Power What's Next with
Ananta Cloud GPU as a Service

High-performance, GPU computing on secure STPI infrastructure, with a clear path to greater scale through Shakti Cloud, India's sovereign AI cloud.

What is Ananta Cloud GPU as a Service?

Ananta Cloud GPU as a Service delivers dedicated and virtualized NVIDIA GPU compute, provisioned in minutes, for AI and machine learning training and inference, deep learning, data analytics, engineering simulation, and visualization workloads. Choose GPU-accelerated virtual machines for flexible, pay-as-you-scale workloads, or dedicated bare-metal servers for maximum, uncontended performance, all hosted within STPI’s secure, redundant data centers. 
For projects that need a larger GPU footprint, the latest NVIDIA architectures such as H100 and L40S, or specialized AI platform services, Ananta Cloud customers can seamlessly extend  their workloads to Shakti Cloud, without leaving the trusted STPI ecosystem.

Why Choose Ananta Cloud GPU as a Service?

High Performance

Native GPU plans are built on NVIDIA Ampere A40 and A100 accelerators, paired with high-core-count Intel Xeon processors, ample RAM, and storage to deliver the throughput required by demanding AI, analytics, and simulation workloads.

Scalability

Start with a single-GPU virtual machine and scale to multi-GPU VMs or bare-metal servers as your workload grows. Beyond native capacity, Shakti Cloud extends this with large GPU fleets and elastic, serverless consumption.

Security & Compliance

Every GPU instance runs inside STPI's secure data centers, giving enterprises, startups, and research teams a trusted, compliant environment to develop and deploy AI workloads on Indian soil.

Predictable, Transparent Pricing

Straightforward monthly rate card pricing for every native plan, so teams can budget GPU costs with confidence, with the option to switch to consumption-based pricing on Shakti Cloud for burst or short-term workloads.

Built for GPU-Intensive Workloads

  • Generative AI, NLP & Computer Vision

    Train, fine-tune, and run inference on language and vision models, accelerating deep learning pipelines and video analytics.

  • Big Data & Predictive Analytics

    Process and visualize large datasets, run real-time analytics, and power fraud- and anomaly-detection pipelines.

  • Healthcare & Life Sciences

    Accelerate medical imaging, genomics, and drug-discovery workloads with GPU-powered processing.

  • Engineering & Scientific Computing

    Run high-fidelity simulations for structural analysis, computational fluid dynamics, and research modeling.

  • Financial Modeling & Risk Analysis

    Accelerate quantitative modeling, risk simulations, and algorithmic research.

Platform Features

GPU Features

Parameter Ampere A100 80 GB Ampere A40
Best suited for suitable application or usage AI training / inference, data science, HPC, gaming AI training, deep learning, data science, HPC
Single precision performance (FP32) 19.5 TFLOPS 37.42 TFLOPS
Half precision performance (FP16) 78 TFLOPS 37.42 TFLOPS (1:1)
GPU memory 80 GB HBM2 48 GB
Form factor PCIe PCIe
GPU memory bandwidth 1,940 GB/sec 696 GB/sec
Power consumption 300 W max TDP 300 W
Interconnect interface PCIe Gen 4: 64 GB/s PCIe Gen 4: 64 GB/s
Thermal solution Passive Passive
GPU base clock 1,065 MHz 1,305 MHz

Native GPU Plans on Ananta Cloud

Plan Configuration MRC (₹)
VM_A40p – Single GPU 20 vCPU
96 GB RAM
1 × 48 GB NVIDIA® Ampere® A40p GPU
250 GB disk
vNIC, up to 10G network, unlimited data transfer		 ₹37,800
VM_A40p – Dual GPU 40 vCPU
192 GB RAM
2 × 48 GB NVIDIA® Ampere® A40p GPU
250 GB disk
vNIC, up to 10G network, unlimited data transfer		 ₹69,300
BM_A40 – Bare Metal PowerEdge R70XA
2 × Intel Xeon Gold 6342 (2.8 GHz, 24C/48T)
1,024 GB RAM
2 × 480 GB SSD
4 × NVIDIA Ampere A40 (192 GB total), passive, double-wide, full-height GPUs		 ₹2,00,340
BM_A100 – Bare Metal 2 × Intel Xeon Platinum 8452Y (2.0 GHz, 36-core, 300 W)
16 × 64 GB RAM (1,024 GB total)
2 × 960 GB + 2 × 1.92 TB NVMe Gen4 SSD
10/25 Gb 2-port Ethernet adapter
1 × NVIDIA A100 80 GB GPU		 ₹1,28,520

  1. Bare Metal GPU Servers - dedicated, uncontended H100 and L40S capacity for peak, large-scale performance.

  2. Shakti Clusters - Kubernetes and SLURM clusters for large-scale distributed training and inference.

  3. Serverless GPUs & AI Endpoints - pay-per-execution GPU capacity and serverless deployment for LLM, vision, and speech models with auto-scaling.

  4. Fine-Tuning & Model Services -managed services to customize and deploy foundation models on sovereign infrastructure.
    Visit shakticloud.ai to explore full specifications, pricing, and to get started.

Need More Scale? Explore Shakti Cloud

For workloads that outgrow native capacity - or that call for the newest NVIDIA GPU generations - Ananta Cloud customers can extend onto Shakti Cloud, India's sovereign AI cloud and one of NVIDIA's global Cloud Partners. Shakti Cloud is built and hosted entirely in India, giving enterprises full data residency alongside NVIDIA H100 and L40S GPU capacity at scale.
On Shakti Cloud, teams can additionally access:
AI Workspace VMs - GPU virtual machines on NVIDIA H100 SXM and L40S, with NVLink-enabled multi-GPU configurations for distributed training.

Get In Touch

Need help choosing the right GPU plan, or want to discuss a workload that needs Shakti Cloud–scale capacity? Our team is here to help you find the right fit across the Ananta Cloud and Shakti Cloud ecosystem.

Contact Us Today