Today, 7 May 2026, AWS Interconnect — multicloud is generally available, the first managed AWS-to-other-cloud private connectivity product the industry has ever shipped, and as of this month a free 500 Mbps interconnect tier per Region begins for every AWS customer. Late April’s GA paired with the May free-tier launch turns a story that has bounced around since AWS re:Invent 2025 into a live decision for UK SMEs running cloud workloads across more than one provider — and it lands at exactly the moment the CMA’s UK cloud verdict has put multicloud friction on the regulatory record.
The headline is simple: AWS and Google Cloud customers can now stand up a quad-redundant, MACsec-encrypted, layer-3 private interconnect between their AWS and GCP estates from the AWS Direct Connect console in minutes, not the weeks or months it used to take with VPN tunnels, colocation cross-connects and third-party SD-WAN fabrics. London is in the GA Region list. Frankfurt is the European fallback. Microsoft Azure and Oracle Cloud Infrastructure are not yet shipping but have publicly committed to adopt the open spec later in 2026. This article unpacks what the launch is, what it means for UK SMEs, where the £0 free tier actually applies, the workloads it does and does not cover, the cost staircase compared with VPN and Direct Connect, and the 30-day SME audit and 12-month plan to get ahead of the Azure cut-over when it lands.
What just launched — and what is still pending
AWS Interconnect — multicloud reached general availability in late April 2026 (AWS’s What’s New page is dated 14 April; the broader GA announcement, including the open-spec drop on GitHub, lands on 29 April per InfoQ). The freshest hook for this month is that the free 500 Mbps local interconnect tier per Region started in May 2026, which is the actual cost-saving moment for UK SMEs that already run a cross-cloud tunnel. The two parts of the AWS Interconnect umbrella are easy to confuse so it is worth being explicit:
AWS Interconnect — multicloud is the new AWS-to-other-cloud-provider managed private connectivity service. It uses physically redundant interconnect facilities and MACsec-encrypted edge routers between AWS and the partner cloud. Customers do not lay fibre, manage colocation, or operate any third-party network fabric. Setup is a console workflow that resolves to a layer-3 private path with quad redundancy. The first launch partner is Google Cloud; Azure and OCI are committed but not yet GA. The interconnect plumbing is exposed as an open spec on GitHub at github.com/aws/AWSInterconnect, so any cloud provider can implement the partner side without bilateral negotiation — that, more than the AWS↔GCP relationship itself, is what reframes this as a standards play rather than a single-vendor deal.
AWS Interconnect — last mile is the sister service announced alongside multicloud. It targets branch offices, data centres and remote locations, providing managed high-speed private connections from those sites into AWS via participating network providers. UK SMEs that run a hybrid estate — a head office in Slough, an Azure-hosted ERP, an AWS analytics workload — will end up using both: last mile for the office-to-AWS leg, multicloud for the AWS-to-Azure leg once Azure ships its side.
Most UK SMEs are Microsoft-first: Microsoft 365 for productivity, Entra ID for identity, Azure for line-of-business. The new AWS Interconnect spec is committed by Microsoft and Oracle but not shipping today. Microsoft has not published a firm GA date beyond “later in 2026.” If your multicloud reality is Azure↔AWS or Azure↔GCP, today’s GA does not yet replace your existing ExpressRoute, VPN or third-party fabric — but you should pre-design the migration path now so the cut-over is a rotation rather than a re-architecture. Treat the Azure interconnect ship date as the trigger event in your 2026 networking plan, not the optional polish item.
The 5-Region GA footprint and why London matters for UK SMEs
AWS Interconnect — multicloud went GA across five AWS Regions, paired with matching Google Cloud regions. London and Frankfurt are both in the launch list, alongside US East, US West and a fifth pair covering the broader US footprint. For a UK business, London being a day-one Region is the difference between a clean go decision and a wait-and-see. Cross-cloud traffic between an AWS London VPC and a Google Cloud europe-west2 (London) project can now stay in-region rather than routing transatlantic, which matters for three reasons UK SMEs care about: data sovereignty (UK GDPR, ICO expectations on personal data residency), latency (analytics ETL, model inference, replication windows), and egress cost (interconnect-priced traffic versus public internet egress).
