Amazon Globalstar and Kuiper (Amazon Leo): What the Deal Means for the Satellite IoT Stack

Amazon has agreed to buy Globalstar, the operator that carries Apple’s satellite emergency features, for about $11.6 billion, and the US Federal Communications Commission (FCC) cleared the spectrum transfer on 27 October 2026. Yet the version of this story that circulated a year ago, with a 2024 announcement, a “Kuiper” constellation already serving customers and a finished D2D network, was wrong on nearly every date. The Amazon Globalstar Kuiper Leo satellite IoT story is more interesting, and less finished, than the headlines made it sound.

It matters now because three things changed in twelve months. Project Kuiper became Amazon Leo. The Globalstar purchase was signed in April 2026 and has now passed its most important regulatory gate. And the 3GPP non-terrestrial network (NTN) standards that let ordinary cellular IoT chipsets talk to satellites are moving from lab to product. If you design devices, tracking fleets or remote telemetry, the question is what changes for your connectivity choices over the next three years.

You leave with a verified timeline, a clear separation between what Amazon owns today and what it only plans, an architecture of how Leo, Globalstar spectrum and AWS could fit together, an honest comparison with Starlink, Iridium and Skylo, and a decision checklist.

What this covers: what was actually announced, the November 2026 state of Amazon Leo, why the spectrum is the prize, a reference architecture for satellite IoT, the standards underneath, competitor positioning, risks, and recommendations.

Context and Background

For most of the last decade, “satellite IoT” meant one of three things. You bought a proprietary module and a subscription from a mobile satellite service (MSS) operator such as Globalstar, Iridium or Orbcomm. You used a store-and-forward nanosatellite service that delivered small messages a few times a day. Or you paid for a broadband terminal such as VSAT and treated the satellite as just another WAN link at a remote site. Each was expensive per device or per megabyte compared with cellular, and each locked you into a single operator’s hardware.

Two shifts broke that pattern. First, low Earth orbit (LEO) broadband constellations, led by Starlink, proved that large satellite fleets can be launched and operated at scale. SpaceX is described in recent coverage as operating roughly 11,000 satellites, while Amazon Leo is still in the low hundreds. Second, 3GPP standardised satellite access for narrowband IoT (NB-IoT) and LTE-M in Release 17, so a cellular modem can in principle use a satellite as a very distant cell tower. That second shift is what makes spectrum ownership strategic, because a standards-based phone or sensor needs a satellite operator that holds licensed mobile spectrum in the right band. For the cellular side of this story, our guide to satellite IoT with NB-NTN versus LoRa satellite and direct-to-device selection covers the radio trade-offs in depth, and this article focuses on the corporate and architectural consequences of the Amazon move.

Amazon’s own satellite effort began as Project Kuiper, approved by the FCC in July 2020 for 3,236 satellites in three shells at roughly 590, 610 and 630 km altitude. Amazon announced the rename to Amazon Leo on 12 November 2025. Throughout this article, “Kuiper” refers to the programme before that date and “Leo” to the same system afterwards. They are not two networks.

Globalstar is the older, smaller company. It operates a LEO constellation of about two dozen satellites, built in the 2000s and replenished since, together with licensed MSS spectrum in the L-band and S-band. Its commercial notoriety comes from one customer: Apple holds roughly a 20 percent stake and uses Globalstar capacity for Emergency SOS via satellite on iPhone 14 and later. Globalstar also sells asset-tracking and personal messaging hardware to industrial and consumer users, which is the IoT business relevant to this article.

For primary reporting on the deal, see the Globalstar investor release of 14 April 2026, which sets out the consideration and conditions, and Engadget’s report on the FCC clearance.

