5G FWA Uplink Power Class improvement by Samsung and Qualcomm

The uplink problem nobody talks about

Fixed Wireless Access has largely been told as a downlink story — gigabit downloads, fiber replacement, last-mile broadband for underserved communities. But the real performance ceiling for next-generation FWA deployments isn't how fast data arrives at the router; it's how efficiently data leaves it. As applications like physical AI inference, AR/VR collaboration, autonomous system telemetry, and real-time industrial monitoring flood the upstream channel, that ceiling is becoming increasingly visible — and costly.

This is the context that makes the Samsung–Qualcomm Power Class 1 validation genuinely significant. On May 6, 2026, Samsung Electronics announced it had successfully validated 5G Power Class 1 (PC1) capability through its fully software-driven virtualized RAN (vRAN) combined with the Qualcomm Dragonwing™ FWA Gen 4 Platform — equipped with the Qualcomm X85 Modem-RF chipset.^[1] The announcement describes it as an industry first, and based on what we know about PC1 specifications and vRAN architecture, that claim holds up.

Understanding Power Class — it's counterintuitive by design

The 3GPP Power Class framework, codified in 3GPP TS 38.101-1 (FR1) and TS 38.101-2 (FR2),^[2],[3] governs the maximum transmit power a User Equipment (UE) can push toward the network. Critically — and this confuses most non-specialists — the numbering is inverted: a lower class number means higher output power. PC3 is the default class at 23 dBm and is used by virtually all mobile handsets. PC2 steps up to 26 dBm for certain FDD/TDD configurations. PC1.5 — an intermediate class targeting bands n41, n77, and n79 — operates at 29 dBm in a 2Tx configuration.^[4] PC1, the pinnacle of the hierarchy, surpasses that threshold and is specifically architected for FWA customer premises equipment (CPE) and vehicle-mounted devices under 3GPP Release 17.^[5]

From a link budget perspective, transmit power is the dominant variable at the cell edge. The 3.5 GHz mid-band — including the 3.7 GHz n78/n77 frequencies used in this test — already imposes 4–5 dB more path loss than legacy 4G FDD bands and 3–4 dB additional penetration loss through building materials.^[6] At the cell edge, these compound. A standard PC3 device is essentially fighting physics; PC1 resets those odds substantially by driving more uplink energy toward the gNB's Massive MIMO receive beams.

The 10x uplink throughput gain over PC1.5 reported by Samsung is a cell-edge measurement — precisely where power headroom matters most.^[1] In radio link budget terms, every 3 dB of additional transmit power roughly doubles coverage area. The 40% coverage extension observed in the test is consistent with what the added power budget enables at mid-band frequencies when paired with Massive MIMO beamforming on the network side.

Why vRAN is the harder part of this story

The throughput numbers are compelling, but the deeper technical significance here lies in the platform on which PC1 was validated: a fully software-driven vRAN. Until this demonstration, PC1 capability had not been proven in a virtualized RAN environment.^[7] That gap mattered operationally because it meant operators pursuing vRAN architectures — for their flexibility, lower hardware dependency, and open interface advantages — could not simultaneously deploy premium uplink performance for FWA CPE.

Samsung's vRAN software stack, running against its 3.7 GHz Massive MIMO radio units, had to correctly interpret and process the higher-power uplink signals from the X85-equipped test device, manage interference budgets accordingly, and apply the appropriate power control loops as defined in TS 38.213.^[8] The fact that this works in a software-layer-only RAN environment confirms that the vRAN processing pipeline is now sufficiently mature to handle the full 3GPP PC1 specification — not just baseline performance modes.

The X85 chipset: what it brings to the equation

The Qualcomm Dragonwing FWA Gen 4 Platform — announced at MWC Barcelona in March 2025^[10] — pairs the X85 Modem-RF with a Hexagon NPU capable of up to 40 TOPS of AI processing, a quad-core application processor with hardware network acceleration, and a tri-band Wi-Fi 7 subsystem. The X85 itself is Qualcomm's 8th-generation modem-to-antenna solution and the first with an integrated dedicated tensor accelerator for 5G Advanced AI inference — enabling on-device optimization of latency, coverage, and power efficiency.^[11]

For PC1 operation specifically, the X85's support for 5G Advanced features including UL carrier aggregation with 4-layer MIMO on sub-6 GHz bands is a critical enabler.^[11] The chipset's Smart Transmit Plus technology manages elevated transmit power within SAR regulatory constraints — a non-trivial RF front-end challenge at PC1 power levels. The platform's mmWave 5G range of 14 km and 8Rx/6Rx antenna configuration further complement the uplink power story by enabling more diverse receive paths at the CPE itself.^[10]

The applications driving urgency

The timing of this announcement is not accidental. Physical AI — the embedding of AI inference directly into real-world systems like robotic arms, quality-control cameras, and autonomous logistics vehicles — generates continuous upstream telemetry that conventional FWA power classes struggle to sustain at the cell edge. Similarly, enterprise AR/VR deployments are inherently bidirectional: the upstream channel carries positional data, control signals, and live video feeds from the user's perspective, where latency on the uplink directly translates to session dropout and motion sickness.^[12]

3GPP Release 18's 5G Advanced framework — which further enhances support for CPE and FWA devices with up to eight uplink antenna ports and simultaneous multi-panel operation^[12] — will run in parallel with the PC1 commercial timeline, creating a compounding improvement in uplink architecture for FWA from 2027 onward.

Field trials and the path to 2027

Beyond the lab validation, Samsung and Qualcomm have already conducted field trials on a U.S. Tier 1 operator network — a significant step that distinguishes this from a purely controlled-environment demonstration.^[1] While the operator has not been named, Samsung's established vRAN relationship with Verizon makes that carrier a natural candidate.^[9] Commercial availability is targeted for 2027, aligning with the typical 18–24 month window between field trial and commercial feature introduction in the U.S. carrier market.

Bottom line

Power Class 1 on vRAN is a technically specific achievement with broad strategic consequences. It resolves the vRAN–uplink performance gap that has limited FWA architecture options. It delivers the uplink throughput needed for the next generation of AI-native and upload-intensive enterprise applications. And with U.S. field trials already underway, it's on a credible path to commercial deployment.

For operators, system integrators, and enterprise network architects evaluating FWA infrastructure in 2025–2027, this validation materially changes the capability conversation. The question is no longer whether 5G FWA can compete with fiber on uplink performance — at the cell edge, with PC1-equipped CPE on a modern vRAN, the answer is increasingly yes.