Nwan
We never let you fail.
One fabric, one control plane and one bill, from a branch in Kolkata to an AI cluster in Hyderabad to the clouds you run on. Fibre partnerships, optical layer, IP and MPLS fabric, managed edge, control software, consoles and billing: HyperNext runs every layer, so every layer answers to one promise.

Every day, a fibre is cut.
Somewhere in India, a road is dug and a cable snaps. For most networks that is where the story stops. A payment fails. A call drops. A hospital waits. Networks fail; that is physics. Letting you fail is a choice.
What goes wrong today
- Fibre cuts, slow repair
- Long-haul routes are cut often and take hours to repair. On most networks, traffic waits for the crew
- Hidden shared risk
- Two routes sold as diverse often share one duct or one bridge. One cut takes down both
- Cloud access in weeks
- Private links into cloud regions take weeks to order and run congested at peak
- Control that sits abroad
- Many controllers and logs live outside India, at odds with the 2026 network data localisation rules
- AI traffic on a shared WAN
- Datasets and checkpoints moving between AI campuses crowd out everyone else
- Bills you cannot check
- SLA and billing figures come from measurements the customer never sees
Eight services. One fabric.
Ordered from the tenant console or the API, live in minutes rather than weeks, and billed from the same telemetry you can see.

Nwan tenant console
Services
- Nwan Port
- 1G to 400G ports at any PoP or HyperNext campus
- Nwan Connect
- Point-to-point circuits, 50 Mbps to 100 Gbps, changed in minutes
- Nwan Cloud Connect
- Private circuits into cloud regions in India, two PoPs by default
- Nwan VPN
- Any-to-any private network across sites and clouds
- Nwan Edge
- Managed edge for branch, data centre and cloud
- Nwan Wave
- 100G, 400G and 800G wavelengths between campuses and metros
- Nwan Internet
- Clean transit, with DDoS cleaned by HyperNext Scrub
- Nwan Secure
- Firewall, intrusion prevention and segmentation, run by our own SOC
Nine levels. One supervisor.
Every failure is answered at the fastest layer that can answer it. Only if that layer cannot restore the service does the next one act. One supervisor watches all nine, so nothing waits for a human.

Operator console · fallback events
The ladder
- F0 · under 50 ms
- Light switches to a protection fibre
- F1 · under 50 ms
- Routers repair the path locally
- F2 · under 1 s
- The fabric re-optimises every path
- F3 · 60 to 180 s
- Wavelengths re-route on other fibre
- F4 · always on
- Alternate carrier carries the load
- F5 · under 1 s
- Edge moves flows to the best link
- F6 · under 1 s
- Cloud traffic shifts to a second on-ramp
- F7 · no impact
- Network runs on if controllers are cut off
- F8 · immediate
- Committed traffic keeps priority
Every target is proven in acceptance testing on the live fabric before launch.
Every connection has a twin.
Each committed connection runs on two paths that share no duct, no bridge and no building. We do not assume diversity. We prove it.

Path trace · primary and backup on separate routes

The fabric on the map of India
Run from India. Watched by our own SOC.
Sovereign, secure, transparent
- Sovereign
- Controllers, telemetry, logs and keys stay in India. Built for the 2026 network data localisation rules. No support access from outside India
- Secure
- Every link between PoPs encrypted, every edge on IPsec. DDoS cleaned on campus by HyperNext Scrub. Incidents reported to CERT-In within six hours
- Transparent
- You see the same telemetry we bill on. A 95th percentile you can check yourself. SLA credits applied automatically
- Cloud in minutes
- Private circuits into the major cloud regions in India, two on-ramps in two PoPs by default plus an encrypted standby path, ordered from the console or API
- Branches that stay up
- Every link measured ten times a second for loss, delay and jitter. Flows move to the best link in under a second. Zero-touch install
- Who it is for
- Hyperscalers, banks and financial services, HyperNext colocation tenants, enterprises and GCCs

Create a connection · live in minutes
Monthly availability, by class.
Four classes
Platinum
Dual edge, dual PoP
- Availability
- 99.995%
- Access
- Two links at once
- Core
- Protected, separate routes
Gold
Dual PoP
- Availability
- 99.99%
- Access
- Separate routes
- Core
- Protected
Silver
Single access
- Availability
- 99.95%
- Access
- Single
- Core
- Protected
Standard
Single access
- Availability
- 99.9%
- Access
- Single
- Core
- Standard
Timing figures are Nwan 1.0 design targets, each proven in acceptance testing before launch. Services and specifications may change before general availability.
Also offered as a service
Talk to the engineering team.
Specifications, drawings and deployment planning for the 800 VDC product line. Designed by HyperNext Engineering, Toronto. Made in India.