What the older coverage got wrong

The stale version of this post, and several recycled summaries, contain claims that do not survive checking against primary sources. The deal was announced on 14 April 2026, not in 2024. The headline value is about $11.5 to $11.6 billion depending on rounding and how the equity is valued, which is consistent with the figure the older post used, but the consideration structure and the date are different. Leo’s constellation is 3,236 satellites on paper and nearly 400 in orbit in practice. Claims of mature Ka-band processing satellites, a fully deployed D2D network or live Leo IoT service are not supported. The rest of this article marks every figure it could not verify.

What Was Actually Announced, and What Has Closed

The Amazon-Globalstar agreement is a definitive merger agreement, signed 14 April 2026, with a separate agreement between Amazon and Apple. It is not yet a completed acquisition. In October 2026 the FCC approved transfer of Globalstar’s licensed mobile-satellite spectrum to Amazon Leo, but the deal itself is expected to close in 2027, subject to conditions.

The consideration

According to Globalstar’s release, each shareholder elects either $90.00 in cash or 0.3210 Amazon shares per Globalstar share, with the stock component valued at no more than $90.00. Cash elections are capped at 40 percent of Globalstar shares, and excess cash elections convert pro rata into stock. Total consideration can fall by up to $110 million if Globalstar misses certain operational milestones. Holders of about 58 percent of voting power approved the merger by written consent, so shareholder approval is largely a formality, and the filings that remain are a Schedule 14C information statement on Globalstar’s side and a Form S-4 registration on Amazon’s.

Light Reading reports termination fees on both sides: roughly $419.8 million payable by Globalstar if it breaches or takes a rival deal, and roughly $592.1 million payable by Amazon in specified circumstances, including failure to win antitrust or foreign investment clearances. Those fee levels are the clearest signal of how Amazon and Globalstar themselves weigh the risk. Amazon is protecting the downside of a regulatory failure and Globalstar is being paid for the risk of waiting. Globalstar’s share price jumped roughly ten percent to about $80 on announcement day, a discount to the $90 offer that reflects timing and execution risk.

The FCC step

On 27 October 2026 the FCC Space Bureau approved Amazon Leo’s application to take over Globalstar’s MSS licences. Reporting on the order describes roughly 25 MHz of spectrum in the L-band near 1.6 GHz and the S-band near 2.4 GHz, plus Globalstar’s Band 53 (n53) terrestrial allocation at 2483.5 to 2495 MHz that supports its private 4G and 5G business. The FCC found no material public-interest harm and treated the sole objection, from a company called Yippy, as a private arbitration matter. Closing is described as dependent on Globalstar meeting milestones for its replacement satellites, referred to in the release as the HIBLEO-4 programme, being built by MDA Space.

Two caveats apply. This article could not verify which antitrust and foreign-investment approvals remain outstanding, and the release does not enumerate them, so treat the 2027 close as Amazon’s expectation and not a certainty. And the FCC order concerns the spectrum transfer. A separate Amazon application for a next-generation direct-to-device (D2D) system of up to 5,105 satellites was reported as still pending at the time of writing.

The Apple agreement

The Apple piece is the commercial anchor of the deal. Amazon Leo will provide satellite connectivity for current and future iPhone and Apple Watch features, including Emergency SOS, Messages, Find My and Roadside Assistance via satellite. The release names iPhone 14 and later and Apple Watch Ultra 3 as supported. Amazon will continue to support existing models on Globalstar’s current and planned constellations and will work with Apple on future services on the Leo network.

Light Reading characterises Amazon Leo as becoming Apple’s primary satellite provider, while noting that Apple can still work with other D2D providers and that carrier satellite offerings remain available. That nuance matters. Apple’s volume gives the combined company a guaranteed anchor tenant with consumer-scale revenue, but it is not an exclusive lock. The financial terms of the Apple agreement were not disclosed.

Amazon Leo in November 2026: A Constellation Still Being Built

Amazon Leo is a partly deployed constellation without open commercial service as of late October 2026. Roughly 390 production satellites are in orbit against a licensed plan of 3,236. Amazon says consumer service begins before the end of 2026 in the US, UK, France, Germany and Canada, with an enterprise preview running since November 2025.

