• Private GPU execution of hundreds of AI models on secure local clusters
• Hybrid cloud flexibility with OpenAI, Bedrock, Vertex AI compatibility
• Smart model routing via real-time selection by cost, speed, context, and data residency

Activeport Group Ltd (ASX:ATV) has announced the launch of its AI Gateway platform, which positions the company to deliver public and private cloud access for the execution of AI-driven applications. The new Activeport AI Platform combines unified routing intelligence, an enterprise-grade API, and secure, locally hosted GPU execution on private infrastructure, providing a sovereign, low-latency private cloud inference solution that can be integrated with public platforms like AWS Bedrock, Google Vertex, Cerebras, and Groq. Key features of the platform include private GPU execution of hundreds of models on sovereign, secure local GPU clusters, hybrid cloud flexibility with full OpenAI, Bedrock, and Vertex AI compatibility, and smart model routing via real-time selection by cost, speed, context window size, and data residency requirements. The launch marks Activeport's strategic expansion from GPU orchestration for cloud gaming into GPU orchestration for AI inference, positioning the company as a preferred software layer for private AI clouds across telecommunications, government, and enterprise sectors. Built on Activeport's proven telco-grade orchestration platform, the new AI solution transforms GPU clusters into private AI inference foundries, allowing customers to access models via standard APIs while maintaining full data sovereignty, compliance, and performance control.

Activeport is particularly excited about opportunities in hosting Diffusion models that can leverage the company's expertise in streaming optimization to boost performance.