DeepSig, a leader in AI-native wireless communications, participated in a discussion on the transformative role of AI-native Radio Access Networks (RAN) in wireless communications at NVIDIA’s GTC Conference. The conversation highlighted how AI-driven RAN enables networks to continuously learn, adapt, and optimize in real-time.
On March 20, 2025, in San Jose, CA, DeepSig CEO and Co-Founder Jim Shea joined industry and academic experts from NVIDIA, MIT WINSLab, Samsung Electronics, and others on the panel, “Defining AI-Native RAN for 6G.” The panel examined how AI will shape the next generation of wireless networks by enhancing efficiency, adaptability, and spectrum utilization.
DeepSig is driving a fundamental shift in AI-native wireless network development, leveraging real-time learning, spectrum awareness, and intelligent air interfaces to enhance network performance and efficiency. At Mobile World Congress Barcelona 2025, DeepSig demonstrated its AI-native air interface, OmniPHY®, running on the NVIDIA AI Aerial platform. By reducing and eliminating traditional pilot overhead and leveraging neural networks to jointly design core air interface functions, OmniPHY® achieved up to 70% throughput gains in certain scenarios while improving waveform power efficiency, highlighting the potential of AI-driven air interfaces in modern wireless networks.
“AI-native RAN is not just an incremental improvement—it’s a paradigm shift in how we design and operate wireless networks,” said DeepSig CEO Jim Shea. “We have a unique opportunity to rethink how networks learn, evolve, and self-optimize. Open collaboration, strong ecosystem partnerships, and advanced AI models will be critical to making next-generation networks more efficient, capable, and intelligent.”
“Our demonstration at MWC showed how AI-driven air interfaces can transform network efficiency, adaptability, and performance. At GTC, we will continue the conversation on how AI will accelerate the transition to 6G-ready networks,” added Shea.
DeepSig leveraged the NVIDIA Aerial Omniverse Digital Twin to highlight site-specific simulations, which illustrate how AI-native RAN can provide additional performance benefits within specific deployment scenarios. RAN digital twins allow DeepSig to train, test and improve AI-driven RAN models and performance in a simulation mirroring a specific real-world deployment.
AI-native RAN marks a breakthrough in wireless evolution, replacing rigid, hardware-defined networks with adaptive, software-driven systems and allowing software-driven systems to adapt and optimize end-to-end. DeepSig is at the forefront of this shift, transforming core communications functions with AI/ML to drive innovation and shape future 6G networks.