XcellHost|Edge Cloud Platform
AI inference and apps that run close to your users, in India.
GPU and CPU virtual machines, storage and managed Kubernetes on a distributed cloud. Requests are answered near the user or device, data stays in India, and you pay by the hour. The core runs in our own Mumbai Tier IV data centre.
GPU-optimised networking
India data residency
Pay as you scale, hourly
✦Ask AI about edge cloudChatGPTClaudePerplexityGoogle AICopilot
Edge topologyMap illustrative
Edge topology · core ↔ edge ↔ devices
LIVE
01Edge topology
Illustrative
Mumbai coreLive · Tier IV
Edge zones*On request
BillingHourly · INR
ResidencyIndia
1999Running business IT since
Tier IVOwn data centre, Mumbai
24×7NOC & SOC, in-house
MeitYEmpanelled cloud provider
ISO27001 · 20000-1 · 27017 · 27018
Per zone*Latency target · SLA per agreement*
Products
One cloud, five building blocks. All billed by the hour.
Start with a single VM. Add GPUs, storage, a private network, a Kubernetes namespace or a managed inference endpoint when you need them. Everything appears on one bill and one dashboard.
Storage & networking
Data and private links- Object storageS3-compatible buckets for models, video and backups. S3 API →
- VPCPrivate networks, subnets and security groups per project. Networks →
- VPN & private connectSite-to-site tunnels from your offices and plants. Connect →
- Load balancerSpread traffic across VMs and zones. LB →
Automation & AI
Run models and workflows- Managed n8nWorkflow automation on your own VM, patched and backed up by us.
- AI inference serviceCall open models by the token, no servers to run. Serverless →
- Model servingvLLM-style endpoints on a dedicated GPU VM. Dedicated →
- Model catalogueOpen-weight models ready to deploy. Catalogue →
Kubernetes
Two ways to run containers- EssentialsNamespace-as-a-service on a shared, managed cluster. Deploy with kubectl or Helm.
- ProA dedicated control plane for your team, with your own nodes and policies.
- API & toolingKubernetes API, Terraform and CI pipelines. K8s API →
- OpenShift optionRed Hat OpenShift, run for you. OpenShift →
Platform & security
Control, cost and supportLocations & edge zones
A Mumbai core, with edge zones near your users.
Mumbai is our own Tier IV data centre and is live today. Edge zones in other Indian cities run with vetted data-centre partners and are set up on request. Pick a city to see what it does.
Core · livePartner edge zone*Devices & users
Illustrative · not to scale
* Partner edge zones run in third-party data centres chosen and contracted by XcellHost. City, availability, services and any latency target are confirmed in writing for your project before go-live. Rings show "nearer is faster" only; they are not measured figures.
Key capabilities
What sits under every VM. Built for real-time work.
Edge workloads fail in small ways: a jittery link, a full disk, a node that does not come back. These are the parts we design and run so that does not happen to you.
High availability
Spread a service across availability zones so one failure does not stop it.
- Multi-AZ placement in the Mumbai core
- Failover between zones and VMs
- Self-healing Kubernetes control plane
AI-optimised fabric
GPU nodes on fast networks with storage that keeps up with the model.
- GPU-optimised networking between nodes
- NVMe volumes for weights and caches
- Low-jitter paths for voice and video
Operational efficiency
One place to run VMs, clusters and storage across zones.
- Unified control plane and API
- Metrics, logs and alerts built in
- Autoscaling for VMs and pods
Security & sovereignty
Your data stays in India, isolated from other customers.
- India data residency by default
- Encrypted data plane and volumes
- DPDP-aligned controls · DPDP platform
Security & compliance
Zero trust by default. Audited every year.
Nothing is open until you open it. Every user, key and service gets only the access it needs, and every change is logged.
- Zero-trust access. MFA for the console, short-lived API keys and no shared root logins.
- Isolated VPC per project. Private subnets, security groups and no public IP unless you add one.
- IAM-scoped access. Roles per project and environment, so a developer cannot touch production by mistake.
- Encryption. TLS in transit; encrypted volumes, snapshots and object storage at rest.
- Audit trail. Who did what, when and from where — exportable for your auditors.
- 24×7 SOC. Our security operations centre watches the platform. Security operations →
ISO 27001Information security
ISO 20000-1IT service management
ISO 27017Cloud security
ISO 27018Personal data in cloud
ISO 9001Quality management
SOC 2Type I & Type II
MeitYEmpanelled
DPDP ActAligned controls →
CNAPPPosture & workload →
Use cases
Where a few milliseconds change the result.
Four patterns we see most. Each one gets faster, cheaper or easier to keep compliant when the compute sits close to the people or machines it serves.
Chat and voice assistants that answer without a pause.
A voice bot that waits for a far-away region sounds broken. Serve the model on a GPU VM near your callers, keep the conversation logs in India, and scale out at peak hours only.