Direct answer first, because the older coverage blurred it: Leo is not yet a competitor to Starlink in any measurable sense. It is a well-funded, late entrant with a credible launch manifest and a very large purchase of spectrum that Starlink cannot easily replicate.

Satellite count and launch cadence

Wikipedia’s tally, current to July 2026, lists 396 production satellites launched plus two prototypes, across 15 flights. The most recent flight in that list was the final of nine Atlas V missions on 2 July 2026, and Ariane 64 debuted in June 2026 carrying 36 satellites. One tracker’s live counter showed 387 in orbit in late October, against a body-text figure of “392 or more” launched, which is the expected gap between satellites launched and satellites still operating or counted. Treat the figure as “roughly 390 to 400” until Amazon publishes an official count.

Amazon has purchased 92 launches in total at a reported cost above $10 billion. The mix spans United Launch Alliance Atlas V and Vulcan, Arianespace Ariane 6 (24 missions after six more were announced in September 2026), Blue Origin New Glenn and SpaceX Falcon 9. The diversity is deliberate. A constellation of this size cannot depend on one launcher.

The regulatory clock

The original FCC authorisation required half the constellation, 1,618 satellites, in orbit by 30 July 2026, and the full system by 30 July 2029. Amazon was nowhere near the first milestone. In June 2026 the FCC waived the deadline but temporarily lowered the spectral priority of satellites launched after it, until Amazon certifies a larger tranche as built. One report puts that certification threshold at 1,616 satellites by 30 October 2027. Because the sources round the threshold differently (1,616 versus 1,618), this article treats the exact number as unverified and the principle as firm: Amazon is operating under a conditional waiver and must accelerate.

That clock explains part of the Globalstar logic. The Globalstar spectrum is licensed, in use and not subject to the same constellation-deployment milestone, so it is the quickest route to a service that does not depend on having 1,600 satellites in orbit.

Service availability and terminals

Amazon’s own announcement for Uzbekistan says the service is in private preview with employees and select enterprise customers, with rollout “later this year” through the local operator Uztelecom and its UZ-SAT subsidiary. A news report that Leo has already launched there conflicts with that statement, and this article follows Amazon’s wording. Other partnerships reported include Australia’s NBN on a wholesale basis, DIRECTV and Sky Brasil in Latin America for 2027, JetBlue as the first airline for 2027, and a South African arrangement with Herotel expected in 2027.

Three self-install terminals are announced. Leo Nano is about seven inches square, targets up to 100 Mbps, and is aimed at residential and mobility use. Leo Pro is larger and targets 400 Mbps, with one tracker citing a price under $400 that Amazon had not confirmed. Leo Ultra targets up to 1 Gbps down and 400 Mbps up in a roughly 20 by 30 inch panel. Monthly service pricing had not been announced in the sources reviewed. For context, the same tracker lists Starlink US residential service at $120 per month.

Figure 1: From the 2020 Kuiper licence to the 2028 direct-to-device target, showing what is closed, what is pending and what is only planned.

Figure 1 lays the sequence out as a dependency chain. The Kuiper licence leads to the Leo rebrand and deployment, which runs in parallel with the Globalstar deal and its FCC clearance. The 2027 close and the 2028 D2D system sit at the end of the chain and depend on regulatory and launch outcomes that are not yet resolved.

Why Globalstar’s Spectrum Is the Real Asset

Amazon is not paying $11.6 billion for 24 ageing satellites. It is paying for licensed mobile-satellite spectrum that works with standard handsets, an anchor customer, and time. A broadband constellation using Ka-band can serve dishes. A phone or a coin-sized sensor cannot use a dish, so only operators with MSS spectrum that handset chipsets already support can offer direct-to-device service in a way consumer devices accept.

