By Nick David, Editorial Lead, SatNews
The third-party merchant RF supply chain is positioning for fiscal 2027. The operators leaning on it have written 2026 forecasts that assume it arrived last quarter.
AST SpaceMobile and Eutelsat OneWeb have committed 2026 subscriber and gateway targets that, by SatNews’s analysis, lean on a third-party merchant RF supply chain its own suppliers position to ramp in fiscal 2027. Amazon Leo, betting on the same 2026 horizon, is a distinct case: it runs its own custom silicon, which makes its terminal-volume ramp, not chip supply, the open question.
When MACOM management told investors in early 2026 that the company was positioned for a “very strong fiscal 2027” in SATCOM, it registered as a forward-looking growth call. Read against the subscriber commitments operators have made for 2026, the comment is closer to a warning.
The dominant US supplier of GaN-on-SiC RF MMICs, the technology layer that sits inside every modern flat-panel satellite terminal and gateway power amplifier, is publicly positioning its SATCOM ramp at fiscal 2027. That timeline sits in tension with the 2026 subscriber and service commitments of the largest commercial LEO operators, though how exposed each one is depends on how much of the supply chain it controls.
The Asymmetry · By the Numbers
2027
Fiscal year MACOM and Qorvo position their SATCOM RF ramp
90K/wk
Starlink terminal capacity at Bastrop, vertically integrated
$1.2B
AST SpaceMobile aggregate contracted revenue commitments across ~60 MNOs
$150.2M
Wolfspeed Q3 FY26 revenue, down 19% YoY, the SiC substrate layer
Starlink is the exception. Public reporting puts SpaceX’s Bastrop, Texas terminal factory at an annual capacity near 4.7 million user terminals, roughly 90,000 units a week, with actual output already running past 70,000 a week and per-unit cost falling over the life of the program. The terminal silicon, the antenna stack, and the satellite hardware are vertically integrated under one operator. Starlink’s subscriber math is calibrated to a supply chain Starlink owns.
Starlink’s subscriber math is calibrated to a supply chain Starlink owns.
The operators positioning to compete with it in 2026 sit on a spectrum of exposure to that outside supply chain.
Amazon Leo’s consumer terminal targets a production cost under $400, anchored to a custom “Prometheus” baseband ASIC used across the terminal, the satellite, and the ground infrastructure. This is Amazon-designed, in-house silicon, distinct from the third-party merchant RF MMIC supply this analysis identifies as the binding constraint; Amazon says it is in production and that the company will begin serving customers in 2026. Whether that vertical integration insulates Amazon from the merchant-layer timeline is the open question; the consumer-volume terminal ramp, not chip supply, is the variable to watch. Amazon’s deployment program still assumes both a satellite ramp and a terminal ramp on overlapping 2026 schedules, against an FCC deployment-milestone clock on which the agency has now acted: in June 2026 the FCC declined Amazon’s requested 24-month extension and instead granted a limited waiver of the July 30 half-constellation (1,616-satellite) deadline, attaching a temporary loss of spectrum priority for satellites launched after that date and leaving the July 2029 full-deployment requirement in place.
Eutelsat OneWeb depends on Intellian and Kymeta as primary terminal suppliers. Kymeta’s flagship multi-band KuKa terminal ships prototypes to key customers from mid-2026, with commercial volume scheduled for 2027. Intellian’s flat-panel ESA line for LEO is in active expansion. The capacity to absorb Eutelsat OneWeb’s enterprise and government pipeline through 2026 sits with third-party manufacturers whose own ramps are public.
AST SpaceMobile has committed to fifteen gateways across five continents in 2026, backed by more than $1.2 billion in aggregate contracted revenue commitments from nearly sixty mobile network operators, AT&T and Verizon among them, with carriers expecting initial US national service this year. AST’s bottleneck sits at the gateway layer: the user device is the smartphone, which shifts the entire constraint onto AST’s gateway buildout. That gateway equipment runs through the same RF foundries and ESA component suppliers that feed every other operator.
The RF supply chain is the binding layer for the operators that rely on it. MACOM demonstrated a 25-watt Ka-band GaN-on-SiC power amplifier at SATShow Week in March 2026 and confirmed it had licensed 40-nanometer GaN technology from Hughes Research Laboratories for E-, W-, and D-band satellite links. Its revenue ramp is positioned at fiscal 2027. Qorvo, whose high-performance analog segment carries the RF GaN it sells into defense and satellite programs, treats GaN as an emerging priority rather than a high-volume line. Neither company is calibrated for a 2026 SATCOM volume ramp, because neither has built the volume capacity for one.
