Tencent Cloud Enterprise launches private cloud for sovereign agentic AI
Tencent Cloud Enterprise has launched two private cloud solutions designed to help highly regulated industries run agentic AI within their own sovereignty boundaries. The Full-Stack Intelligent AI Suite and Lite Deployment combine infrastructure, model services, agent runtime controls and application support.
- Tencent Cloud Enterprise launched the Full-Stack Intelligent AI Suite and Lite Deployment.
- The suite covers infrastructure, TokenHub model services, agent runtime controls and applications.
- Lite Deployment supports a full-stack cloud from eight servers and AI-only deployments from three.
- The company says data, model weights and inference traffic remain within the customer's sovereignty boundary.
- Tencent Cloud Enterprise announced joint research reports with Gartner and Omdia.
A full stack from infrastructure to agents
The Full-Stack Intelligent AI Suite is Tencent Cloud Enterprise's flagship release of the year. The company says it represents a shift from a full-stack cloud to a full-stack intelligent cloud, with four selectable layers: a converged compute, storage and networking foundation; a model service layer; an agent runtime layer; and an agent application layer.
The model layer is built on TokenHub, Tencent Cloud Enterprise's private MaaS platform for large language model services. TokenHub turns computing resources into token services that can be called, metered and operated through a unified inference API. The company says the platform supports multi-tenant isolation and token-level metering for individual users and API keys.
The agent runtime includes a cloud-native sandbox, egress control and identity credentials. Tencent Cloud Enterprise says agent behavior is fully auditable, runtime environments can be provisioned in minutes, and unified training and inference scheduling can raise heterogeneous compute utilization by up to 60%.
Performance and sovereignty claims
Tencent Cloud Enterprise reports more than 250,000 TPM of measured single-accelerator throughput for its built-in inference acceleration, cache hit rates above 90%, and throughput gains of up to 220% on alternative silicon. These are company-reported figures.
The suite is compatible with mainstream x86 and ARM processors and multi-vendor accelerators. Tencent Cloud Enterprise says it has been validated at 10,000-GPU scale and supports financial-grade resilience with an RTO of no more than 2 minutes and an RPO of 0.
The company also says data, model weights and inference traffic never leave the customer's sovereignty boundary. Vice President of Tencent Cloud Kaibo Sha described the central issue for agents accessing core business data and critical systems as “where the data is processed, who operates the system, and what foundation the agent runs on,” according to the announcement.
Lite Deployment starts with eight servers
Tencent Cloud Enterprise Lite Deployment is aimed at organizations that want to begin with a smaller private cloud footprint and expand without re-platforming. A complete full-stack cloud can run from eight servers: three master nodes, three block storage nodes and two network function nodes. An AI-only deployment can start from three servers with private model services included, according to the company.
The same architecture can scale from a single availability zone to metro dual-AZ active-active deployment and then to cross-region disaster recovery. The software runs on existing hardware, requires no capital-heavy refresh and carries no minimum spend, Tencent Cloud Enterprise says.
The company cites a large futures brokerage that used this approach for a full-stack private cloud supporting core trading systems across heterogeneous chip architectures. The brokerage started its AI deployment on three servers and expanded as volumes grew.
Research points to demand for controlled AI infrastructure
Tencent Cloud Enterprise also announced joint research with Gartner and Omdia. Its guide with Gartner, A Practical Guide to Building Sovereign AI Cloud Platforms, presents sovereignty as a spectrum based on data, operations and technology. It outlines adoption paths for financial services, healthcare, public service cloud and telecommunications.
The Gartner research cited in the guide says that by 2028, strict data sovereignty and privacy regulations will compel more than 25% of enterprises in regulated industries to repatriate their primary GenAI workloads on-premises. It also says that through 2029, dependence on proprietary foundation models will leave more than 50% of GenAI adopters locked into a single public cloud provider's ecosystem, while only 17% of organizations report having enough talent to support AI adoption.
The Omdia report, Telco Cloud and AI Transformation: Global Market Opportunities and Case Studies, projects $160 billion in telco-led AI data center investment between 2026 and 2031. It identifies full-stack industry cloud platforms, national cloud infrastructure and AI monetization beyond GPU-as-a-Service as key transformation trends. Together, the launches position Tencent Cloud Enterprise's offering for customers that need private deployment, hardware flexibility and a path from small pilots to larger production environments.