Frankfurt is the European fallback, which means SMEs running EU workloads can keep traffic on EU soil with the same plumbing and the same console. This matters in two specific cases: a UK SME with an EU-resident customer base under data-residency contractual commitments, and a UK SME running a Frankfurt-hosted disaster-recovery copy of a London production workload. Both of those used to require explicit colocation contracts in Telehouse Frankfurt or Equinix FR2 and a careful network engineering design; both are now on-console operations.
The fifth Region is, per AWS’s materials and InfoQ’s coverage, paired with a corresponding Google Cloud region in the broader US/Asia-Pacific footprint. The point for UK SMEs is not the specific fifth Region — it is that AWS has gone for global parity in the GA wave, which signals that follow-on Regions and partner clouds are sequenced rather than hopeful. Customers who delay because “Sydney isn’t in the list” or “Mumbai isn’t in the list” today can plan against a roadmap, not a wait.
How the new path replaces the old one — technically
The legacy UK SME cross-cloud reality has been one of three increasingly painful options. Option one was a site-to-site IPsec VPN over the public internet between an AWS Transit Gateway and a Google Cloud Cloud Router. Cheap to start, slow under load, no SLA, variable latency, encryption overhead on every packet, and a constant operational tax on whoever owns the BGP session. Option two was AWS Direct Connect plus Google Cloud Partner Interconnect via a colocation provider — Megaport, Equinix Fabric or similar — with cross-connects running between the two clouds inside a metro carrier hotel. Reliable, fast, but costing thousands of pounds a month in port and cross-connect fees and taking weeks to provision. Option three was an SD-WAN overlay (Cisco Meraki MX, Fortinet Fabric, Aviatrix) running cross-cloud tunnels with vendor-managed orchestration. Better operationally than DIY VPN but adding licensing, hardware in some cases, and a third party to coordinate during incidents.
AWS Interconnect — multicloud collapses those three into a console workflow. The connection is layer 3, not layer 2, so customers route between VPCs and VPC networks without bridging concerns. Encryption is MACsec between the AWS and Google Cloud edge routers, which is wire-speed and offloaded from the application path. Redundancy is quad — physically redundant interconnect facilities and physically redundant routers per facility — so the SLA is a managed product SLA rather than a cobble of vendor SLAs. Continuous monitoring is built in for proactive failure detection. Integration with AWS Transit Gateway and AWS Cloud WAN means a single interconnect can fan out to many VPCs and many Regions on the AWS side without the customer designing a hub-and-spoke from scratch. None of those are individually new ideas; the consolidation into a managed product with a console-only setup and a free tier is.
The migration timeline — from re:Invent preview to today
VPN vs Direct Connect colo vs AWS Interconnect — the like-for-like
The bar chart below normalises five comparison axes between the three cross-cloud options UK SMEs face. Throughput, setup time, monthly cost (anchored to a 250 Mbps sustained workload at a 20-employee SME), encryption, and redundancy. The chart is indicative and assumes London Region/Equinix LD5 as the reference geography; actual numbers vary with carrier choice and bandwidth committed.
UK SME multicloud workload mix — where the savings actually fall
Not every workload moves to interconnect on day one. The donut below is the typical UK SME multicloud workload distribution we see across our customer base, and it shows why the Azure pendency matters: nearly half of an average UK SME’s cloud spend sits behind Microsoft 365 and Azure. The free 500 Mbps tier is unlocking value on the AWS↔GCP slice today; the bigger prize lands when Azure ships.
The interpretation is straightforward: the AWS↔GCP slice (analytics, AI, BigQuery, Vertex, S3 data lake to Vertex training) is exactly the slice that is now on the free tier from May 2026. The Microsoft 365 / Azure slice — identity, productivity, line-of-business, ERP — remains tied to existing connectivity until the Azure side ships. This is why we say “don’t add a second cloud just because it’s now easier”: optimisation for the slice you already have is a stronger 2026 move than expansion for its own sake.
The 30-day UK SME multicloud networking audit
The score grid below is the single most useful thing a UK SME can do this month: the eight checks below decide whether you are eligible for the free tier today, what your migration cost looks like, and whether your Azure plan is real or aspirational. Use it as a board-ready “what changed for us this week” sheet.
