Amazon Web Services (AWS) y Google Cloud They launched AWS Interconnect – multicloud on December 2, 2025, joining a wave of multicloud alliances that are redefining the business infrastructure market. Oracle expanded in October 2025 Oracle Database@Azure to 28 regions and announced Multicloud Universal Credits, while Cloudflare integrated its platform with Oracle Cloud Infrastructure in October and acquired Nefeli Networks in 2024 to launch Magic Cloud Networking. 89% of companies operate multicloud strategies according to Flexera 2025, driven by AI workloads that require distributed access to data across multiple providers.
The solution eliminates weeks of manual configuration that traditionally required establishing private connectivity between clouds. AWS Interconnect – multicloud combines pre-built capacity pools with encryption MACsec IEEE 802.1AE at a physical level and quadruple redundancy between facilities. Customers can provision 10 Gbps or 100 Gbps connections via web console, CLI or API, reducing time-to-production for critical multicloud architectures. The architecture allows databases in AWS Aurora communicate efficiently with analytics in Google BigQuery through dedicated private connections that bypass the public internet.
Oracle adopted a radically different strategy: physically deploy Oracle Exadata within AWS, Azure and Google Cloud data centers. Oracle Database@AWS reached general availability in July 2025 in Virginia and Oregon, expanding in December to Frankfurt, Tokyo and Ohio, with 20 additional regions planned for 2026. Oracle Database@Azure operates in 28 regions from October 2025. Oracle Database@Google Cloud It is available in 8 regions with 9 additional ones planned. This model offers microsecond latency between Oracle Database and hyperscaler services (Amazon Bedrock, Microsoft Fabric, Google Vertex AI) when removing network hops. The fundamental difference: AWS-Google Cloud provides L3 connectivity between separate clouds, while Oracle runs workloads directly within the competitor’s infrastructure under OCI management.
Agreements between competitors
89% of companies operate multicloud strategies according to Flexera 2025, driven by AI workloads that require distributed access to datasets across multiple providers. Robert Kennedy, VP of network services at AWS, defines the solution as a fundamental change in multicloud connectivity. Salesforce became the first marquee customer. Jim Ostrognai, SVP Software Engineering, confirmed that AWS Interconnect allows you to establish critical bridges to Google Cloud as easily as deploying internal AWS resources. Salesforce operates Hyperforcecloud-native architecture that requires integration with heterogeneous IT ecosystems of Fortune 500 clients.
Cloudflare acquired Nefeli Networks in 2024 and launched Magic Cloud Networkingproviding a unified orchestration layer to connect and secure public cloud environments. In October 2025, Cloudflare natively integrated its platform with Oracle Cloud Infrastructure (OCI)enabling joint customers to leverage Cloudflare security, performance, and resiliency directly from OCI across hybrid and multicloud architectures. According to Forrester at Cloudflare Connect 2025, 73% of enterprise networking decision-makers adopt multicloud networking. Cloudflare positions its connectivity cloud as the connective tissue for modern digital operations, abstracting complexity from DNS, digital certificates and routing through its global network of 330+ edge locations. The difference with AWS-Google: Cloudflare operates as a neutral layer on multiple hyperscalers simultaneously, while AWS Interconnect specifically connects two peer-to-peer providers.
Oracle announced in October 2025 Multicloud Universal Creditsthe industry’s first flexible cross-cloud consumption model. The credits are usable in Oracle Database@AWS, Oracle Database@Azure, Oracle Database@Google Cloud y OCIallowing simplified procurement with consistent contracts between all clouds. Dave McCarthy, Research VP at IDC, notes that procurement and governance are traditionally roadblocks for innovation. Oracle eliminates contractual friction: Enterprises can provision Oracle AI Database in any AWS, Azure, Google Cloud, or OCI region under a single consumer agreement. Customers can purchase via private offers in each hyperscaler’s marketplaces, activating reseller networks. This represents a procurement change more than a technological one: the technical integration Oracle-hyperscalers already existed, Universal Credits eliminates administrative barriers.
Industrial validation of the service
Salesforce became the first marquee customer. Jim Ostrognai, SVP Software Engineering, confirmed: “AWS Interconnect – multicloud allows us to establish these critical bridges to Google Cloud as easily as we deploy internal AWS resources.” Salesforce operates Hyperforcecloud-native architecture that requires integration with heterogeneous IT ecosystems of Fortune 500 clients. The ability to connect AWS (where Hyperforce primarily hosts) with Google Cloud (where clients operate data warehouses) without physical network management overhead directly reduces operational complexity in enterprise deployments.