What the bands are

Globalstar’s licences cover L-band uplink near 1610 to 1618.725 MHz and S-band downlink 2483.5 to 2500 MHz, as described in the separate D2D filing. The S-band overlaps with 3GPP band n53 at 2483.5 to 2495 MHz, a band Globalstar has promoted for private terrestrial networks. Roth Capital, quoted by Light Reading, calls n53 Globalstar’s most valuable asset. That is an analyst view, and it is plausible for a simple reason: a spectrum block that can host both a terrestrial cell and a satellite downlink can be used for hybrid services where the satellite fills coverage gaps.

The relevance to IoT is that a chipset which already supports an n53-compatible or neighbouring S-band configuration can reach the satellite with modest firmware changes, instead of requiring a new radio front end. Apple’s iPhone 14 and later already do this. Many industrial IoT modules do not, and that is the gap Amazon would need to close with partners.

What Amazon says it will build

Amazon says that starting in 2028 it will deploy a next-generation D2D satellite system, designed for higher spectrum efficiency than legacy direct-to-cell offers, integrated with the first and second generations of the Leo broadband system and supporting mobile network operators. The separate FCC application for up to 5,105 D2D satellites, if approved, would be the vehicle. Until that happens, Amazon will continue to operate Globalstar’s existing fleet and the replacement satellites under construction by MDA Space alongside the Leo broadband satellites.

An analyst quoted by Light Reading suggests the larger future Globalstar C-3 constellation of 48 operational satellites plus spares was not addressed in the announcement and may be cancelled. That is one analyst’s speculation and not confirmed by either company.

Why this changes competitive dynamics

Direct-to-device has been a three-way contest. SpaceX has been building a D2D product with mobile operators and, per recent reports, buying spectrum from EchoStar. AST SpaceMobile has pursued large phased-array satellites with operators. Iridium has taken a different path by turning its existing global constellation into a standards-based NB-IoT service. This article did not independently verify the SpaceX and EchoStar transaction values and treats them as reported.

Adding Globalstar gives Amazon something it lacked: a revenue-generating licensed MSS business and a direct relationship with the world’s most profitable handset maker. It does not give Amazon a finished D2D network. The distance between those two statements is roughly two years and some billions of dollars of satellites.

A Reference Architecture for Satellite IoT with Leo, Globalstar and AWS

If Amazon delivers on the plan, an IoT device in 2029 or 2030 might reach AWS through one of three paths: a legacy Globalstar-style MSS link, an NB-NTN link using Leo D2D satellites, or a Leo broadband terminal acting as a site gateway. This section describes the architecture as an inference from Amazon’s public statements. Amazon has not published an IoT reference architecture for Leo, and nothing here should be read as a product announcement.

Figure 2: Three device access paths converge on ground gateways, a core network and cloud ingestion.

Figure 2 shows the likely layers. Devices at the left attach through either direct-to-device narrowband links or a site gateway with a broadband terminal. Satellites relay traffic to ground stations. A core network function authenticates devices and routes data. The cloud ingestion layer terminates the protocol and feeds storage and analytics.

Path A: narrowband direct to device

A battery-powered sensor with an NB-IoT or LTE-M modem that supports NTN would register on a satellite as it does on a cell tower, send a few hundred bytes, and sleep. Link budgets are the hard constraint. A LEO satellite at several hundred kilometres is thousands of times further than a terrestrial tower, so data rates are tiny, messages are short and latency is dominated by waiting for a satellite pass unless coverage is continuous. The 3GPP NTN work adds timing and frequency pre-compensation, because the device must correct for satellite Doppler shift and delay using its own GNSS position fix.

For this reason, narrowband satellite IoT suits messages measured in tens to hundreds of bytes: a pipeline pressure reading, a container position, a tank level. It does not suit firmware updates, video or anything chatty. Our piece on ambient IoT in 3GPP Release 19 and 20 covers the other end of the power spectrum, and it is a useful reminder that satellite links are the most power-hungry way to send a byte.

