Oblivus offers a cloud platform that provides on-demand and reserved access to a range of NVIDIA GPUs and modern CPUs with per‑minute billing and no quota limits. Users can launch, modify, and scale instances instantly via a fast API or dashboard, benefiting from pre‑installed drivers, deep‑learning libraries, and high‑speed networking at prices up to 80% lower than major providers.
Funding
Funding not disclosed
Founders
Product
Problem
Access to high-performance GPU and CPU compute resources is often limited by high costs, complex provisioning processes, and restrictive quota systems, making it difficult for developers, researchers, and enterprises to scale AI, rendering, or HPC workloads efficiently.
Solution
Oblivus provides a cloud platform that offers on-demand and reserved access to a wide range of NVIDIA GPUs and modern CPUs with per‑minute billing and no quota restrictions. Users can launch, modify, and rescale instances via a fast API or web dashboard, with deployment times as short as 30 seconds. The service includes pre‑installed drivers and a custom Ubuntu‑based OS (OblivusAI) with deep‑learning libraries, as well as dedicated, fully isolated hardware resources. Pricing is transparent and up to 80 % lower than major cloud providers, with optional reservations that further reduce costs. Tier‑3/4 data centers ensure ISO‑certified security and high availability, while network connectivity ranges from 10 Gbps to 100 Gbps.
Target Audience
Primary customers are AI/ML engineers, data scientists, VFX studios, cloud gaming developers, and academic or corporate research groups that require scalable GPU‑accelerated compute.
Features
- Access to eight enterprise‑grade NVIDIA GPU SKUs (e.g., H100, A100, RTX 4090) and latest CPUs, selectable per instance
- Per‑minute, pay‑as‑you‑go billing with optional discounted reservations; no hidden fees or setup costs
- Quota‑free provisioning; instances can be launched, modified, and scaled via a lightning‑fast REST API or dashboard
- Dedicated hardware resources with full GPU pass‑through and root access; no resource sharing between customers
- Pre‑installed NVIDIA drivers and OblivusAI OS (Ubuntu 22.04) bundled with common deep‑learning frameworks for immediate model training
- High‑speed networking (10 Gbps–100 Gbps) and replicated 3‑x enterprise storage included in instance pricing
- ISO/IEC 27001, 27017, 27018, 27701 certified Tier‑3/4 data centers delivering 99.99 % uptime