Los AWS Transit Gateways They operate as region-scoped services. A TGW in us-east-1 only connects to Interconnects in corresponding us-east-1, forcing multi-region architectures to provision separate connections per region pair. Managing IP address spaces requires coordination: Amazon VPC CIDR blocks cannot overlap Google Cloud VPC ranges. Oracle Database@ solves this differently: when running within the hyperscaler data center, latency is microseconds versus 60-120ms for AWS-Google Cloud cross-region. However, Oracle Database@ requires physical deployment of Exadata, limiting geographic flexibility versus on-demand provisioning from AWS Interconnect. Vendor lock-in migrates from physical connectivity to managed services: database abstractions, serverless platforms, and AI services remain vendor-specific in both models.
Multicloud expansion roadmap in 2026
AWS Interconnect – multicloud is in preview in five region pairs: US East (Virginia), US West (Oregon and California), EU West (London) y EU Central (Frankfurt). Microsoft Azure It will be integrated in 2026 according to CIO Dive. Oracle Database@AWS plans 20 additional regions in 2026 including Canada, Hyderabad, Ireland, London, Melbourne, Milan, Mumbai, Osaka, Paris, São Paulo, Seoul, Singapore, Spain, Stockholm, Sydney and Zurich. Oracle Database@Azure will add Brazil Southeast, France South, North Central US, South India and West Europe. Leo Leung, Group VP of OCI, told CIO Dive that 70% of workloads remain on-premise, allowing cooperation between competitors to capture cloud migrations without cannibalizing existing business. AWS and Google Cloud published the Connection Coordinator API Specification under OpenAPI 3.0 for third party integration.
The multicloud wave from October to December 2025 marks industrial convergence: AWS-Google Cloud L3 connectivity, Oracle Database@ physical deployment inside-hyperscaler, Cloudflare neutral orchestration layer, and Oracle Multicloud Universal Credits contractual simplification They represent complementary, not competitive, strategies. Each solution addresses different use cases: AWS Interconnect for distributed workloads requiring on-demand connectivity, Oracle Database@ for microsecond latency in analytics-database integration, Cloudflare Magic Cloud Networking for unified multi-vendor management, Universal Credits for friction-free procurement.
The Register points out the irony: AWS and Google Cloud defended before the UK Competition and Markets Authority that there were no technical barriers to multicloud interoperability, and then launched a specific solution solving precisely those barriers. The reality: vendor lock-in operated at the level of operational complexity and contractual friction, not physical or technical capacity. The 2025 initiatives eliminate these barriers through automation (AWS Interconnect), co-location (Oracle Database@), abstraction (Cloudflare), and unified procurement (Universal Credits).
For enterprise architectures with multi-region disaster recovery requirements, heterogeneous cloud service dependencies, or GDPR/NIS2/DORA compliance, These solutions represent fundamental infrastructure. Confirmation of Azure integration into AWS Interconnect 2026, Oracle Database@ expansion to a combined 40+ regions, and Cloudflare OCI adoption suggest that managed interoperability is becoming a commodity. The 70% of on-premise workloads cited by Oracle’s Leo Leung indicates a large enough market for hyperscaler cooperation without cannibalization, prioritizing capture of cloud migrations over zero-sum competition.
Beyond technical efficiency and compliance, this convergence towards interoperability responds to a strategic need for survival: mitigation of risk due to dependence on a single supplier or vendor lock-in. According to the Flexera 2025 State of the Cloud Report, 89% of organizations has already adopted multicloud strategiesa trend accelerated by the demand for operational resilience in the face of systemic falls and the concept of “explosion radius.” In the European context, regulations such as DORA and NIS2 have transformed infrastructure diversification from a competitive advantage into a legal compliance requirement, forcing companies to avoid concentrating risks in a single hyperscaler to guarantee business continuity in the face of critical contingencies.
In parallel, interoperability is redefining cloud economics through the FinOps discipline. By removing administrative barriers through models like Oracle Universal Credits and facilitating technical connectivity, Companies gain freedom of movement that was previously unthinkableallowing them to shift workloads to the provider that offers the best cost-performance ratio at all times. This “cloud arbitrage” capability turns infrastructure into a fluid commodity; Organizations are no longer purchasing space in an isolated cloud, but rather access to a distributed ecosystem where on-demand provisioning and unified governance can dramatically reduce resource waste and optimize return on investment in high-intensity AI projects.