Path B: site gateway on a broadband terminal

The second path needs no special modem. A remote site, such as a mine, wind farm, vessel or oil pad, places a Leo Nano or Pro terminal at the edge and connects local devices through a conventional gateway running MQTT or OPC UA. Throughput is orders of magnitude higher, so the economics are different: you pay for a service plan for the whole site, not per device. The satellite becomes a WAN link, and standard edge patterns apply. For store-and-forward and offline buffering at the edge, see our guide to MQTT store-and-forward at an edge gateway.

Amazon has said enterprise customers are in preview since November 2025, but the sources reviewed describe no IoT-specific enterprise product, so how Leo terminals will be integrated into AWS IoT services is unverified. The architecture above is plausible because AWS already offers IoT Core and Greengrass, but plausible is not announced.

Path C: legacy MSS during transition

During the transition, Globalstar’s existing network continues to serve its installed base. Amazon has said it will support Apple’s existing models on the Globalstar constellations. By extension, Globalstar’s industrial tracking customers presumably keep service, but Amazon has made no public commitments about those products, and sunset timelines are unknown. Anyone with a Globalstar-based asset tracking fleet should ask the vendor directly.

Where the data lands

At the cloud end, the pattern is conventional. A satellite operator hands off packets at a ground-station gateway or core network. The protocol then needs normalising: NB-IoT NTN commonly carries small UDP or CoAP payloads, whereas site gateways carry MQTT. A unified namespace pattern with Sparkplug B is a sensible way to merge both feeds into one digital twin model, so the twin sees a tank level whether it arrived by narrowband satellite or a broadband gateway. The twin must also record link type and age, because a reading that is nine hours stale needs different treatment than one that is nine seconds old.

The Standards Underneath: NB-NTN, Spectrum and Device Reality

Whether any of this matters to an IoT engineer depends on one question: can the chipset you already buy talk to the satellite? The answer is set by 3GPP non-terrestrial network specifications, by which bands a module’s radio supports, and by whether the satellite operator exposes a standards-based interface or a proprietary one.

3GPP NTN in practice

Release 17 introduced NTN support for NB-IoT and LTE-M, and later releases refine it. The core adaptations are timing advance and frequency pre-compensation, longer timers to cope with round-trip delay, and a requirement that the device have a GNSS fix before transmitting. These adaptations are what allow a modem designed for a cell a few kilometres away to talk to a satellite hundreds of kilometres away. This summary comes from general knowledge of the specifications, and exact release-by-release behaviour should be checked against the 3GPP documents before you design a product around it. Our overview of 3GPP Release 20 and industrial IoT shows how the cellular roadmap continues.

The chipsets are arriving. Iridium’s January 2026 announcement said its first mobile-originated message over NTN Direct was sent using a Nordic Semiconductor nRF9151 LTE-M, NB-IoT and NTN module. That is a vendor press release, and the claim of 100 percent global coverage is Iridium’s own. It shows that mainstream cellular IoT silicon is being qualified against satellite networks.

Operator models compared

The market splits into operators that own licensed MSS spectrum and offer a standards interface, operators that sell proprietary links, and operators that wholesale capacity to carriers. Skylo is the best-known example of the first group in the consumer and enterprise IoT space, working through carrier and device partnerships on NB-NTN over MSS spectrum. This article did not re-verify Skylo’s current partner list or coverage and describes the model generally.

Iridium is the second path. It plans to run standards-based NB-IoT over its existing constellation, with Deutsche Telekom and Vodafone IoT announced as integration partners in 2025. Its January 2026 release said commercial service is planned for 2026 and beta testing was about to begin. Whether commercial service has actually launched was not confirmed in the sources reviewed, and this article marks it as unverified.

Globalstar is the third. Its MSS spectrum is the one most handset chipsets already support, because Apple’s use has driven support in iPhone radios. Under Amazon, the question is whether Globalstar-band support spreads from phones to IoT modules.

