AWS quietly did something on August 20, 2026 that networking engineers have wanted for years: it gave Direct Connect users fine-grained control over which routes flow into their AWS environment, at the same time its newer multicloud product, AWS Interconnect, kept adding partner clouds. Oracle Cloud Infrastructure joined the list in early August, following an April general-availability launch and a free-tier bandwidth offer in May. Put together, the moves mark AWS’s clearest attempt yet to own the plumbing between clouds, not just the compute sitting inside them. For an industry that spent the last five years debating whether multicloud is real or marketing spin, AWS just built a toll road for it.
This matters because the economics of connecting AWS, Azure, Google Cloud, and Oracle Cloud have historically been ugly: expensive dedicated circuits, per-GB egress fees that can blow up a monthly bill, and vendor lock-in dressed up as convenience. AWS Interconnect – multicloud, combined with a new free 500 Mbps tier and Direct Connect’s inbound prefix controls, changes that math for a meaningful slice of enterprise network architects. This piece breaks down what shipped, what it costs against Azure ExpressRoute and Google Cloud Interconnect, and what it signals about where cloud networking spend is headed in 2026 and 2027.
What AWS Actually Shipped in August 2026
Two separate AWS networking announcements landed in the same window and are easy to conflate, so it’s worth separating them. First, AWS Direct Connect introduced inbound prefix controls on August 20, 2026, a feature that lets customers allocate and manage route-prefix allocations for private and transit virtual interfaces (VIFs) based on workload needs. Engineers can now assign up to 1,000 prefixes each for IPv4 and IPv6 on dedicated and hosted connections, according to AWS’s own release notes. That’s a route-management upgrade for an on-premises-to-AWS service that’s been around since 2011.
Second, and more strategically significant, is AWS Interconnect – multicloud, a distinct product that went generally available on April 14, 2026. AWS describes it as offering “simple, resilient, high-speed private connections to other cloud service providers,” per the GA announcement. Unlike Direct Connect, which links a customer’s data center to AWS, Interconnect – multicloud links AWS directly to other public clouds without traffic ever touching the public internet. On August 3, 2026, AWS confirmed in its weekly roundup that Oracle Cloud Infrastructure connectivity through Interconnect had reached general availability, calling the product “the first purpose-built multicloud connectivity product of its kind.”
Inside AWS Interconnect – Multicloud: How the Plumbing Works
Technically, Interconnect – multicloud provisions a private circuit between an AWS region and a partner cloud’s network edge, bypassing the public internet entirely. That matters for two reasons: latency stays more predictable than routing through internet exchange points, and traffic never crosses a network AWS doesn’t control, which security teams tend to like when regulated data is involved. AWS positions the product as filling a gap that used to require third-party network-as-a-service vendors or manually stitched VPN tunnels between cloud VPCs.
The product is explicitly aimed at enterprises running hyperscale workloads across more than one provider, not small teams testing a side project. Oracle is the current headline partner, building on the two companies’ existing database interconnect work, but AWS’s language leaves room for additional CSP partnerships. AWS hasn’t published a full regional availability matrix, so engineers evaluating the service should check the console for their specific region and target cloud before committing architecture decisions to it.
The Free 500 Mbps Tier: What It Covers and What It Doesn’t
On May 29, 2026, AWS added a free 500 Mbps tier to Interconnect – multicloud, letting customers privately connect workloads on AWS and other public clouds with no recurring circuit fee up to that bandwidth threshold, according to AWS’s pricing update. That’s a real shift from how private cloud connectivity has been priced for the past decade. Azure ExpressRoute and Google Cloud Interconnect both charge a port fee starting at the lowest tier; AWS is charging nothing at all up to 500 Mbps.
The catch, unstated in AWS’s marketing but implicit in every free-tier cloud offer: 500 Mbps is thin for production database replication or large-scale data pipeline traffic between clouds. It’s plenty for control-plane traffic, API calls between microservices split across providers, or a proof-of-concept multicloud deployment. Teams moving real production bandwidth will graduate to paid tiers quickly, and AWS hasn’t published exact pricing beyond that free threshold in its public documentation, which makes budget planning harder than it should be for a service this consequential.
