AWS and Google Launch Managed Cross-Cloud Networking, Ending Third-Party Dependency

AWS–Google managed connectivity turns multi-cloud networking into a cloud-native service

AWS and Google Cloud jointly released a preview of managed, private cross-cloud connectivity between their networks. The service provisions on-demand connections between Google Cloud VPCs and AWS VPCs in minutes, replacing what has been a bespoke IPsec tunnel or third-party SD-WAN build for most enterprises. AWS confirmed Microsoft Azure will join in 2026, creating a native three-cloud connectivity path.

The announcement changes the economics of multi-cloud networking. Instead of deploying routers, managing VPN appliances, or contracting with interconnect providers like Equinix Fabric or Megaport, enterprises can treat cross-cloud connectivity as a line item on their cloud bill. This moves spend from up-front capital expenditure on hardware and colocation to predictable operational expenditure under existing cloud agreements.

For enterprises that avoided multi-cloud because of networking complexity, AWS and Google have removed the primary technical objection. The service makes deliberate two-cloud strategies viable for workloads that previously required single-cloud architectures to avoid custom routing.

Impact on third-party network-as-a-service vendors

Third-party providers that built businesses around cross-cloud connectivity now face direct competition from the hyperscalers themselves. Equinix Fabric, Megaport, Colt's SD-WAN offerings, and newer NaaS platforms lose their value proposition in basic AWS-Google topologies. These vendors must differentiate on policy-based routing, observability across more than two clouds, or edge connectivity that extends beyond hyperscaler VPCs.

The move also pressures telco-cloud integrators and managed service providers that broker multi-cloud networking. When AWS and Google offer first-party, API-driven cross-cloud links, the case for outsourcing connectivity weakens unless the integrator delivers value beyond simple VPC peering.

Global enterprise cloud infrastructure spending exceeds $61 billion per quarter, according to Synergy Research Group data. A meaningful portion of that spend has gone to third-party networking and colocation to connect clouds. Native cross-cloud services capture that spend directly.

Budget and operational implications for enterprise buyers

Enterprise network teams can now reallocate engineering hours from managing custom multi-cloud tunnels to building cloud-native networking expertise. The operational risk of misconfigured cross-cloud routing drops when connectivity is centralized in each provider's console and API. Security teams gain clearer audit trails because traffic stays within managed hyperscaler networks rather than traversing third-party fabrics.

The trade-off is increased strategic lock-in at the networking layer. Once cross-cloud traffic flows through proprietary AWS–Google primitives, migrating to a third provider or leaving one of the two becomes harder. The 2026 Azure timeline gives enterprises a roadmap for three-cloud architectures, but it also creates dependency on hyperscaler connectivity services rather than portable, vendor-neutral networking.

Buyers should update RFPs to include native cross-cloud connectivity as a capability criterion when evaluating cloud providers. The question is no longer whether multi-cloud networking is feasible, but whether the enterprise wants to build it using cloud-native services or maintain independence through third-party fabrics.

Amazon and Anthropic's $25 billion multi-cloud AI bet

Amazon and Anthropic announced a $25 billion joint investment to develop AI infrastructure and chips designed for multi-cloud deployment. The partnership explicitly targets flexible AI services that operate across multiple cloud platforms, not just AWS. At the same time, OpenAI ended its exclusive hosting agreement with Microsoft Azure, enabling broader multi-cloud distribution of its models.

The investment signals that AWS will support cross-cloud AI deployments through Anthropic, countering the Azure–OpenAI pairing. For enterprises planning large-scale AI workloads, this reduces platform risk. Buyers can now negotiate AI consumption across multiple clouds rather than depending on Azure alone for OpenAI models or AWS alone for Anthropic.

The $25 billion figure reflects a broader $100 billion-plus investment race in AI chip manufacturing driven by multi-cloud strategies. While specific pricing is not yet public, the scale of capital suggests aggressive pricing or multi-cloud discounts will follow as AWS, Google, and Azure compete for AI workload share. GPU and accelerator supply constraints should ease over time, making multi-cloud AI strategies more feasible from a capacity standpoint.

What to watch

Track pricing announcements when AWS and Google move the cross-cloud connectivity service from preview to general availability. The per-connection cost will determine whether native connectivity undercuts third-party fabrics on economics or only on operational simplicity.

Monitor how Microsoft Azure integrates into the AWS–Google service in 2026. If Azure joins on equal terms, three-cloud native networking becomes the default architecture for large enterprises. If integration is limited or requires separate agreements, enterprises will face fragmentation between two-cloud and three-cloud strategies.

Watch for Anthropic pricing and deployment terms as the Amazon partnership scales. Enterprises that commit early to multi-cloud AI on AWS may gain pricing leverage, but they will also lock in architectural choices before the market matures. The OpenAI–Azure decoupling creates competitive pressure that should benefit buyers in the near term.