- Voice agents for contact centres and IVR
- Customer and employee chat copilots
- RAG over your own documents · RAG as a Service
Typical stack
GPU VM · L4 or L40SvLLM-style endpointObject storageK8s Essentials
AI as a Service →Analyse camera and sensor feeds where they are recorded.
Sending every CCTV stream to a distant region is slow and costly. Run detection next to the cameras, keep the footage local, and send only events and clips to the core.
- CCTV and VMS analytics for sites, retail and campuses
- Industrial IoT and machine data · IoT Cloud
- Edge inference for plants · Smart manufacturing · Digital twin
Typical stack
GPU VM · L4NVMe volumesSite-to-site VPNObject storage
Edge network →Host your SaaS and websites close to Indian customers.
Run web, API and database tiers on CPU VMs with a load balancer in front. Add a Kubernetes namespace when you outgrow single servers, and pay only for the hours you run.
- Web apps, APIs and storefronts
- Microservices on managed Kubernetes
- Dev and test environments that switch off at night
Typical stack
G1.large CPU VMsLoad balancerVPCK8s Pro
General Compute →Keep a copy of your systems ready to run.
Replicate servers and data from your office or another cloud into the Mumbai core. Test a failover whenever you like and pay for the recovery VMs only while they run.
- Snapshots and object-storage backups
- Disaster recovery for SMBs · Cloud DR
- Ransomware-resilient backup · Cloud backup
Typical stack
Object storageSnapshotsStandby VMsVPN
Cloud DR →Who it is for
Teams that need speed, control and an Indian bill.
AI startupsServe models without buying GPUs. Start small with credits.
EnterprisesData in India, audit trails and one GST invoice.
Video analytics teamsProcess feeds near the cameras, store less.
DevelopersAPI, Terraform and VMs in about a minute.
Platform teamsManaged Kubernetes with your own policies.
SaaS buildersHost close to Indian users and pay by the hour.
How it works
From sign-in to running workload in five steps.
Sign in
Create an account with your business details and GSTIN. MFA is on from the start.
Choose zone, plan, name
Pick Mumbai or a partner edge zone*, a size and a name such as web-01.
Launch or deploy
Start a VM, or deploy containers to a managed Kubernetes namespace.
Track cost
See hourly spend for every VM, volume and cluster on one dashboard.
Migrate or review
Optional: we move existing servers for you, or review your current cloud bill.
Services
We move it for you. Then we check the bill.
Migration service
Lift-and-shift VMs from AWS, Azure, Google Cloud, VMware or your own servers. We plan the cut-over, copy the data, test it with you and switch over in an agreed window.
- Inventory and sizing of current servers
- Data copy and app testing before cut-over
- Rollback plan and hypercare after go-live
Free cloud cost review
Send us your last three cloud invoices. Our FinOps team maps each line to an Edge Cloud equivalent, flags idle resources and shows what you would pay here, at no charge.
- Idle VMs, oversized instances and orphan volumes
- Egress and data-transfer charges explained
- A written comparison you can share internally
Programmes
For startups building on us, and partners selling with us.
Startup programme
For early-stage Indian companiesCloud creditsUp to ₹49,999*
First monthFree* on CPU VMs
- For DPIIT-recognised or incubator-backed startups*
- An architecture session with our engineers
- A path from credits to normal hourly billing, no lock-in
Indicative — confirm. Credit amounts, eligibility and the free month are set per application.
Partner programme
Resell, refer or build on Edge CloudReferralResellerMSPSystem integratorDistributorTechnology partner
Tiers
RegisteredSilverGoldPlatinum
Margins, commissions and tier benefits are set per partner agreement.
Pricing
Hourly or monthly. The same rate either way.
Hourly price = monthly price ÷ 730 hours. Run a VM all month and you pay the monthly figure; switch it off and you stop paying for compute. CPU VM rates match our General Compute page.
| Size | vCPU | RAM | Hourly | Monthly | Add |
|---|
Indicative · excl. 18% GST · Mumbai core. Boot volume (₹6 per GB a month), public IPv4 (₹299 a month) and Windows Server licence (₹1,499 a month) are extra, as on General Compute.
NVIDIA L424 GB · inference, video AI
NVIDIA L40S48 GB · inference, fine-tuning
NVIDIA RTX PRO 600096 GB · generative media, 3D
NVIDIA H200141 GB · 70B-class models
GPU VMs are always priced on quote — supply and rates change week to week. Quotes are in INR with GST and usually arrive within one business day. GPU memory per NVIDIA; full list on GPU Servers.