The pattern most SMEs hit: yes to questions 1, 2 and 3, no firm answer to 4–8. That cluster — existing AWS↔GCP traffic, on a VPN, under 500 Mbps — is exactly the migration the May free tier is designed for. Saving the VPN appliance overhead and reclaiming engineer attention is realistic this month.
Cost staircase by SME size — what saving looks like
The table below sizes the saving by SME headcount, anchored on the typical cross-cloud workload mix at each size band. Costs are illustrative monthly totals at London Region/Equinix LD5 reference points; actual figures vary by carrier, port speed and Direct Connect commit.
| Business size | Current path | Current monthly £ | AWS Interconnect monthly £ | 12-month saving £ |
|---|---|---|---|---|
| Micro — 1–9 staff, AWS+GCP analytics, sub-200 Mbps | Site-to-site VPN over internet | £200 | £0 (free tier) | £2,400 |
| Small — 10–49 staff, light AI workloads, 250–500 Mbps | VPN + SD-WAN appliance | £700 | £0 (free tier) | £8,400 |
| Medium — 50–249 staff, BigQuery + S3 data lake, 1 Gbps committed | Direct Connect + Partner Interconnect colo | £2,400 | £850 (paid above 500 Mbps) | £18,600 |
| Larger SME — 250+ staff, multi-Region DR, 10 Gbps committed | Dual Direct Connect + dual colo | £6,500 | £3,200 | £39,600 |
| Multi-site SME — 50+ staff, branch + cross-cloud, hybrid AI | SD-WAN + VPN + occasional colo | £1,800 | £420 (free tier + branch last-mile) | £16,560 |
Two cautions on the table. First, the “free tier” is bandwidth-bounded; sustained workloads above 500 Mbps fall back to bandwidth- and Region-priced billing, but at AWS’s own framing “Interconnect traffic is priced well below standard egress.” Specific per-GB rates for the paid tier above the free allowance are not yet published; we expect them to undercut current Direct Connect data-out pricing by a wide margin. Second, the saving against Direct Connect colo is what it is — sit down with your existing colo contracts before assuming you can flip immediately. Some are 12- or 36-month commits; the saving lives in the renewal window, not the same week as the GA.
Old way vs new way — side-by-side
Old way: cross-cloud as a project
Reactive, vendor-coordinated, weeks of lead time
- Procure a colocation port at Equinix LD5 or Telehouse North — weeks to provision
- Negotiate cross-connect with carrier and partner cloud — multiple bilateral contracts
- Configure BGP, route filters, IPsec keys and SD-WAN policy — weekly engineering tax
- Layer MACsec or IPsec encryption manually if regulated workloads require it
- Procure SLA exceptions across each vendor in the chain
- Pay port + cross-connect + bandwidth + SD-WAN licensing every month
- Re-engineer when traffic patterns change — not a config tweak
New way: cross-cloud as a console workflow
Proactive, console-driven, minutes of lead time
- Open the AWS Direct Connect console, choose the partner cloud and Region pair
- Let AWS provision quad-redundant interconnect facilities and routers behind the scenes
- MACsec encryption is wire-speed and on by default between AWS and partner edge routers
- Integrate with Transit Gateway/Cloud WAN so one connection fans out to many VPCs
- Free 500 Mbps tier per Region replaces an entire VPN appliance bill
- Single product SLA covers the whole path — no multi-vendor finger-pointing
- Scale the bandwidth from the console as traffic patterns change
Vendor partner readiness — what is and is not GA today
The progress section below tracks the partner-cloud side of the open spec adoption. AWS shipped both halves of its own contribution at GA. Google Cloud’s side is fully in production. Microsoft Azure has signalled adoption and product collateral has appeared in preview channels but the Azure side is not yet GA. Oracle has signalled, with no firm date. Alibaba Cloud has not signalled. UK SMEs should plan against this readiness curve, not against AWS’s own roadmap.
UK SME multicloud readiness gauge
The gauge below summarises the collective UK SME readiness picture for the new world of console-managed cross-cloud connectivity. It blends three inputs: how much UK SME workload is on AWS+GCP today (where the GA lands cleanly), how much sits behind Azure (where the cut-over still needs designing), and how many SMEs have a documented multicloud network plan at all. The needle position is meant to be honest, not flattering — most SMEs are halfway, not finished.
