Joint Solution Enables Telcos to Deploy Small Language Models at Near-Edge Infrastructure, Powering Next-Generation AI Services with Industry-Leading Economics.
WASHINGTON, Oct. 28, 2025 /PRNewswire-PRWeb/ -- At NVIDIA GTC DC, Lanner Electronics and Personal AI today announced a strategic partnership to deliver a complete hardware and software platform enabling telecommunications providers to deploy advanced AI services at the network edge. The integrated solution combines Lanner's NVIDIA MGX-certified Edge AI servers with Personal AI's Small Language Model (SLM) technology to unlock new revenue streams for telcos while preparing networks for the 6G era.
As the telecommunications industry transitions to AI-native networks, telcos have an opportunity to monetize their extensive edge infrastructure investments. The Lanner-Personal AI platform directly addresses this by enabling service providers to offer AI-powered solutions to over 350 million telecom subscribers in the US.
AI-RAN: The Foundation for 6G Networks
The joint platform is purpose-built for AI Radio Access Networks (AI-RAN), a critical building block for next-generation 6G infrastructure. By consolidating AI workloads and RAN functions on shared, distributed computing infrastructure at near-edge hub sites, telcos can transform single-purpose cell sites into multipurpose AI factories that generate new revenue.
At Lanner, our mission is to accelerate edge-native AI deployment across telco infrastructure," said Jeans Tseng, CTO of Lanner Electronics. "Our carrier-grade, short-chassis Edge AI servers with scalable AI accelerators are pre-validated with Personal AI's Small Language Model technology, delivering a production-ready platform that telecom operators can deploy at existing cell sites. This collaboration not only shortens time-to-market but also empowers telcos to launch new AI-driven services and fully unlock the potential of their edge networks."
Small Language Models: A Quantum Leap in Edge AI Economics
At the core of the solution is Personal AI's approach to edge deployment using Small Language Models (SLMs); models ranging from hundreds of millions to 2 billion parameters. Validated by NVIDIA research demonstrating that SLMs are "sufficiently powerful, inherently more suitable, and accountably more economical" for agentic AI systems, PLMs deliver >20x better cost efficiency than traditional large language models while maintaining comparable accuracy for specialized tasks.
"Telcos are uniquely positioned to become the infrastructure layer for distributed AI, but legacy cloud-based LLM economics simply don't work at telco scale," said Jonathan Bikoff, Chief Business Officer at Personal AI. "Our Small Language Model platform, running on Lanner's edge-optimized hardware, enables telcos to offer AI services at 90% gross margins. These economics work for both the telco and their customers."
Technical Architecture: Built for Telco Edge
The platform centers on Lanner's ECA-6051 NVIDIA MGX Edge AI server, featuring:
- Intel Xeon 6 processors (up to 144 cores) for high-density compute
- Support for up to 3x NVIDIA L40S GPUs in air-cooled 2U form factor
- 1TB DDR5 6400MHz memory for concurrent model serving
- Carrier-grade reliability with NEBS Level 3 certification
- Sub-800ms latency for real-time voice AI applications
For far-edge deployments at cell sites, the solution also incorporates Lanner's EAI-I233 NVIDIA Jetson AGX Orin platform, enabling ultra-low-latency AI processing directly at the radio access network edge.
Privacy, Security, and Data Sovereignty
By processing Small Language Models at the telco edge rather than in centralized clouds, the platform addresses critical privacy and regulatory requirements. Customer Proprietary Network Information (CPNI) and sensitive business data never leaves the telco's infrastructure, ensuring compliance with FCC regulations and providing competitive differentiation against hyperscale cloud AI services.
The distributed architecture also enables telcos to support sovereign AI initiatives, allowing countries and regions to maintain data residency while still accessing state-of-the-art AI capabilities.
Media Contacts:
Lanner Electronics
Personal AI
Media Contact
Jonathan Bikoff, Personal AI, 1 +14155800085, [email protected], https://www.personal.ai/
SOURCE Personal AI
Share this article