Direct Connect’s Inbound Prefix Controls, Explained for Network Engineers
The August 20 Direct Connect update solves a narrower but genuinely annoying problem. Before this release, network teams managing BGP route advertisements into AWS over Direct Connect had to lean heavily on router-side ACLs and prefix lists on their own edge equipment to control what got announced. Inbound prefix controls move part of that governance into AWS itself, letting teams allocate and cap route-prefix counts per virtual interface directly through the Direct Connect configuration.
Here’s roughly what that looks like from the AWS CLI side, for a team scripting the rollout across a fleet of virtual interfaces:
aws directconnect describe-virtual-interfaces \
--virtual-interface-id dxvif-xxxxxxxx \
--query 'virtualInterfaces[0].[virtualInterfaceId,routeFilterPrefixes]'
aws directconnect create-transit-virtual-interface \
--connection-id dxcon-xxxxxxxx \
--new-transit-virtual-interface '{
"virtualInterfaceName": "prod-transit-vif",
"vlan": 101,
"asn": 65000,
"mtu": 9001,
"directConnectGatewayId": "dxgw-xxxxxxxx"
}'With up to 1,000 prefixes supported per VIF for both IPv4 and IPv6, large enterprises running dozens of route-summarized subnets across hybrid environments get meaningfully more headroom than earlier defaults allowed. It’s an unglamorous feature, but unglamorous is exactly what keeps a Fortune 500 network team from filing a support ticket at 2 a.m.
AWS Interconnect vs Azure ExpressRoute vs Google Cloud Interconnect: Pricing Compared
The most useful thing for anyone budgeting a multicloud network is a straight pricing comparison, and here the picture is genuinely lopsided at the low end. Azure ExpressRoute’s Metered Standard circuit pricing, published on Microsoft’s own pricing page, starts at $55 a month for 50 Mbps and climbs to $436 a month for 1 Gbps, plus $0.025 per GB of outbound data in Zone 1. Google Cloud’s Dedicated Interconnect runs on an hourly model: $2.328 an hour for a 10 Gbps circuit, which works out to roughly $1,688 a month at continuous use, according to Google’s published pricing.
| Provider / Product | Entry Tier | Entry Price | Mid Tier (approx. 1 Gbps class) | Egress / Data Fee |
|---|---|---|---|---|
| AWS Interconnect – multicloud | Up to 500 Mbps | $0/mo (free tier) | Not publicly listed beyond free tier | Not publicly disclosed |
| Azure ExpressRoute (Metered) | 50 Mbps | $55/mo | $436/mo (1 Gbps) | $0.025/GB (Zone 1) |
| Azure ExpressRoute (Unlimited) | 1 Gbps | $5,700/mo | Included in flat fee | Included |
| Google Cloud Dedicated Interconnect | 10 Gbps | ~$1,688/mo ($2.328/hr) | 100 Gbps: ~$16,914/mo | ~$0.02/GB (typical) |
| Google Cross-Cloud Interconnect | 10 Gbps | ~$4,060/mo ($5.60/hr) | 100 Gbps: ~$21,750/mo | Separate egress applies |
Two things stand out. AWS’s free tier undercuts every rival at the entry level, full stop, at least for the 500 Mbps threshold it covers. And Google’s per-hour billing model for Cross-Cloud Interconnect, the product most functionally similar to AWS Interconnect – multicloud, starts well above AWS’s free allowance even before egress. Azure has no directly branded first-party multicloud interconnect product at all right now; ExpressRoute is built for on-premises-to-Azure connectivity, with Global Reach and partner arrangements filling some of the multicloud gap indirectly.
