amazon has requested an extension to 2028, acknowledging it will reach roughly 700 by the original deadline. That gap is real and material. SpaceX's Starlink operates over 10,000 satellites today. Amazon is buying time as much as it is buying satellites.
The question is not whether satellite connectivity will matter to enterprise AI workloads. It already does. The question is who controls the infrastructure layer when it does.
Apple's Position After This Deal Is More Complicated Than It Looks
Apple had taken a roughly 20 percent equity stake in Globalstar in late 2024, accompanied by approximately $1.1 billion in infrastructure prepayments. That stake made Apple a party to any acquisition discussion, and the new Amazon-Apple agreement is the outcome of those negotiations. Apple's satellite features will continue uninterrupted, built on Amazon Leo's expanded network rather than Globalstar's standalone one. Amazon has committed to supporting current iPhone and Apple Watch models using Globalstar's existing constellation, and to collaborating with Apple on future satellite services on the expanded Leo network. The next-generation constellation, under construction by MDA Space, is part of what transfers in the acquisition.
Apple's 20 percent stake is now Amazon's problem to honor. The satellite features keep running. What Apple no longer controls is the pricing conversation as Leo scales and Amazon's enterprise and government customers start sharing the same infrastructure that runs iPhone's consumer features.
Every AI-Connected Device Is Making a Connectivity Assumption That May Not Hold
As on-device artificial intelligence processing advances, devices are moving from cloud-dependent inference toward hybrid architectures. A phone makes a local inference decision, then calls back for context enrichment, model updates, or data synchronization. That handshake assumes reliable connectivity. In practice, enterprise field teams, logistics fleets, remote sensors, and anyone driving through exurban Maryland knows the assumption frequently fails.
Amazon Leo, fully integrated with Amazon Web Services, is being positioned as a connectivity layer that also serves as a cloud-native pipe. Per Amazon's own positioning, Leo will enable enterprises and governments to move data for storage, analytics, and AI processing without relying on terrestrial network availability. For organizations running distributed workloads, that is a wide area network architecture decision, not a device feature.
The Leo terminal lineup reflects this. The Leo Nano delivers up to 100 megabits per second for mobile scenarios. The Leo Pro targets 400 megabits per second for standard enterprise deployments. The Leo Ultra targets 1 gigabit per second for industrial and Internet of Things applications. Amazon claims uplink performance six to eight times better than current satellite alternatives (unaudited, per company materials). Those numbers have not been independently validated at commercial scale.
The Regulatory Friction Is the Story Within the Story
Amazon's request for an FCC deadline extension to 2028 is a live risk. The commission is expected to rule in coming months. If the extension is denied, Amazon faces a hard operational constraint at exactly the moment it is absorbing Globalstar's infrastructure and managing the Apple transition. For a company announcing an $11.57 billion acquisition, admitting it cannot meet its own satellite deployment commitments on the original schedule is a material signal that CIOs evaluating Amazon Leo for enterprise connectivity should weigh carefully.
AST SpaceMobile is building a competing direct-to-device constellation designed to work with unmodified smartphones. Its reported revenue for the fourth quarter of 2025 was $54.30 million, a figure the company describes as representing more than 2,700 percent year-over-year growth (unaudited, per company disclosures). That growth rate reflects a small base, but the technical approach of reaching unmodified devices has enterprise implications worth tracking.
The more immediate competitive reality is that Starlink has density, coverage, and a commercial head start. Amazon's advantage, if the regulatory and deployment timeline holds, is the AWS integration and the spectrum assets that now come with the Globalstar acquisition. Those are durable structural advantages. They just require years to materialize.
What This Means Before the Deal Even Closes
Before committing to Leo as part of any network architecture, enterprise teams need to know what they are actually buying today versus what Amazon is promising to build. Between now and the anticipated 2027 close, Amazon is operating two parallel infrastructure realities: Leo's growing but still thin constellation, and Globalstar's existing operational network. The Apple agreement bridges those two realities cleanly for consumer device features. The enterprise path is less defined.
Organizations already committed to AWS infrastructure have the clearest near-term case for evaluating Leo as a wide area network complement to fiber and 5G. The integration with AWS cloud services means satellite connectivity is not just a backup link. It is a potential primary path for workloads in locations where terrestrial networks consistently underperform. Remote operations, logistics telemetry, field service automation, and edge inference scenarios all belong in that evaluation.
The places Amazon is actually targeting have never had a carrier debate at all.
Amazon Leo's enterprise proposition is that satellite connectivity plus AWS integration eliminates the coverage gaps that limit your distributed AI workloads. Before that claim can be tested at your organization, Amazon has to close the Globalstar deal, satisfy or renegotiate its FCC satellite deployment obligations, and deliver the Leo Ultra at the throughput numbers it is claiming. Ask your AWS account team specifically which enterprise connectivity SLAs Amazon will put in writing for Leo coverage in your operating geographies before 2028, and what the contractual remedies are if deployment milestones slip again. A constellation that is still being built is infrastructure, not a service commitment.
