Operating from the US and Great Britain The entire team in Oblivus is Turkish and they provide services all over the world.

themselves”Democratized GPU cloudDefining it as “, Oblivus’ goal is to provide a reliable, fast, scalable and economical infrastructure that not only large organizations but also everyone can easily access.

Oblivus, which has two multi-tenant data centers in Montreal, Canada and Oslo, Norway, offers its users the On-Demand service in these data centers that anyone can register and access immediately. In addition, there are 7 single-tenant data centers and these data centers are only used for a few customers with large clusters (+512 GPUs).

Building an infrastructure from scratch is a financial challenge, and this is where Oblivus comes into play:

Setting up its own infrastructure for a startup interested in artificial intelligence in the current conditions Financially it is very difficult. That’s why startups often turn to CSPs. However, CSPs also have some problems:

1. Prices: While it is more economical for startups to build their own infrastructure, even the most basic virtual machines with graphics cards can reach tens of thousands of dollars per month in large CSPs. Unless you have a large investment, this is not an amount that can be spent on just anyone.
2. Quota and limits: Each CSP has certain quota limits. You cannot create a virtual machine with a graphics card without increasing these limits. To increase the limits, you need to contact different departments within CSP, submit a request, send emails and explain in detail why you need your quota. In today’s ‘AI Boom’ era, many medium and small organizations are unable to meet these quotas. If you are not affiliated with a company, your quota request will be rejected immediately. This limits access to graphics cards and lengthens the software development process.
3. State-of-the-art GPUs: Most CSPs offer GPUs with the latest technology (NVIDIA H100, etc.), which are currently available on the market and accessible only through long-term reservations. Unless you want to spend a few million dollars, these GPUs are very difficult to access and scale your usage as needed.

Oblivus provides highly critical and beneficial solutions to these problems:

1. While they offer the same level of service as large CSPs, they also offer prices that are up to 50% more affordable. This price difference allows small and medium-sized businesses to implement artificial intelligence projects without making large investments.
2. Since there is no quota or limit on the platform, any registered user can use hundreds of graphics cards at the same time, as long as there are sufficient funds in their account and there are enough GPUs in the infrastructure. This allows users to configure their infrastructure as desired and smoothly scale this infrastructure if necessary. As of today, Oblivus has 9 data centers and more than 18,000 GPUs.
3. Oblivus offers the latest technology NVIDIA H100 and all other GPU models directly to all its customers with its On-Demand service, without the need for long-term reservations.

With its infrastructure and over 29,000 registered users, Oblivus also supports many sectors other than artificial intelligence. Its users include major conglomerates such as VNG Corporation, AI-focused Akash, Bittensor and Strong Compute, architectural firm Edge Design Dubai, leading universities Stanford University, UCLA, UC San Diego and the University of Notre Dame. These organizations are provided with clusters consisting of hundreds of graphics cards.

If you would like detailed information about Oblivus’ services and much more, you can visit their website by clicking here.