Essentials
₹2,999*/ namespace / month
Indicative · excl. 18% GST
- Namespace on a shared, managed cluster
- CPU and memory quota you set
- kubectl, Helm and CI access
- Cluster upgrades and patching by us
For platform teams
Pro
₹9,999*/ cluster / month
Indicative · excl. 18% GST · plus worker VMs at CPU or GPU rates
- Dedicated control plane, highly available
- Your own worker nodes, CPU or GPU
- Your network and admission policies
- Self-healing nodes and autoscaling
| Item | Unit | Price | Source |
|---|---|---|---|
| NVMe block volume | per GB a month, billed hourly | ₹6 | As on General Compute |
| Snapshots | per GB a month | ₹2 | As on General Compute |
| Object storage (S3-compatible) | per TB a month | ₹1,999* | Indicative |
| Public IPv4 | per IP a month | ₹299 | As on General Compute |
| Internet egress | first 1 TB a month per account | Included* | Indicative |
| Egress beyond 1 TB | per GB | ₹0.99* | Indicative |
| Managed n8n | per instance a month, plus the VM | ₹1,999* | Indicative |
| AI inference service | per token or per endpoint | On quote | Serverless inference |
Indicative · excl. 18% GST. * Placeholder rates to be confirmed by XcellHost. Traffic inside a VPC and between your VMs in the same zone is not charged.
Hourly calculator
Price your workload by the hour.
Choose a size, how many hours it runs, how many copies and how much storage. The estimate updates as you go and adds to your cart in one click.
Your estimate · G1.xlarge
₹0
incl. 18% GST · for the hours you run
Indicative · excl. 18% GST before tax line. Hourly rate = monthly ÷ 730. Storage is prorated for the same hours. Object storage at ₹1,999* per TB a month.
Console preview
Everything on one screen. Spend by the hour.
A drawn preview of the Edge Cloud console. Names and figures are samples — click a row to see its details.
Running4
This hour₹13.97
Month to date₹0
Alerts0
| Name | Type | Zone | Status | ₹ / hour |
|---|---|---|---|---|
| web-01 | G1.xlarge | Mumbai | Running | 6.57 |
| api-02 | G1.large | Mumbai | Running | 3.29 |
| infer-01 | GPU VM · L4 | Mumbai | Running | Quote |
| shop-prod | K8s Essentials | Mumbai | Healthy | 4.11 |
FAQ
Edge Cloud, answered.
It is a distributed cloud for workloads that need to respond quickly: AI inference, video analytics, real-time copilots and web apps. You get CPU and GPU virtual machines, storage, private networking and managed Kubernetes, billed by the hour in INR. The core runs in XcellHost's own Mumbai Tier IV data centre, with partner edge zones in other Indian cities on request.
AI startups serving models, enterprises that need data to stay in India, video analytics teams, developers, platform teams running Kubernetes and SaaS companies hosting for Indian users.
Linux or Windows VMs, containers on Kubernetes, model-serving endpoints on GPU VMs, n8n workflows, databases, web servers and backup targets. If it runs on a standard x86 server or an NVIDIA GPU, it runs here.
CPU VMs start at ₹1,199 a month for 1 vCPU and 4 GB RAM, or ₹1.64 an hour — the same rates as our General Compute service. NVMe storage is ₹6 per GB a month. GPU VMs are priced on quote. All prices are indicative and exclude 18% GST. Use the hourly calculator for an estimate.
Yes. Our migration service lifts and shifts VMs from AWS, Azure, Google Cloud, VMware and on-premises servers. We size the targets, copy and test the data, and cut over in a window you agree, with a rollback plan.
Data stays in India by default, access is scoped by IAM, data is encrypted and every action is logged. These controls support your obligations under the Digital Personal Data Protection Act, 2023; compliance itself also depends on your own processes. Our DPDP Compliance Platform covers consent, notices and data-principal rights.
Three things. Compute can sit closer to your users through edge zones, you are billed in INR with a GST invoice and a published egress allowance, and you talk to an Indian team that runs the platform 24×7. Many customers use both: a hyperscaler for global services and Edge Cloud for India-facing, latency-sensitive workloads.
Essentials gives you a namespace with quotas on a shared cluster we manage — the fastest way to deploy containers. Pro gives your team a dedicated, highly available control plane with your own worker nodes, network policies and admission rules. Essentials is ₹2,999* per namespace a month; Pro is ₹9,999* per cluster plus workers (indicative).
Send three recent cloud invoices. Our FinOps team maps each line to an Edge Cloud equivalent, points out idle and oversized resources and egress charges, and sends a written comparison. There is no charge and no obligation.
Mumbai is XcellHost's own Tier IV data centre and is live. Edge zones in cities such as Delhi NCR, Pune, Bengaluru, Hyderabad, Chennai and Kolkata run with partner data centres and are set up on request. Availability is confirmed for your project before go-live.
Uptime SLAs and any latency targets are written into your customer agreement, per zone and per service. We do not publish one number for all cases because latency depends on where your users are and how they connect.*
Our NOC and SOC run 24×7 from Mumbai. Call +91 22 6711 1555, WhatsApp +91 86570 32540, or email [email protected]. Sales questions go to [email protected].
Tell us where your users are. We'll place the compute.
Share your workload, cities and expected traffic. We'll reply with a zone plan, VM sizes and an INR estimate — usually within one business day.*
ISO 27001 · ISO 20000-1SOC 2 Type I & IIMeitY empanelled24×7 NOC & SOCMumbai Tier IVSince 1999