Starlink is the fourth. SpaceX’s direct-to-cell work is aimed at phones through mobile operators, and the August 2026 coverage found in research suggests it is extending its spectrum holdings. The details were not verifiable here beyond headlines. Starlink’s broadband terminals are fully commercial and widely used as site WAN links. In the comparison table below, Starlink’s IoT-specific D2D service is labelled as unverified.

Figure 3: Where each operator sits on spectrum ownership and standards-based IoT access, and how mature the offering is.

Figure 3 is a positioning map built from public statements, not a benchmark. The vertical question is whether the operator holds licensed spectrum that phones and modules support. The horizontal question is whether the service is live or announced.

Decision matrix

Operator path	Spectrum position	Device interface	Commercial status (Nov 2026)	Best fit
Amazon Leo broadband	FCC licence for the Leo broadband system	Leo terminals, site gateway	Private preview, consumer launch planned late 2026	Remote site WAN, vessels, mines
Amazon Leo plus Globalstar D2D	Globalstar L and S band MSS, n53	Standards D2D planned for 2028	Pending FCC D2D approval and deal close	Future phone and sensor access
Globalstar legacy	Same MSS spectrum	Proprietary modules, Apple features	Operating	Existing trackers, Apple devices
Iridium NTN Direct	Iridium L-band MSS	3GPP NB-IoT NTN	Beta then commercial planned 2026, status unverified	Global narrowband telemetry
Skylo style NB-NTN	MSS spectrum via partners	3GPP NB-NTN	Operating through partners, not re-verified	Phones, wearables, trackers
Starlink broadband	Broadband licences	Starlink terminals	Operating	Site WAN, high bandwidth

The table deliberately avoids price and latency numbers, because the sources reviewed did not give comparable, verified figures. Leo’s page-level claim of 20 to 40 ms latency is a list figure from a tracker, not a measured result.

Deeper Analysis: What Amazon Is Really Building

Reading the filings and press coverage together suggests a three-part strategy. The first part is the anchor revenue stream: the Apple agreement. The second is a spectrum position that makes Leo’s D2D ambition credible without waiting for a full constellation. The third is vertical integration with AWS, which Amazon has hinted at but not specified for IoT. Each part has a different risk profile.

The anchor tenant and its limits

Apple gives the combined company a predictable demand source for a satellite service that otherwise has no consumer product yet. It also gives a strong signal to chipset vendors that an Amazon-owned MSS band will remain supported in future iPhones. That second effect is what makes the spectrum attractive to IoT module makers: if the same band is guaranteed to be in a billion phones, supporting it in a sensor module costs little.

The limit is that Apple is a customer, not a captive. Reports say Apple may work with other providers, and Apple’s satellite features are small data services, such as short messages and location sharing. They are a good fit for narrowband capacity but are not a growth story by themselves.

The 2028 gap

Amazon’s own timeline puts the next-generation D2D system in 2028. Globalstar’s current constellation is old, and the replacement satellites are part of the closing conditions. In the meantime, the new owner inherits a network with limited capacity, and any promised IoT expansion must be carried by the legacy fleet and a relatively small number of new satellites. A realistic reading: through 2027, nothing in the IoT product experience changes for Globalstar customers, because the deal has not closed and the new network does not exist.

For planning, that gap is the most useful piece of information in the whole story. If you are choosing connectivity for devices that will be installed in 2027 and run for ten years, an Amazon D2D option is a possible future migration path and not a present option.

Integration with AWS: inference, not announcement

The older version of this article asserted that Leo would integrate natively with AWS, including edge compute on satellites. Nothing in the primary sources reviewed supports on-orbit processing or an AWS satellite edge product. Amazon’s own materials describe Leo as a broadband network with enterprise and government customers and a D2D system for mobile operators. What is realistic is the conventional cloud pattern: terrestrial gateways landing traffic in AWS regions, private connectivity from ground stations to a virtual private cloud, and device management through existing services. Think of it as plumbing that Amazon can bundle and price aggressively, not as a new compute paradigm.

