DC Suite · Operate

Turn your datacenter into a GPU cloud.

DC Suite is the operating system for your GPU datacenter — the same platform that runs Podstack, licensable to run inside your own facility. Orchestrate, fractionalize, meter, bill and govern multi-tenant GPU workloads across NVIDIA and AMD.

See the architecture
ISO 27001Multi-tenantNVIDIA + AMDSelf-hostableSovereign-ready
console.podstack.ai / dc-suite / fleet
Fleet overview live
GPUs online
128
Tenants
37
Metered / hr
₹18,669
H100 node-01
88%
H100 node-02
75%
L40S node-07
85%
MI300X node-11
62%
A100 node-14
88%

Proven in production

The platform running Podstack’s own ISO 27001 GPU cloud — now licensable for yours.

The Gap

You have the GPUs. Turning them into a cloud is the hard part.

Buying or leasing accelerators is the easy 10%. Making them a self-serve, multi-tenant, billable, governed product is years of platform engineering — the part operators underestimate.

Raw GPUs aren't a product

No multi-tenancy, no self-serve provisioning, no quotas or isolation. A bare fleet can't be sold by the slice or the minute.

Utilization bleeds money

Whole-GPU allocation strands 40-70% of capacity. Without fractionalization, inference tenants each hold a full card they barely use.

Billing & governance are DIY

Metering, invoicing, cost analytics, RBAC, audit trails, residency controls - building this stack in-house takes a team and 18+ months.

One Control Plane

Your hardware. Your tenants. A cloud business — out of the box.

DC Suite sits on top of any NVIDIA or AMD fleet and turns it into a governed, monetized, multi-tenant GPU cloud. Five modules, one platform, your brand.

Scheduler
K8s-native orchestration
PodVirt
Fractional GPUs
BillOps
Metering & invoicing
FinOps
Cost analytics
Governance
Sovereignty & RBAC

How It Fits Together

From bare metal to billable cloud

DC Suite is the layer between your GPUs and your customers. It federates the fleet below and presents a governed, self-serve cloud above.

⚉ Your tenants & customers

Self-serve portal
REST API & CLI
SLAs & quotas
White-label billing
DC Suite control planeruns on Podstack · self-hostable
Scheduler
K8s-native orchestration
PodVirt
Fractional GPUs
BillOps
Metering & invoicing
FinOps
Cost analytics
Governance
Sovereignty & RBAC

▣ Your GPU fleet — any hardware, any datacenter

NVIDIA H100H200A100L40SAMD MI300XRTX 4090Bare-metal & colo

Path to Production

Live in weeks, not the 18 months it takes to build it

01

Deploy on your GPUs

Install DC Suite on your fleet - Kubernetes-native, container/OCI, on-prem or colo. No rip-and-replace.

02

Fractionalize with PodVirt

Slice each GPU from 12.5% to 100% across NVIDIA and AMD. Safe oversubscription, no vendor SDK lock-in.

03

Meter & bill with BillOps

Per-minute usage metering, prepaid wallets, invoicing and white-label billing - your customers, your brand.

04

Optimize with FinOps

Per-tenant and per-GPU cost analytics, utilization dashboards and chargeback to keep every card earning.

Module · PodVirt

Sell a GPU by the slice

PodVirt is Podstack’s software-defined fractionalization layer. It carves any accelerator into right-sized slices so ten inference tenants can share one card safely — and it runs on NVIDIA and AMD without a vendor SDK.

  • Fractions from 12.5% to 100% of any GPU
  • OEM-agnostic - NVIDIA + AMD, one control plane
  • Safe oversubscription for SaaS-like gross margins
  • Per-tenant isolation enforced in the control plane
1× NVIDIA H100 · 80GButilization 88%
Tenant A · 42%
B · 28%
C · 18%
free
Training job Inference API Notebook Idle headroom
Metered revenue · 7d▲ 23% vs prior
₹4.9L
Mon
Tue
Wed
Thu
Fri
Sat
Sun

Module · FinOps & BillOps

Every card accounted for, every minute billed

BillOps meters usage per minute and turns it into invoices; FinOps turns the same data into margin. See revenue, cost and utilization per tenant, per GPU, per datacenter — and forecast where the next dollar comes from.

