AWS and Azure Launch Direct Private Interconnect in Public Preview

AWS and Microsoft build the bridge they said wasn't needed

AWS and Microsoft have jointly moved a direct networking link between AWS and Azure into public preview, combining AWS Interconnect – multicloud with Azure and Azure Multicloud Interconnect. The service provides private, MACsec-encrypted Layer-3 connectivity between AWS and Azure regions through a managed L3 fabric, initially available in US East (N. Virginia), US West, Sydney, and Frankfurt, with a roadmap to 100 Gbps per connection at general availability.

This is the first purpose-built, open-specification multicloud networking product of its kind from two hyperscalers. More importantly, it undercuts both companies' long-standing narrative that enterprise buyers didn't need deep cross-cloud networking. AWS and Microsoft are now building exactly the infrastructure they previously dismissed as unnecessary.

What this means for network architecture and vendor lock-in

Enterprises with significant estates on both AWS and Azure can now design direct, provider-managed private connectivity with MACsec encryption and targeted 100 Gbps throughput, instead of piecing together IPsec VPNs or third-party MPLS/SD-WAN. This lowers operational friction—one managed experience, standardized connectivity model—and increases design incentives to put different workloads on different clouds based on best-of-breed capabilities rather than sticking with one to avoid network pain.

The technical shift is material. Previously, cross-cloud connectivity meant going through the public internet with VPN encryption, using third-party colocation providers like Equinix Fabric or Megaport, or deploying carrier MPLS circuits. Each approach added latency, complexity, or cost. The AWS-Azure interconnect eliminates those trade-offs for the specific AWS-Azure pairing, with performance targets that match intra-cloud private connectivity.

Oracle already launched AWS Interconnect – multicloud with Oracle Cloud Infrastructure on July 29, 2026, providing high-speed private connectivity between AWS and OCI. The AWS-Azure announcement extends the pattern: hyperscalers are now competing on interoperability, not lock-in.

Budget implications and missing pricing

Public preview documentation emphasizes architecture and performance. Concrete per-Gbps or per-port pricing has not yet been published. Early buyers should treat consumption pricing as an open item and expect changes before general availability. Network spend can shift from carrier/colo interconnect and DIY VPN appliances to native cloud networking services once pricing is known, potentially consolidating contracts.

The lack of published pricing in preview means buyers should plan for a proof-of-concept or limited rollout in FY26-27 budgets but avoid assuming specific unit economics until GA rate cards are released. The availability in limited regions—N. Virginia, US West, Sydney, Frankfurt—means geography remains a constraint. Risk and architecture teams must validate disaster recovery and business continuity patterns in regions where the service is not yet offered.

Market data confirms multicloud is structurally normal

A cloud market-share analysis published September 14, 2026 reports AWS at 28% market share and Google Cloud at a record 15%, framing this shift alongside the AWS-Azure interconnect preview as evidence of genuinely multi-vendor cloud architectures. The Flexera 2026 State of the Cloud Report shows multicloud adoption increased by 2 percentage points year over year among organizations, confirming a continuing upward trend rather than plateau.

Falling AWS share and rising Google Cloud share, coupled with continued Azure strength, create a three-way competitive field where enterprises are more likely to standardize on two primary clouds plus a third for specific workloads, and demand data access and network interoperability rather than commit to single-vendor lock-in. Hyperscaler behavior is now aligned with this reality: they are building technical primitives for multicloud rather than actively resisting it.

What to watch

CIOs and CFOs can now justify formal multicloud strategies to boards using concrete adoption metrics—Flexera's +2 percentage points year over year—and market-share diversification. Procurement teams should expect more cross-cloud negotiations, such as discounts tied to multicloud use or cross-platform credits, as hyperscalers compete not just to be the only cloud but to win incremental workloads.

As multicloud becomes normal, networking, observability, and security line items need to be modeled for cross-cloud baselines—interconnect, multicloud logging, IAM federation—rather than single-cloud assumptions. Governance frameworks must be updated for multi-provider compliance, covering different regions, residency rules, and shared-responsibility models. The upside: reduced concentration risk is now supported by mainstream tools and market trends, not just vendor promises.

The AWS-Azure interconnect is a lagging indicator, not a leading one. The hyperscalers are acknowledging what enterprise architecture teams have already decided: multicloud is the default, and the network needs to catch up.