IBM is out of the Public Cloud Infrastructure Magic Quadrant (IaaS + PaaS).
After several years among the “niche players,” Big Blue does not appear in the latest edition.
The momentum is more favorable for Alibaba Cloud. Still a “niche player” three years ago, it is now among the “leaders.” In this category, it sits alongside Amazon, Google, Microsoft… and Oracle, which continues to narrow the gap with the trio.
The Magic Quadrant is structured along two axes. One, called “Execution,” reflects the ability to meet demand (customer experience, pricing, quality of products/services…). The other, called “Vision,” focuses on strategies (sales, marketing, innovation…).
The situation on the “Execution” axis:
| Rank | Provider | Year-over-year change |
| 1 | AWS | = |
| 2 | = | |
| 3 | Microsoft | = |
| 4 | Oracle | = |
| 5 | Alibaba Cloud | = |
| 6 | Tencent Cloud | + 2 |
| 7 | Huawei Cloud | = |
On the “Vision” axis :
| Rank | Provider | Year-over-year change |
| 1 | = | |
| 2 | Microsoft | + 1 |
| 3 | AWS | – 1 |
| 4 | Oracle | = |
| 5 | Alibaba Cloud | = |
| 6 | Huawei Cloud | + 1 |
| 7 | Tencent Cloud | + 1 |
A Real Stack AI at Alibaba Cloud
From XuanTie CPUs and Zhenwu accelerators to Qwen models, and up to integration with its DingTalk communications service, Alibaba Cloud stands out for its “full-stack AI” approach. Gartner also praises its handling of hybrid cloud within the Apsara Stack. It also notes the suitability of its Cloud Parallel File Storage (CPFS) for AI workloads.
… but services with regionally variable availability
The engineering culture at Alibaba Cloud shapes its sectoral approach. It tends to produce technical metrics that are often hard to map to business indicators. Also beware of a partner ecosystem that is limited outside Asia. And of the unavailability of certain offerings (Lingjun AI Cluster, Qoder, Dataworks…) depending on geographic regions.
A multicloud pivot praised at AWS…
Last year, Gartner gave AWS credit for its mastery of the GenAI stack… excluding in-house models. It also praised its community, unparalleled in size in this market. And its operational scale, with a history of strong availability and a robust ability to provision resources, especially GPUs.
This year, AWS’s track record is highlighted from another angle: Amazon’s cloud unit is well positioned to capitalize on AI. Gartner points to the “agent stack” built around the Bedrock offering. It also notes the network effect of AWS’s market position on its partner network and talent pool. Another plus: the breadth of availability of AWS Interconnect – multicloud.
… but architecture and reliability questioned
Last year, the multicloud assessment was less favorable: minimal support, only a few components that could actually run outside the AWS cloud without specific hardware. Gartner also noted AWS’s lag in SaaS. And the lack of competitiveness of its Nova AI models.
This element remains relevant, as does the SaaS lag. Gartner questions AWS reliability, citing a major outage in fall 2025 with global impact. It also notes that some endpoints, such as those of the Marketplace, retain regional dependencies.
Google has an agentic stack that sets it apart…
Like Alibaba, Google has gone far in vertical integration in the AI domain. Gartner had highlighted this last year, praising a comprehensive sovereign cloud portfolio, including local controls (Data Boundary), an air-gapped option, and partner-operated dedicated regions.
This year again, Google is credited with a solid standing for its sovereign options. And for its AI stack “that sets it apart” (chips, data services, Gemini models, agentic components, SaaS integrations). It is also credited for its DeepMind subsidiary, a research and development engine that shapes its cloud offering, and for its contributions to open source (Kubernetes and TensorFlow).
… but beware the Gemini-centric approach
The legacy remains a sensitive point despite progress through the VMware partnership, Gartner noted last year. It also pointed to the lack of explicit and clear integration between GCP and Google Workspace. As well as inconsistent support and account management quality; especially for premium customers and for partners outside North America.
This year, the partner ecosystem is presented as a weakness in that it is less developed than AWS and Microsoft in certain industries and regions. Also pay attention to the emphasis placed on Gemini, where integrations and incentives tend to align. More broadly, innovation tends to focus on AI and data, at the expense of pure infrastructure.
Context layer, modernization and SDLC, Microsoft’s strengths…
Microsoft appears as the most capable of putting AI in the hands of end users, Gartner said last year. It also praised the integration of Azure services with the rest of its portfolio. And the level of support for multicloud (Azure Local, Azure Arc and Azure IoT).
This year, the firm highlights Microsoft’s “context layer” called IQ, which it views as a differentiator. Other strengths: modernization offerings for .NET and Java, described as “accessible and affordable.” On the SDLC ecosystem, it points to a development toolkit: coding assistant (GitHub Copilot), code hosting (GitHub), development environments (Visual Studio, VS Code) and continuous integration (Azure DevOps).
… but the house-built AI bricks still have to find their footing
On Azure, expenditure-management capabilities are limited, Gartner noted last year. It also cited capacity shortages across regions. And the ongoing dependence on third parties for the AI offering, beginning with OpenAI.
Capacity constraints remain real. Microsoft teams tend to be opaque about this, risking delays and added costs. As for in-house AI models, they still need to prove themselves. Like the Maia chips. On the sovereignty front, Azure Local lacks market traction, while Foundry Local remains in preview.
Oracle stands out again on multi-cloud…
Last year, Oracle benefited from a good point on distributed and sovereign cloud, thanks to its ability to maintain functional and price parity with the rest of the offering. Another positive: the level of support for multi-cloud. Both for network links (private connections between OCI and Azure / GCP at the time) and for databases (Exadata and Autonomous Database deployed with the “big three” cloud players).
The parity on deployment models, functional and price-wise remains. The same goes for multi-cloud; this year, with a focus on the database portion. Gartner adds Oracle’s AI strategy, mainly on three counts: price/performance ratio, a conscious choice not to monopolize R&D on in-house models, and a focus on the OCI Supercluster infrastructure for large-scale training.
… less on alignment with its partners
In last year’s Magic Quadrant edition, Gartner pointed to Oracle’s handling of the early 2025 incident that enabled, among other things, the exfiltration of customer credentials. It also noted that, unlike competitors offering sector-specific use cases directly on their platform, Larry Ellison’s company ships its solutions (Fusion, Industry Applications) separately, positioning OCI only as an infrastructure backbone.
This remark persists, even if expressed differently (a sectoral approach linked to the applicative offerings rather than being natively integrated into OCI). It is also accompanied by a warning about contractual transparency: some secondary costs are disclosed only late in the negotiation cycle. Also beware of a lack of alignment between Oracle and its partners, for sales and for support.
