Container Platforms: A Market Starved of Innovation

Despite the lack of tools to manage them, container deployments on bare metal are expected to grow in 2025, given the upheaval in the virtualization market.

Gartner had put this prediction forward in the Magic Quadrant 2024 for container platforms, published in September. Nearly a year later, the firm finds that these deployments remain “relatively few in number.” There is growing interest, however, in deploying AI workloads, boosted by the tie-ins that vendors are building with DevOps toolchains.

Overall, progress in this market is now more about evolution than revolution, Gartner says. It also notes a “lack of innovators.” Commentary about players across the board reflects this: from year to year, they tend to look alike… quite a lot.

VMware is no longer “leader,” Huawei becomes one

The movement is more noticeable in how vendors are positioned within the Magic Quadrant. With gains on the “vision” axis—centred on strategies: commercial, marketing, geographic, vertical, etc.—Huawei moves from “challenger” to “leader,” joining Alibaba Cloud, AWS, Google, Microsoft, Red Hat, and SUSE. Broadcom (VMware) moves in the opposite direction.

Last year, twelve players were listed. This time there are fifteen, with the entry of Spectro Cloud, Nutanix (which built its offering on the basis of the D2iQ acquisition)… and a European player: Kubermatic (Germany).

On the “execution” axis, which reflects the ability to effectively meet market demand (customer experience, pre-sales performance, quality of products/services…), the situation is as follows:

On the “vision” axis:

Alibaba Cloud, still constrained by export restrictions

In 2024, Gartner praised the depth of Alibaba Cloud’s portfolio, especially in edge computing and hybrid cloud. It again highlights the functional breadth of its offering—unmatched outside American providers. It also notes the heavy usage within the Alibaba group, demonstrating its capabilities to operate at scale.

The American export restrictions that affected Alibaba Cloud last year have persisted. As a result, it does not possess the latest NVIDIA GPUs. The company also closed cloud regions in 2024 (India, Australia). And it saw its global IaaS market share decline, struggling to gain traction outside of China and Southeast Asia.

AWS remains limited on multicloud

As in 2024, AWS has the advantage of the depth of integration between its container platforms and its broader infrastructure and management services. Gartner again praises its partner ecosystem (larger than those of competitors) and the breadth of its serverless offerings.

The absence of native tools for fleet management remains a sticking point. Likewise, limited multicloud support (the main option being to deploy EKS Distro on other public clouds). And there is ongoing difficulty choosing between ECS and EKS depending on use cases.

Google and its modest footprint on the “legacy” stack

The AI Hypercomputer offering remains a differentiator for Google, with features others do not have, such as clusters up to 65,000 nodes. While last year Gartner focused on the simplicity of the catalog and fleet-management capabilities, this time it notes the American group’s leadership in open source—being the creator and main contributor to Kubernetes; involvement in related projects like Istio, Kubeflow, and Ray. It also notes Google’s tendency to meet needs with fewer services than competitors.

The criticisms this year resemble those of last year. On the one hand, limited presence in traditional lifting-and-shifting and application modernization. On the other, a MSP ecosystem that is smaller than that of rivals. Gartner also adds concerns about variable customer-support quality, especially outside the United States.

Huawei, in the same position as Alibaba

With a brisk pace of feature delivery, Huawei narrows the gap with other “leaders.” Like Google, its contribution to open source is notable (roughly 80 CNCF projects including Kmesh, KubeEdge, and Volcano). The Chinese group also benefits from a certain maturity in on-prem infrastructure and networking, which helps it address more accurately related use cases.

As with Alibaba Cloud, international sanctions continue to affect its activity. Ascend NPUs can stand in as an alternative to NVIDIA GPUs, though this creates potential interoperability issues. Huawei hardware—starting with its AI accelerators—does not always align with industry standards.

Microsoft, not the most advanced on edge and bare metal

For Microsoft, one enduring advantage cited last year remains: the integration points with GitHub and Azure DevOps. This is complemented by the management of AI workloads; in particular, the cost efficiency for customers of inference services.

Managing clusters on other public clouds or on vSphere remains a weakness. Microsoft also lags in edge, bare metal, and air-gapped deployments, while fleet-management experience is fragmented across multiple tools.

Beware lock-in with Red Hat

Red Hat continues to have a strong grasp of the market, and consequently its product aligns well with the majority of evaluated use cases. OpenShift sits at the core of its strategy and that of IBM.

From year to year, pricing remains relatively high (it’s hard to justify the costs for deployments that don’t require advanced features). And caution about lock-in remains relevant (the Advanced Cluster Management component provides interoperability, but it’s only included in the OpenShift Platform Plus subscription or otherwise as an add-on). This, as well as competition from partner hyperscalers, remains a consideration.

SUSE still struggles to differentiate

Rancher, by contrast, supports avoiding lock-in thanks to its history of supporting multiple distributions. This observation from last year still holds. It also notes competitive pricing and a strong position in edge deployments. Gartner adds a few elements about the roadmap, including developer experience and server-virtualization capabilities.

All the weaknesses flagged in 2024 remain again: difficulty differentiating, a lack of developer-focused features, and the potential impact of the evolving relationships with hyperscalers.