Feature-by-Feature: What Each Product Is Actually Built For
Pricing tells only part of the story, because these products aren’t functionally identical. AWS Direct Connect and Azure ExpressRoute both solve the same core problem: private connectivity between a customer’s own data center and a single cloud. Google’s Dedicated and Partner Interconnect do the same job for GCP. AWS Interconnect – multicloud and Google’s Cross-Cloud Interconnect solve a different problem entirely: private connectivity between two clouds, with no customer data center in the loop at all.
| Product | Connects | GA Date | Max Published Capacity | Confirmed Partner Clouds |
|---|---|---|---|---|
| AWS Direct Connect | On-premises ↔ AWS | August 2011 | Up to 400 Gbps (via Direct Connect Direct) | N/A (hybrid product) |
| AWS Interconnect – multicloud | AWS ↔ other clouds | April 14, 2026 | Not publicly disclosed | Oracle Cloud Infrastructure (GA Aug 3, 2026) |
| Azure ExpressRoute | On-premises ↔ Azure | May 12, 2014 | Up to 400 Gbps (ExpressRoute Direct) | N/A (hybrid product) |
| Google Dedicated Interconnect | On-premises ↔ GCP | September 7, 2017 | 100 Gbps circuits | N/A (hybrid product) |
| Google Cross-Cloud Interconnect | GCP ↔ other clouds | 2019 | 100 Gbps circuits | AWS, Azure, Oracle (per Google documentation) |
Read that table carefully and the competitive gap becomes clear: Google actually beat AWS to a first-party multicloud interconnect product by roughly seven years, launching Cross-Cloud Interconnect in 2019. AWS’s framing of Interconnect – multicloud as “the first purpose-built multicloud connectivity product of its kind” is, at minimum, aggressive marketing language. What AWS does have going for it is the free tier and, notably, Oracle as a partner cloud that both Microsoft and Google have already courted hard through Oracle Database@Azure, announced back in 2023.
Why Oracle Cloud Infrastructure Is the Partner That Matters
Oracle occupies an unusual position in the hyperscaler wars: it doesn’t have anywhere close to AWS, Azure, or Google Cloud’s overall market share, but it runs the database layer underneath a disproportionate share of enterprise workloads that live elsewhere. Oracle Database@Azure, launched in 2023, proved that customers wanted low-latency links between OCI’s database services and a hyperscaler’s compute and AI stack without a full migration. AWS connecting to OCI via Interconnect – multicloud extends that same logic to Amazon’s customer base, letting enterprises keep Oracle databases in OCI while running application logic, AI inference, or analytics on AWS with a private, low-latency link between them.
For Oracle, being the connective partner cloud across all three US hyperscalers turns OCI into something closer to enterprise database infrastructure than a standalone cloud competing for greenfield workloads. That’s a smaller, more defensible business than trying to out-scale AWS on raw compute, and it’s one Oracle appears to be leaning into deliberately.
Market Impact: Hyperscaler Growth Puts Pressure on Network Spend
This networking news lands against a backdrop of extraordinary hyperscaler growth. Alphabet reported Google Cloud revenue rose 82% year over year to $24.8 billion in its most recent quarter, while Microsoft posted 43% growth in Azure and Amazon reported 37% growth in AWS, according to reporting from The Motley Fool. Enterprise spending on cloud infrastructure passed $143 billion in Q2 2026, a year-on-year growth rate of 43%, according to Synergy Research Group figures cited by The Register. AWS still leads overall market share at roughly 31%, per the same Synergy tracking.
Analysts covering hyperscaler stocks flagged Azure’s commercial backlog specifically: it hit $678 billion, up 84% year over year, a figure that signals customers are pre-committing to years of cloud spend rather than buying month to month, per 247WallSt’s coverage of the earnings cycle. Growth at that scale, spread across three or four providers instead of consolidating into one, is exactly the condition that makes cheap, reliable interconnects between clouds commercially urgent rather than a nice-to-have.
The Egress Fee Problem Interconnect Products Don’t Fully Solve
Cheaper private circuits don’t automatically mean cheaper multicloud bills, and Google’s own pricing moves this year prove it. Google raised its networking egress prices for CDN Interconnect, Direct Peering, and Carrier Peering effective May 1, 2026, roughly doubling North American rates from $0.04 per GB to $0.08 per GB, with Europe rising to $0.08 per GB and Asia to $0.085 per GB, according to SpendArk’s breakdown of 2026 GCP pricing changes. That single change was described by the same analysis as “the single most consequential Google Cloud pricing event of 2026” for networking budgets.
