SINGAPORE, Sept. 21, 2026 (GLOBE NEWSWIRE) -- KAYTUS, a leading provider of AI infrastructure solutions, today announced a major upgrade to MotusAI, its enterprise AI platform. The release empowers enterprises to custom-build, govern, and scale on-premises “Token Factories,” deploy AI agents into production, and keep sensitive data within their own infrastructure, all while reducing annual token-related operating costs by 30%–50%.

The Challenge: Scaling AI Agents While Maintaining Control

As enterprise AI advances from simple LLM queries to autonomous multi-agent systems, tokens are becoming the computing currency powering core business workflows. Deploying agentic AI across the enterprise, at scale, brings three critical infrastructure challenges into focus:

• Security & Compliance Risks: Sending proprietary source code, customer records, and core business logic through public clouds, LLM APIs can expose sensitive data and create regulatory compliance risks. 
• Latency & Service Reliability: Multi-step agent reasoning and tool orchestration trigger unpredictable traffic spikes. Without elastic scheduling, compute resource bottlenecks delay Time to First Token (TTFT) and cause failed requests, compromising service-level agreements and user experience. 
• Uncontrolled Token Costs: Unmonitored model usage and missing departmental quotas drive escalating API costs, leaving enterprises without clear spending accountability across business units.

MotusAI: A Solid Foundation for End-to-End Token Lifecycle Management

MotusAI addresses these challenges with a secure, on-premises foundation unifying token production, distribution, and operations.

1. Enterprise Data Sovereignty

MotusAI keeps inference processing, model weights, and context within the enterprise security perimeter, eliminating reliance on public cloud implications. Organizations in financial services, healthcare, and government can scale AI while retaining control over sensitive data and compliance policies.

2. Reliable Service Quality

MotusAI transforms on-premises GPU clusters into a resilient token production engine, sustaining sub-second responsiveness even during peak demand:

• High-Speed Multi-Turn Inference: Prefill-Decode (PD) Disaggregation, dynamic KV caching, and dynamic batching reduce TTFT and end-to-end latency, enabling instant-fast and responsive multi-turn interactions. 
• Peak-Traffic Resilience: Integrated with inference runtimes such as vLLM and SGLang, MotusAI dynamically scales compute resources based on real-time telemetry of TTFT, Tokens per Second (TPS), and GPU utilization metrics. In production testing, autoscaling activated within 21 seconds of peak traffic, successfully expanded capacity to 16 instances within two minutes, automatically restoring SLA compliance. 
• Self-Healing Availability: Real-time cluster monitoring triggers automatic failover when node anomalies occur, maintaining round-the-clock availability for mission-critical agentic AI workflows.

3. Unified API Gateway for Faster AI Development

MotusAI enterprise-grade API gateway streamlines model deployment and access for internal development teams:

• Zero-Code Seamless Model Switching: OpenAI-compatible APIs let developers connect, test, and switch between open-source and commercial models without rewriting application code, preserving flexibility and avoiding vendor lock-in.
   
• Million-Token Context Support: Native long-context processing enables complex document analysis, advanced reasoning, and automated code generation.

4. Precise Governance & Lower Cost

MotusAI delivers end-to-end operational visibility to eliminate waste compute resources.

• Real-Time Performance Insights: Interactive dashboards provide tracking of latency, token throughput, and cache efficiency through live TTFT, TPS, and cache hit metrics, helping teams assess service level quality and optimize compute utilization. 
• Dynamic Resource Pooling: Fine-grained GPU partitioning and intelligent scheduling across workloads, maximize hardware efficiency, increasing utilization from 68.9% to 95.7% in benchmark tests. 
• Multi-Tenant Isolation & Chargeback: Administrators enforce departmental quotas, control access, and set internal billing rates to strengthen spending accountability. Filtering redundant requests enables enterprises to reduce annual token operating costs by 30%–50%.

Proven Successfully in Production Worldwide

MotusAI runs AI workloads in enterprise and commercial production environments across global markets:

• Financial Services: An overseas fintech firm replaced its existing platform with MotusAI across eight GPU servers, enabling metered token services with centralized governance for internal risk analysis and security compliance. 
• GPU Cloud Providers: A Japanese cloud provider runs its core platform on MotusAI, offering shared GPU resources and end-to-end training and inference workflows to more than 30 enterprise clients. 
• NeoCloud Operators: A Southeast Asian provider chose KAYTUS’s integrated hardware and software solution over an international competitor, using MotusAI’s built-in multi-tenancy and billing capabilities.

“Enterprises don't just need more GPUs—they need the ability to govern, settle, and scale token services reliably,” said Darren Cox, GM of KAYTUS Europe. “MotusAI bridges the gap between hardware and token operations, empowering organizations to run AI agents securely within their own data centers.”

Advancing the Future of Enterprise AI at Scale

With MotusAI, KAYTUS brings secure, high-throughput production on premises, helping enterprises protect sensitive data, operate independently of cloud APIs, and further increase the GPU utilization. The upgraded MotusA platform enables organizations worldwide to deploy and scale agentic AI securely and efficiently. 

About KAYTUS 

KAYTUS is a leading provider in AI infrastructure and liquid cooling solutions, delivering a diverse range of innovative, open, and eco-friendly products for cloud, AI, edge computing, and other emerging applications. With a customer-centric approach, KAYTUS is agile and responsive to user needs through its adaptable business model. Discover more at KAYTUS.com and follow us on LinkedIn and X

Media Contacts: media@kaytus.com

 

Tags：