  • Per-minute metering with prepaid wallets
  • White-label invoicing & multi-currency
  • Cost & margin analytics with chargeback
  • Utilization forecasting per tenant and GPU

Run on Podstack — or run it yourself

Podstack is the only GPU platform available both as a managed cloud and as a licensable operating system you deploy in your own datacenter.

Managed Cloud

Run on Podstack

Use QuickPods, TrainPods and Inference on Podstack’s own ISO 27001 cloud — nothing to operate.

  • Fractional & serverless GPUs, per-minute billing
  • Data residency options - US, EU, and India - zero egress
  • Live in minutes, no infrastructure to manage

Licensable Platform · DC Suite

Run it yourself

License the full proprietary stack — scheduler, PodVirt, BillOps, FinOps — and run it inside your own facility.

  • Data never leaves your walls - true sovereignty
  • Your hardware, your brand, your customers
  • Same platform proven running Podstack in production
12.5%
smallest GPU slice
100%
of a card, on demand
2×
NVIDIA + AMD
99.9%
uptime SLA
1min
metering granularity

Why DC Suite

Built for cloud economics — not just scheduling

Generic orchestrators schedule pods. DC Suite runs a GPU-cloud business: fractioning, metering, billing, cost analytics and sovereignty, in one licensable platform.

CapabilityBuild in-houseGeneric orchestratorDC Suite
Fractional GPUs (12.5-100%)Months of R&DPartial✓ PodVirt
NVIDIA and AMD, no SDK lock-in
Per-minute metering & invoicingDIY billing stack✓ BillOps
Cost & margin analyticsBolt-on tools✓ FinOps
Multi-tenant isolation & RBACPartialPartial
Sovereignty & residency controls
Self-hostable in your datacenterVaries
Time to a billable cloud18+ monthsStill no billingWeeks

Sovereign by Design

Compliance that’s enforced in software, not a PDF

The reason regulated buyers and sovereign operators pick Podstack. DC Suite enforces residency, isolation and access at the control-plane layer — and self-hosting means data never leaves your walls.

In-region data residency

Pin every workload to a datacenter and region. Data and control both stay inside the boundary you choose.

ISO 27001 & audit trails

Certified security processes, full audit logging of who accessed what and when, and named-personnel access controls.

Per-tenant isolation

Each tenant is walled off from every other - compute, storage and network - with quotas and RBAC enforced by the platform.

Self-hostable

Run the entire stack in your own datacenter. For the most sensitive workloads, nothing has to touch anyone else's cloud.

OEM-agnostic

No dependence on a single silicon vendor's SDK - de-risk supply and mix NVIDIA and AMD in one fleet.

Your brand, end to end

White-label the tenant portal, API and billing. Your customers see your cloud - DC Suite stays invisible.

Full certification detail lives in the Trust Center →

Built For

Anyone who owns GPUs and wants a cloud

Datacenter & colo operators

Monetize idle GPU capacity as a self-serve, billable cloud.

Sovereign & regional clouds

Stand up a compliant national GPU cloud on your own soil.

Regulated enterprises

BFSI and healthcare running private, governed GPU platforms.

Telcos & GPU resellers

Turn edge and core capacity into an AI-compute product line.

Research & HPC centers

Fair-share fractional GPUs across labs with usage chargeback.

Neoclouds scaling up

Skip 18 months of platform build and go to market now.

Frequently asked questions

DC Suite

License the platform that runs Podstack.

DC Suite is enterprise-licensed and deployed with our team. Tell us about your fleet and we’ll map the fastest path to your own GPU cloud.