Alternative to VMware Tanzu (TKG) Series (5): Building Cloud-Native Infrastructure for AI Scenarios

To help users better understand which products on the market can replace VMware, this series introduces ZStack’s VMware replacement solutions, including product details, functional comparisons, technical features, and industry practices across virtualization, enterprise cloud, and container cloud platforms.

Previously, we introduced four alternative solutions to VMware Tanzu (TKG):

Article 1: VMware Tanzu (TKG) Alternative Series (1) Building with a Container Cloud Platform

Article 2: VMware Tanzu (TKG) Alternative Series (2) Unified Virtual Machine and Container Cloud Platform

Article 3: VMware Tanzu (TKG) Alternative Series (3) Building a Container Service Platform on Virtual Machines and Physical Servers

Article 4: Alternative to VMware Tanzu (TKG) Series (4) Building an Enterprise Cloud-Native Platform

In this article, we continue exploring ZStack’s solutions for replacing VMware Tanzu, focusing on applications in AI and intelligent computing scenarios.

Broadcom recently announced that it will end support for VMware vSphere 7.x and vSAN 7.x on October 2, 2025. After that date, customers will no longer receive product support, security patches, or updates—creating uncertainty around VMware’s long-term roadmap.

According to Gartner, more than 75% of VMware users are actively seeking alternatives due to concerns over rising license costs, audit restrictions, and unclear technology direction. ZStack’s replacement solutions not only meet VMware Tanzu’s functional requirements but also enable enterprises to build AI-ready, cloud-native infrastructures that overcome compute bottlenecks, accelerate R&D, and simplify operations.

Challenges of VMware Tanzu in AI Scenarios

Rising Costs

As VMware Tanzu shifted to per-vCore billing, AI cluster costs have increased significantly. AI training and inference require massive parallel computing, and growing core counts directly raise long-term TCO. This creates budget uncertainty for enterprises scaling AI capacity.

Coarse-Grained GPU Scheduling

AI and large model training heavily depend on GPUs. However, within VMware Tanzu environments, GPU scheduling is limited—often allowing only full-card allocation. This leads to high memory waste: if a virtual machine uses only 20% of a GPU’s memory, the remaining 80% sits idle and unavailable to others. Such coarse scheduling is inadequate for flexible compute allocation needs.

High Operational Complexity

In AI R&D and deployment, enterprises must manage CPU, storage, and GPU resources simultaneously. VMware Tanzu separates container and virtualization environments, forcing administrators to maintain multiple toolchains. This adds operational overhead and slows AI engineering iteration.

Core Enterprise Requirements for AI Infrastructure

During large-scale AI deployment, enterprises expect the new container service platform to:

• Overcome compute bottlenecks and improve GPU/heterogeneous resource utilization;
• Boost AI engineering efficiency for faster model training and iteration;
• Simplify operations through automation and unified management;
• Avoid vendor lock-in to ensure long-term flexibility and control.

These expectations drive enterprises to prioritize platforms capable of both AI compute optimization and cloud-native agility when seeking VMware Tanzu replacements.

Building Cloud-Native Infrastructure for AI and Intelligent Computing

(ZStack Zaku: Cloud-Native Scenarios for Replacing VMware Tanzu (TKG))

ZStack Zaku enables fine-grained GPU memory segmentation at the MB level without requiring additional software licenses, greatly improving heterogeneous compute utilization. It supports flexible scheduling of fragmented GPU resources for various AI workloads—overcoming resource bottlenecks, accelerating engineering iterations, and reducing operational complexity to achieve higher efficiency and lower costs.

AI development not only drives the evolution of intelligent computing centers but also creates a positive feedback loop with cloud-native infrastructure. Gartner notes that container management with GPU orchestration effectively allocates GPU resources in containerized environments for machine learning, AI, and generative AI (GenAI). It simplifies deployment and management of GPU-intensive workloads, with key use cases including container scheduling based on GPU requirements, dynamic GPU allocation, auto-scaling of GPU resources, and GPU lifecycle management.

Typical Application Scenarios

AI R&D and Large-Model Training

In AI labs and model R&D, ZStack Zaku helps significantly improve GPU utilization and reduce hardware investment. Automated scheduling and monitoring accelerate model training cycles, shortening time from research to deployment.

Intelligent Finance and Risk Control

Financial institutions rely on AI for massive data analysis and real-time risk control. Through ZStack’s GPU segmentation and scheduling, they can build elastic AI container platforms that remain compliant while supporting high-frequency trading and risk analysis.

Manufacturing and Smart Quality Inspection

Manufacturers handling large volumes of image and sensor data can deploy diverse AI models on ZStack platforms to enhance production intelligence while keeping costs under control.

ZStack: Let Every Company Have Its Own Cloud

Founded in 2015, ZStack is one of the leading cloud infrastructure software companies in China. Driven by R&D, ZStack delivers secure, stable, and controllable cloud infrastructure to enterprise customers worldwide.

ZStack holds a top-five position in China’s cloud system software market according to IDC, ranking first among independent vendors. Its success is rooted in a fully self-developed architecture, independent intellectual property, and complete source code ownership—not reliant on secondary development of open-source projects.

ZStack’s full-stack VMware replacement product line includes:

• ZStack ZSphere: Covers over 95% of VMware’s VM lifecycle management features with enhanced operation efficiency.
• ZStack Cloud: Enterprise-grade IaaS platform supporting large-scale compute, storage, and network resource pooling.
• ZStack Zaku Container Service Platform: Purpose-built for cloud-native and AI workloads, supporting multi-cluster, CI/CD, microservice governance, and GPU scheduling.
• ZStack Cube HCI Appliance: Integrates compute, storage, and networking for simplified deployment and delivery.

To date, ZStack has been deployed in over 30 countries, completing more than 1,000 VMware replacement projects across industries such as finance, manufacturing, energy, healthcare, education, and public services. With extensive practical experience and continuous innovation, ZStack is not only the ideal VMware Tanzu alternative but also a long-term partner for building future AI-driven, cloud-native infrastructures—empowering over 4,000 enterprises in their digital transformation journey.

FAQ

Q: Why do AI scenarios require a VMware Tanzu replacement?

A: Because VMware’s subscription model leads to high costs and insufficient GPU scheduling, which cannot support large-scale AI deployment.

Q: How does ZStack Zaku improve GPU utilization?

A: By MB-level GPU memory segmentation and fragmented scheduling, allowing multiple tasks to run concurrently on one GPU without resource waste.

Q: What challenges does ZStack solve in building intelligent computing centers?

A: ZStack overcomes compute bottlenecks, accelerates AI engineering iteration, and simplifies large-scale AI operations through unified management.

Q: Which industries benefit most from ZStack’s container platform alternative?

A: Finance, manufacturing, energy, healthcare, and education—especially where AI training, real-time inference, or data analytics are key.

Q: What experience does ZStack have in VMware replacement?

A: ZStack has completed over 1,000 VMware replacement cases across 30+ countries, offering proven expertise and industry depth.