This is where Amazon’s advantage lies. A buyer that already runs AWS can purchase connectivity and cloud together, with a single bill and a single support relationship. That is the same bundling logic that made cloud adoption sticky, and it is a more durable edge than raw satellite performance.

Figure 4: A single narrowband sensor message from wake-up to digital twin update, showing where delay and staleness enter.

Figure 4 follows one reading. The device wakes, gets a GNSS fix, waits for a satellite pass, transmits, and the ground gateway forwards to the cloud, which updates the twin. The long arrow is the wait for a pass, not the radio transmission. Designing for that wait is the central problem of satellite IoT, and it is largely independent of who owns the satellite.

Worked example: a tank farm

Consider an illustrative fleet of 2,000 battery-powered tank-level sensors across remote sites, each reporting 40 bytes every six hours. That is four messages per day per device, or 8,000 messages daily, and about 320 KB per day for the whole fleet before protocol overhead. At that volume, a narrowband satellite service is physically feasible: the data is tiny, and the constraint is per-message cost and battery life, not bandwidth.

Now change the requirement: the tanks also need a monthly 2 MB firmware update. That is 4 GB per month across the fleet, and a narrowband link at a few kilobits per second per device makes this impractical. The same fleet would instead use a gateway at each site with a broadband terminal for updates, and the narrowband path would stay for emergency alerts. The numbers here are illustrative arithmetic, not vendor data. They show why the Leo Nano-class terminal and the D2D service are complements and not substitutes.

Trade-offs, Gotchas, and What Goes Wrong

The risks fall into four groups: regulatory, schedule, technical and commercial. Most of them are already visible in the sources.

Regulatory and deal risk. The FCC spectrum approval is done, but closing remains conditional on other clearances and on Globalstar hitting satellite milestones. The termination fee structure, with Amazon owing a large sum if antitrust or foreign-investment clearances fail, shows the parties themselves see non-trivial risk. The separate D2D satellite application is a second regulatory gate. And the FCC waiver for Leo’s deployment milestone is conditional.

Schedule risk. Amazon needs to certify more than 1,600 satellites built by late October 2027 under the reported terms of the waiver, while operating roughly 390 today. Launch supply has been the bottleneck. Nine Atlas V missions are finished, and the rest depends on Vulcan, Ariane 6, New Glenn and Falcon 9 maintaining cadence. Slips in any single vehicle compound.

Technical risk. Direct-to-device from LEO is hard. Link budgets are tight, handset antennas are small and interference between satellite systems sharing adjacent spectrum needs coordination. Amazon claims higher spectral efficiency than legacy direct-to-cell systems, and that is an unverified design claim until the satellites fly.

Commercial risk. Globalstar’s installed base of industrial users has no stated migration path. Vendor lock-in is real: a device built around one operator’s proprietary module cannot move easily. The consolidation also reduces the number of independent MSS operators, which is a concentration risk for any buyer who wants competitive tenders.

Anti-patterns to avoid

Designing a product around a satellite service that has not launched, with no fallback.
Assuming that because Apple phones work with a band, an industrial module will too.
Treating narrowband links as if they were real-time. A reading delivered hours later must be tagged as late in your data model.
Sizing batteries from datasheet transmit currents without counting GNSS acquisition and retries, which often dominate the energy budget.
Reading a press release figure for coverage or latency as a measured service level.

Practical Recommendations

For most IoT teams the right response to this deal is to wait and design for portability, not to switch operators. Nothing Amazon has announced changes what you can buy in 2026. What changes is how you should structure contracts and architectures for the 2028 to 2030 window.