The lesson for anyone excited about a free AWS interconnect tier: circuit fees and data-transfer fees are separate line items, and providers have shown they’re willing to raise one while cutting the other. A free 500 Mbps circuit that then routes through an egress fee that just doubled isn’t necessarily a net win. FinOps teams need to model both numbers together, not just the headline connectivity price.
Historical Context: From Dedicated Circuits to Multicloud by Default
The hybrid-cloud decade
AWS Direct Connect launched in August 2011, built to solve one problem: getting enterprise data centers a private, low-latency path into AWS instead of routing sensitive traffic over the open internet. Azure ExpressRoute followed in May 2014, and Google’s Dedicated Interconnect arrived in September 2017. For most of the 2010s, “cloud networking” meant hybrid connectivity between a company’s own servers and exactly one hyperscaler.
The multicloud pivot
Google broke that pattern first, launching Cross-Cloud Interconnect in 2019 to let GCP customers privately link to AWS, Azure, and Oracle. It took AWS until April 2026, seven years later, to ship an equivalent first-party product. That gap is notable given AWS’s market-leading position; it suggests AWS spent years betting customers would consolidate onto a single cloud rather than run production workloads split across several, and only recently accepted that multicloud architectures are now a durable enterprise default rather than a transitional phase.
Competitive Response: Will Azure and Google Move Next?
Microsoft is the most exposed of the three right now. It has no first-party product branded as directly as AWS Interconnect – multicloud or Google Cross-Cloud Interconnect; Azure customers wanting private links to AWS or GCP have had to rely on ExpressRoute Global Reach, third-party network-as-a-service vendors, or manually configured VPN tunnels. Given Azure’s $678 billion commercial backlog and its aggressive AI infrastructure push, a dedicated Azure multicloud interconnect product looks less like a hypothetical and more like a matter of when, not if.
Google, having shipped first, has the most to lose if AWS’s free-tier strategy pulls price-sensitive customers away from Cross-Cloud Interconnect’s hourly billing. Expect Google to either introduce its own bandwidth-limited free tier or lean harder on its existing egress-fee waiver programs to keep multicloud customers from defecting purely on entry-level pricing.
What This Means for Network Engineers and FinOps Teams Today
For teams actively running or planning multicloud architectures, the immediate action items are straightforward. Check whether AWS Interconnect – multicloud is available in your target region and whether your secondary cloud is on AWS’s partner list; right now that means Oracle Cloud Infrastructure. If you’re already paying for Google Cross-Cloud Interconnect or a stitched-together VPN solution purely for AWS-to-OCI traffic, the free 500 Mbps tier is worth testing against a non-production workload before committing budget elsewhere.
For teams managing existing Direct Connect deployments, the inbound prefix controls update is worth a router-config audit regardless of whether you touch multicloud interconnects at all. Moving prefix governance out of on-premises router ACLs and into AWS’s own configuration reduces one class of misconfiguration risk that’s caused real outages in the past when route-prefix limits were exceeded unexpectedly during a failover event.
Predictions: Where Cloud Interconnect Pricing Goes From Here
- Azure ships a first-party multicloud interconnect product within 12 months. The competitive gap against AWS and Google is now too visible for Microsoft’s enterprise sales teams to keep working around with partner-based Global Reach configurations.
- AWS publishes paid-tier pricing for Interconnect – multicloud by early 2027. A production-grade multicloud product without public pricing above 500 Mbps is an unusual gap for AWS to leave open for long, especially with enterprise procurement teams demanding cost predictability.
- More partner clouds join AWS Interconnect – multicloud beyond Oracle. Given the precedent Google set by connecting to AWS, Azure, and Oracle simultaneously, AWS is likely to add at least one more major partner cloud connection within the next few quarters.