Beneath the RF MMIC layer sits the SiC substrate supply. Wolfspeed, the dominant US wafer supplier, reported fiscal Q3 2026 revenue of $150.2 million, down nineteen percent year over year, with the CFO publicly identifying improving factory utilization as the company’s primary lever for margin recovery. That is the language of capacity management. The 200mm SiC line at Mohawk Valley is real, but most of its output is allocated to EV-grade power semiconductors. RF-grade SiC for satellite-band GaN MMICs is a smaller, slower-to-scale node.
The RF supply chain, layer by layer
SUBSTRATE
SiC wafers
Wolfspeed, output mostly EV-grade
CAPACITY-LIMITED
RF MMIC
GaN-on-SiC
MACOM, Qorvo, ramp set for FY2027
BOTTLENECK
HARDWARE
Terminals & gateways
Intellian, Kymeta, volume 2027
OPERATORS
Subscriber targets
AST & OneWeb lean on third-party RF; Amazon Leo more integrated
AST & OneWeb’s 2026 commitments lean on a 2027 RF supply chain
Starlink owns every layer; Amazon Leo runs its own custom silicon.
The ground-segment counterpoint is Kratos. The company reported $2.01 billion in total backlog at the end of fiscal Q1 2026, with Government Solutions backlog climbing from $1.212 billion to $1.635 billion in a single quarter and management citing space among the drivers. Kratos sells software-defined ground-segment systems to operators and defense customers building or modernizing teleports. The demand is real. It is also a different layer of the problem: subscriber-facing user terminals run on a separate supply chain Kratos does not address.
The strongest counterargument is that the supply chain will catch up. Wolfspeed is scaling its SiC platform. MACOM has new technology nodes coming online. Intellian and Kymeta are launching new terminal lines. Capex is flowing. The constraint, the argument goes, dissolves before the 2026 forecasts fully break.
That argument has a timeline problem. Capacity additions in compound-semiconductor manufacturing run on twelve-to-twenty-four-month qualification cycles from capex commit to first qualified wafer. The Wolfspeed ramp is allocated primarily to EV-grade power. Kymeta’s flagship multi-band terminal moves to volume in 2027. MACOM’s SATCOM ramp is fiscal 2027. For the operators that buy from this layer, the math of the 2026 forecasts assumes a supply chain the foundry layer has publicly told investors will not arrive until next year.
The pattern is hard to miss. Of the four major commercial LEO subscriber stories of 2026, one belongs to an operator that built its own terminal pipeline. Starlink will hit its numbers because Starlink built the supply chain to serve them. AST SpaceMobile and Eutelsat OneWeb, by SatNews’s analysis of the public record, have made 2026 commitments that lean on a third-party merchant RF and terminal supply chain whose suppliers have positioned their SATCOM ramps for fiscal 2027; the operators have not characterized their commitments this way, and the assessment is ours. Amazon Leo is a distinct case: it designed its own Prometheus baseband silicon and is integrating it end to end, so its 2026 bet rests less on that merchant chain than on proving out its own terminal-volume ramp. Where operators depend on the merchant layer, the likely outcome is a wave of selective subscriber-target softening through Q3 and Q4, attributed to demand softness, regulatory friction, or financing constraints. The underlying cause would be foundry capacity.
Key Takeaway
By SatNews’s analysis of public filings, AST SpaceMobile and Eutelsat OneWeb have committed to 2026 subscriber numbers that lean on a third-party merchant RF supply chain their suppliers have told investors ramps in fiscal 2027. Starlink, which built its terminal pipeline end to end, is the clear exception; Amazon Leo, running its own custom baseband silicon already in production, is a more vertically integrated case whose open question is consumer-terminal volume rather than chip supply.
Watch MACOM’s fiscal Q4 SATCOM commentary in August. Watch Qorvo’s defense and aerospace bookings over the next two quarters. Watch Kymeta’s first customer deliveries on KuKa. Watch the subscriber updates from Amazon Leo and AST SpaceMobile against the gateway and terminal manifests they have published.
One operator solved the ground segment. Its competitors have committed to 2026 subscriber numbers that, in differing degrees, still depend on a third-party supply chain whose own timeline points to 2027.
About the Author
A storyteller at heart, Nick David covers space policy, satellite markets, defense, and the technologies reshaping how humanity operates beyond Earth. With a background in creative direction, brand strategy, and editorial storytelling, he brings a modern lens to complex subjects and a relentless curiosity about what comes next.