First, separate the application protocol from the radio. Keep payloads compact and transport-agnostic so a sensor can move from one satellite operator to another by swapping a module, not rewriting the backend. Use a gateway or ingestion layer that normalises messages, and record link type and message age in the data model.

Second, prefer standards-based modules. A 3GPP NTN-capable NB-IoT or LTE-M module gives you more operator options than a proprietary one. When evaluating, confirm in writing which satellite bands the module’s radio supports, because standards compliance does not guarantee band coverage.

Third, treat Amazon Leo as an option to evaluate for site connectivity once consumer and enterprise service opens, with pricing known. Run a pilot at one remote site, measure real latency and availability, and compare against Starlink and incumbent VSAT.

Fourth, ask existing satellite vendors direct questions. If you use Globalstar-based trackers, ask for written commitments on service continuity through the transition. If you evaluate Iridium NTN Direct, ask for the current commercial status and coverage evidence, since this article could not confirm it.

Checklist

[ ] Identify devices that need under 200 bytes per message and tolerate hours of delay.
[ ] Identify sites that need a broadband WAN link and cannot get terrestrial service.
[ ] Verify module band support against each operator’s spectrum, in writing.
[ ] Add link type and age to every telemetry record in the twin.
[ ] Define a fallback path for any service not yet commercial.
[ ] Re-check deal closing, the FCC D2D decision and Leo pricing in each quarter of 2027.
[ ] Budget for GNSS acquisition and retransmissions in battery calculations.

Frequently Asked Questions

Did Amazon complete the Globalstar acquisition?

No. Amazon signed a definitive merger agreement on 14 April 2026, and the FCC approved transfer of Globalstar’s mobile-satellite spectrum on 27 October 2026. The deal itself is expected to close in 2027, subject to remaining conditions that include Globalstar meeting milestones on its replacement satellites. Treat any article describing the acquisition as finished as premature. Check Amazon’s and Globalstar’s filings for the closing announcement before relying on the combined company’s services.

How much is Amazon paying for Globalstar?

Globalstar’s release says shareholders can elect $90.00 in cash or 0.3210 Amazon shares per share, with cash capped at 40 percent of shares and total consideration reducible by up to $110 million if operating milestones are missed. Press coverage rounds the equity value to roughly $11.5 to $11.6 billion. The exact figure depends on how the stock component is valued at closing, so quote the per-share terms when precision matters.

Is Project Kuiper the same as Amazon Leo?

Yes. Amazon announced on 12 November 2025 that Project Kuiper would be branded Amazon Leo. It is the same LEO broadband constellation, licensed by the FCC in 2020 for 3,236 satellites. Older articles using the Kuiper name describe the same system, but they may predate the rebrand, the Globalstar agreement and the 2026 FCC deployment waiver, so check their dates.

Is Amazon Leo available to customers yet?

Not generally, as of late October 2026. Amazon describes a private preview for employees and select enterprise customers, with initial rollout planned before the end of 2026 starting in the US, UK, France, Germany and Canada. Partners in other countries are expected in 2027. Monthly pricing had not been announced in the sources reviewed. Join the waitlist or enterprise preview if you need a firm date, and do not plan production deployments around the planned launch.

What does this mean for satellite IoT devices?

In the short term, nothing changes for devices already deployed. In the medium term, Amazon plans a next-generation direct-to-device system from 2028, using Globalstar’s licensed spectrum, but it needs further FCC approval. If you are choosing hardware now, favour standards-based 3GPP NTN modules, verify band support, and keep your data pipeline independent of any single satellite operator.

How does Amazon Leo compare with Starlink?

Starlink is far ahead on scale and service: reports describe roughly 11,000 satellites against Leo’s roughly 390, and Starlink has commercial terminals and plans, with a US residential price of $120 per month cited by one tracker. Leo’s differentiators are its AWS ownership, its Globalstar spectrum and the Apple relationship. Leo has not yet shown independent performance data, so any speed or latency comparison is a vendor target.