What about Cisco Meraki SD-WAN — do I rip it out?
UK SMEs running Cisco Meraki MX SD-WAN sometimes ask whether the AWS Interconnect launch makes Meraki redundant. It does not. Meraki MX’s job is the branch and SD-WAN layer: connecting offices, retail sites and mobile workers to your core network with policy, application visibility and zero-touch deployment. AWS Interconnect’s job is the cloud-to-cloud layer: AWS↔GCP and (later) AWS↔Azure private fabric. The two meet at AWS Transit Gateway. The right pattern for most multi-site UK SMEs is Meraki on the branch side, AWS Interconnect on the cross-cloud side, Transit Gateway as the join. If anything, the new console-managed cross-cloud product makes Meraki easier to size correctly — you no longer need to overprovision Meraki tunnels for cross-cloud volume.
At-a-glance summary — what UK SMEs need to know
| Question | Answer (May 2026) |
|---|---|
| Is AWS Interconnect — multicloud GA? | Yes, late April 2026, 5 Regions including London & Frankfurt. |
| Is the free tier really free? | Yes — one local 500 Mbps interconnect per Region, started May 2026. |
| Is Azure GA? | No. Committed for “later in 2026”, no firm date. |
| Is Oracle Cloud GA? | No. Committed, no firm date. |
| Does it replace ExpressRoute? | Not yet. ExpressRoute is the Azure-side path; wait for the Azure interconnect ship. |
| Does it replace Direct Connect? | No. Direct Connect handles AWS↔on-prem; Interconnect handles AWS↔other-cloud. |
| Does it replace SD-WAN? | No. SD-WAN is branch/edge; Interconnect is cloud-to-cloud. |
| Is encryption included? | Yes, MACsec wire-speed between AWS and partner cloud edge routers. |
| Is the SLA managed end-to-end? | Yes — single product SLA covering quad-redundant facilities and routers. |
| Does it work in London? | Yes, GA Region. Frankfurt also. |
| What about UK data sovereignty? | London Region paired with europe-west2 keeps in-UK data on UK fabric. |
| How long does setup take? | Minutes from the AWS Direct Connect console. |
| What if my throughput is over 500 Mbps? | Bandwidth-priced above the free tier, “well below standard egress”. |
| Is it open standard? | Yes — spec on github.com/aws/AWSInterconnect, any cloud can adopt. |
| Should I act this month? | If you run AWS+GCP under 500 Mbps on a VPN, yes. Otherwise plan for Azure ship. |
How this lands against the rest of the 2026 cloud story
This GA does not arrive in a vacuum. It lands inside a 2026 cloud landscape where the CMA’s UK cloud market verdict has already named multicloud friction as a competitive concern, where Azure outage waves have UK businesses re-evaluating reliance on a single cloud, and where AI workloads are forcing UK SMEs into multi-vendor footprints they did not choose. We covered the regulatory backdrop in our piece on the CMA cloud market verdict and what UK multicloud businesses do next. The reliability angle is in our analysis of the Microsoft 365 and Azure outage wave earlier this year. The AI-driven multicloud demand pattern was visible in our coverage of Microsoft 365 Copilot agent mode. And the longer-running sovereignty thread is in our piece on UK reliance on US tech giants. Read together with today’s GA, the picture is consistent: 2026 is the year multicloud became a console feature for UK SMEs.
Get a UK SME-grade multicloud network plan in 30 days
Cloudswitched runs the audit, designs the AWS↔GCP cut-over, sequences the Azure plan against the “later in 2026” window, and integrates Meraki SD-WAN at the branch. London-based, UK-SME-focused, capability-led.
See our Cloud Networking solutionFrequently asked questions
Talk to the Cloudswitched cloud-networking team
Book a 30-minute scoping call. We will run the 8-question audit live, model your free-tier saving against your current VPN or colo bill, and put a 12-month plan against the Azure interconnect ship date. London-based, UK-SME-focused, no obligation.
Plan our cross-cloud move
CloudSwitched
London-based managed IT services provider offering support, cloud solutions and cybersecurity for SMEs.
