At IBC 2026, SDMC showcased an AI Home architecture designed to combine cloud intelligence, Edge AI and emerging Physical AI capabilities across connected home devices and services. The demonstration comes as SDMC works with Google Cloud through its Proactive Physical AI Acceleration Program, with a focus on technology co-creation, scenario validation and real-world service pilots.
The next phase of consumer AI may not be defined by a single smart speaker, television or home appliance. Instead, intelligence could increasingly operate as a distributed layer across devices, networks and computing resources inside the home.
That is the architecture SDMC presented at IBC 2026, combining cloud-edge intelligence, multimodal AI, AI agents and shared edge computing into a broader platform for connected-home services.
The Hong Kong-listed technology provider is also participating in Google Cloud’s Proactive Physical AI Acceleration Program, where the companies are working across technology development, scenario validation and service pilots.
The collaboration is aimed at extending AI Home services across areas including entertainment, security and other emerging connected-home applications.
Moving AI From Devices to the Edge
At the center of SDMC’s architecture are Cedar and AI Station.
Cedar combines a multimodal AI model with an AI agent designed to understand context and coordinate actions across devices and services. SDMC says the system is built on Google’s Gemma open model, giving it an edge-oriented foundation for multimodal understanding and reasoning.
When broader reasoning capabilities are required, Google’s Gemini models can complement processing in the cloud.
That creates a hybrid architecture rather than requiring every AI task to be processed remotely.
The distinction is important for home environments, where latency, connectivity and privacy can affect the usability of AI services. Processing sensitive or time-critical workloads locally can reduce dependence on cloud round trips, while cloud infrastructure can provide additional computing resources when local processing is insufficient.
Shared Edge Compute for Multiple AI Applications
SDMC’s AI Station addresses another challenge: the hardware requirements associated with running increasingly capable AI applications.
The system uses NVIDIA Jetson T5000 hardware as shared edge infrastructure for Cedar and other AI workloads. Rather than placing a dedicated high-performance AI computer inside every connected endpoint, multiple applications can share a common computing resource.
The approach resembles an edge-computing model already emerging in industrial and enterprise environments, but applies it to residential and operator-managed ecosystems.
SDMC says AI Station is intended for operators, small businesses and premium households, with latency- and privacy-sensitive workloads remaining within the local environment.
For telecommunications operators, this architecture could also create a way to add AI capabilities to existing broadband and connected-home infrastructure without replacing every endpoint.
AI Across Entertainment and Network Operations
The IBC demonstrations extended beyond conversational AI.
Together with Amlogic, SDMC demonstrated Edge AI capabilities in an AI Home Display and AI Streaming Box, while a Google TV 4K Projector provided a large-screen entertainment component.
The company also demonstrated AI applications for network operations.
Its AI Operation technology is designed to assist with network diagnosis, proactive optimization and more autonomous operations and maintenance. Meanwhile, the AI Fiber Gateway combines connectivity hardware with AI capabilities intended to improve network performance and operational efficiency.
This operational layer is significant because AI Home infrastructure is not only about how consumers interact with devices. For operators, the economics can depend equally on how efficiently those devices, networks and services can be managed at scale.
From Generative AI Toward Physical AI
SDMC is also positioning its architecture as a foundation for Physical AI, where AI systems interact with physical environments rather than remaining confined to software interfaces.
Potential applications cited by the company include connected-home control, autonomous cleaning and robotic manipulation.
The transition from digital AI to Physical AI introduces additional technical requirements. Systems need to perceive their surroundings, reason about physical conditions and translate decisions into actions through connected hardware.
SDMC’s cloud-edge approach could provide a distributed computing layer for those applications, although the company’s Physical AI scenarios remain an area of ongoing development rather than established mass-market deployments.
The Operator Opportunity
The broader commercial opportunity lies in turning AI Home from a collection of individual AI-enabled products into a coordinated service platform.
Telecom operators already control important parts of the home technology stack, including broadband connectivity, gateways, streaming services and connected devices. An edge AI architecture could allow those operators to introduce new AI services across existing infrastructure while keeping some processing close to users.
The challenge will be balancing AI performance, hardware costs, interoperability, privacy and deployment complexity.
SDMC’s IBC showcase illustrates one approach: use edge computing for responsive and privacy-sensitive workloads, cloud infrastructure for more demanding reasoning, and an agent layer to coordinate actions across the home.
If that architecture can be deployed economically, the AI Home could evolve from a collection of smart endpoints into a distributed AI computing environment capable of understanding context, coordinating devices and eventually interacting with the physical world.
Market Landscape
The consumer AI market is moving toward multimodal, agentic and edge-based computing. While cloud AI remains essential for large-scale model inference, edge processing is increasingly relevant for applications requiring low latency, privacy or continuous operation.
Smart-home platforms face an additional integration challenge because homes contain devices from multiple manufacturers and technology generations. Shared edge infrastructure and software agents could provide an abstraction layer between those devices and AI services.
The emerging Physical AI category adds another dimension. Robots, autonomous appliances and intelligent home systems require AI to connect perception and reasoning with real-world action.
For telecom operators, these technologies potentially create a new service layer on top of broadband and connected-home infrastructure. The commercial question is whether operators can deliver useful AI experiences at sufficient scale without creating excessive hardware and operational complexity.
Top Insights
- SDMC is combining cloud AI, Edge AI and emerging Physical AI into a distributed architecture for connected-home services.
- Cedar uses multimodal AI and an agent layer to understand context and coordinate actions across devices and services.
- AI Station uses NVIDIA Jetson T5000 hardware as shared edge infrastructure for multiple AI workloads.
- Google Gemini can complement edge intelligence when applications require broader cloud-based reasoning capabilities.
- SDMC’s operator-focused strategy extends AI beyond consumer interaction into network diagnosis, optimization and autonomous operations.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
