MicroPhase Ships AntSDR T510 AI Combining RFSoC and Jetson
MicroPhase Technology has announced the AntSDR T510 AI, a single-board platform listed as "coming soon" on Crowd Supply that pairs an AMD Zynq UltraScale+ RFSoC ZU47DR with an NVIDIA Jetson module to perform RF capture, deterministic RF-domain processing, and GPU-accelerated AI inference on one board. Per the Crowd Supply listing, the board offers eight 14-bit ADC channels sampling up to 5 GSPS, eight 14-bit DAC channels up to 9.85 GSPS, an 8T8R synchronized transmit/receive architecture, and RF coverage from 1 MHz to 6 GHz. Platforms that colocate RF front ends and GPU acceleration reduce interconnect latency and simplify timing chains in phased-array, massive MIMO, and spectrum-sensing research, shortening development cycles for edge wireless systems.
What happened
MicroPhase Technology announced the AntSDR T510 AI, a crowdfunding campaign listed as "launching soon" on Crowd Supply. The platform integrates an AMD Zynq UltraScale+ RFSoC ZU47DR alongside an NVIDIA Jetson module on a single board. Per the Crowd Supply listing, the board provides eight 14-bit ADC channels sampling to 5 GSPS, eight 14-bit DAC channels to 9.85 GSPS, synchronized 8T8R operation, and direct RF coverage from 1 MHz to 6 GHz. Hackster covered the announcement, describing the board as capable of capturing, processing, analyzing, and responding to RF signals without external systems.
Technical details
Per the Crowd Supply page, the RFSoC handles deterministic RF tasks such as digital upconversion/downconversion, interpolation, decimation, and multi-channel synchronization, while the Jetson module (listed as Jetson NX at 15W typical power) supplies GPU-accelerated AI workloads. Each RF channel supports up to 2 GHz of baseband bandwidth. The design includes a 100G QSFP28 optical output, multi-board synchronization for scaling to 16, 32, or more channels, and typical power consumption of 45W under full 8-channel operation plus Jetson NX. Storage includes 4GB + 2GB DDR4, 32GB eMMC, and M.2 SSD expansion.
Software and open-source
The platform ships Ubuntu 22.04 with CUDA pre-configured. Bundled tools include WaveSight (multi-channel RF visualization and replay), GNU Radio and SoapySDR compatibility, and the open-source IQTAXI driver framework for RFSoC control and IQ data acquisition. A "SignalLab AI" demo provides GPU-accelerated real-time classification of Wi-Fi, Bluetooth, and modulation types from captured RF data. MicroPhase plans to publish hardware reference files, RFSoC firmware source, and Jetson integration examples in the public MicroPhase/T510-AI GitHub repository.
Industry context
Colocating high-speed RF data converters and local GPU acceleration is an emerging pattern for edge wireless research, because it reduces data-movement overhead and cross-device synchronization complexity in applications like phased-array radar, massive MIMO, and intelligent spectrum monitoring. Developers working on real-time ML-in-the-loop signal processing often face latency and bandwidth limits when splitting acquisition and inference across separate systems.
What to watch
Pricing and campaign launch date have not yet been announced on Crowd Supply. Practitioners should monitor SDK maturity - specifically the data path from RFSoC fabric into Jetson GPU - plus power and thermal performance under sustained wideband workloads. The 100G QSFP28 interface and multi-board synchronization path are notable for large-array prototyping.
Scoring Rationale
Interesting specialized hardware integrating RFSoC and GPU on one board, with a solid open-source software stack, but the campaign has not yet launched and the audience is narrow (RF/edge wireless research). Relevant to practitioners in phased-array, massive MIMO, and spectrum-sensing domains, but too niche and early-stage to rank above Solid.
Practice with real Telecom & ISP data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Telecom & ISP problems