- Egress and circuit pricing decouple further, and FinOps platforms race to model both. Products like North.cloud’s newly launched North v3, which explicitly targets unified cloud, AI, and data spend visibility, reflect growing enterprise demand for tools that can reconcile free connectivity tiers against rising per-GB data fees in one dashboard.
- Multicloud interconnect becomes a standard RFP line item for large enterprises by 2027. As backlog figures like Azure’s $678 billion signal multi-year enterprise commitments, procurement teams will increasingly demand interconnect roadmaps and pricing commitments upfront rather than discovering gaps after migration.
How This Compares to the Broader Cloud Networking Landscape
Put in context, this is one piece of a much larger networking buildout happening across every hyperscaler in 2026. Cloudflare has spent the year expanding its own edge and AI-agent connectivity products, Azure has been rolling out incremental updates to Event Hubs and Fabric connectivity, and AWS itself has been active on multiple networking fronts simultaneously, from Bedrock pricing cuts to Direct Connect route management. None of these individually rewrite the cloud market. Together, they describe an industry where the connective tissue between clouds, not just the compute or storage inside any single one, is becoming a genuine competitive battleground.
Frequently Asked Questions
What is AWS Interconnect – multicloud?
It’s an AWS networking product, generally available since April 14, 2026, that provides private, high-speed connections directly between AWS and other cloud service providers, bypassing the public internet. It’s distinct from AWS Direct Connect, which links customer data centers to AWS.
Is AWS Interconnect – multicloud free?
AWS offers a free tier up to 500 Mbps of multicloud connectivity, introduced May 29, 2026. Pricing for bandwidth above that threshold has not been publicly disclosed by AWS as of this writing.
Which clouds can I connect to with AWS Interconnect – multicloud?
Oracle Cloud Infrastructure reached general availability as a connected partner cloud on August 3, 2026. AWS has not published a full list of additional partner clouds at this time.
How does AWS Interconnect – multicloud compare to Azure ExpressRoute?
They solve different problems. ExpressRoute connects on-premises data centers to Azure, starting at $55/month for 50 Mbps and $436/month for 1 Gbps. AWS Interconnect – multicloud connects AWS directly to other clouds like Oracle Cloud Infrastructure, with no equivalent first-party Azure product currently available.
What are AWS Direct Connect’s new inbound prefix controls?
Launched August 20, 2026, this feature lets customers allocate and manage inbound route-prefix allocations for private and transit virtual interfaces, supporting up to 1,000 prefixes each for IPv4 and IPv6 on dedicated and hosted connections.
Does Google Cloud have a multicloud interconnect product too?
Yes. Google Cross-Cloud Interconnect launched in 2019, seven years before AWS’s equivalent product, and connects GCP to AWS, Azure, and Oracle Cloud Infrastructure, priced from roughly $5.60 an hour for a 10 Gbps circuit.
Why is Oracle Cloud Infrastructure connected to all three major hyperscalers?
Oracle has positioned OCI as a database-layer partner cloud rather than a direct compute competitor, building on precedents like Oracle Database@Azure, announced in 2023. Being interconnected with AWS, Azure, and Google Cloud keeps enterprise Oracle database customers in OCI without forcing a full migration.
Will multicloud interconnect pricing get cheaper?
Circuit pricing is trending down, evidenced by AWS’s free tier, but egress and data-transfer fees are moving the opposite direction in places, as shown by Google’s May 2026 egress price increase. Total multicloud networking cost depends on both figures together, not circuit pricing alone.
Related Coverage
- AWS vs GCP vs Azure: GCP Cuts SQL Costs 30% [2026]
- ECS vs EKS: $0 vs $438/Mo Control Plane [2026]
- Azure Kubernetes Security GA as Backlog Hits $678B [2026]
- AWS Outage Hits 28 Hours, Third us-east-1 Failure [2026]
- Google Cloud Hits 63% Growth, Outpaces AWS, Azure [2026]
- AWS Lambda Serverless Tutorial: 12 Steps, 45 Min [2026]
For more cloud computing news and analysis, visit the Cloud Computing section.
