News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise Expansion
EveryNeocloud
Christofari

Christofari

Sberbank's Nvidia-based supercomputers in Moscow for neural network training

Christofari is a pair of supercomputers, Christofari (2019) and Christofari Neo (2021), built by Sberbank with Nvidia hardware and run with its SberCloud subsidiary. They train neural networks and run inference, science and commercial work, and are offered to outside users through the SberCloud ML Space cloud platform.

Share on WhatsApp

  • Launched December 2019; Christofari Neo followed in November 2021
  • About 6.7 petaflops for Christofari, 11.95 petaflops for Christofari Neo
  • First financial-sector supercomputer open to outside users
  • Built on Nvidia DGX-2 with Tesla V100, then DGX A100 with 80 GB A100 GPUs
  • Housed in a Sberbank data centre at Skolkovo, Moscow

What it is

Christofari is the name of two supercomputers run by Sberbank of Russia together with its SberCloud subsidiary, built on Nvidia hardware. The first machine went live in 2019 and the stronger Christofari Neo followed in 2021. Their main job is teaching neural networks, and they also handle inference of trained models, scientific research and commercial calculations. Outside organisations in business, science and government can reach the machines through the SberCloud ML Space cloud platform. The systems sit in the Top 500 list of the strongest commercially available computers. The name honours Nikolai Christofari, the holder of Sberbank's first savings account passbook and its first customer.

How it came about

Sberbank and SberCloud showed the first Christofari on 8 November 2019 at the AI Journey conference in Moscow, presented by CEO Herman Gref and group CTO David Rafalovsky. Commercial service began that December. Within a year its capacity underpinned the ML Space machine learning platform, announced in December 2020. Christofari Neo was unveiled at the same conference in November 2021 by Rafalovsky. In March 2020 Christofari was described as the only Russian supercomputer built specifically for AI work, and as the fastest in the country, beating the previous leader Lomonosov-2 by more than twice.

The compute

Christofari uses Nvidia DGX-2 nodes, each with 16 Tesla V100 accelerators, 512 GB of GPU memory, dual Intel Xeon Platinum 8168 CPUs and 1.5 TB of system memory. Mellanox switches link the nodes over InfiniBand EDR at 100 Gbit/s. Christofari Neo runs DGX A100 units with eight A100 GPUs of 80 GB each and dual AMD Rome 7742 processors, with InfiniBand giving up to 1600 GB/s per node. Software is a modified Ubuntu 18.04 LTS with Nvidia drivers, CUDA Toolkit, cuDNN, NCCL and containerised workloads.

Where the GPUs are

Both machines sit in a Sberbank data centre at the Skolkovo Innovation Center in Moscow. Christofari fills a single machine room and was assembled in under a year. The systems are certified in Russia for handling personal data.

Customers and deals

Sber uses the machines internally for speech recognition, voice generation in its call centre, where bots already answer 40% of customer enquiries, and for reading CT scans of lungs. SberDevices and Sber AI were first to use Christofari Neo, building a Russian-language image generation service based on DALL-E. Scientific, commercial and government bodies can also rent capacity through SberCloud ML Space.

Its place in AI

Christofari was the first supercomputer of a financial institution anywhere to be opened to third-party users, and the first Russian system certified for personal data work. At launch it ranked 40th in the world, 7th in Europe and 1st in Russia. Christofari placed 1st among the 50 strongest computers in the CIS in September 2021 and 80th globally in June 2022. Christofari Neo reached 7th in the HPL-AI rating in November 2021 and 46th worldwide in June 2022.

Questions people ask about Christofari

What is Christofari?

It is a pair of supercomputers owned by Sberbank of Russia and built with Nvidia hardware, run alongside the SberCloud subsidiary. Christofari started service in 2019 and Christofari Neo in 2021. They train neural networks and run inference, research and commercial jobs.

Who can use it?

Sber uses it for its own work, such as call centre speech and lung CT analysis. Outside scientific, commercial and government organisations can access the capacity through the SberCloud ML Space cloud platform. SberDevices and Sber AI were the first teams on Christofari Neo.

What hardware does it use?

Christofari runs Nvidia DGX-2 nodes with 16 Tesla V100 accelerators each, linked by Mellanox InfiniBand EDR at 100 Gbit/s. Christofari Neo uses DGX A100 units with eight 80 GB A100 GPUs and InfiniBand reaching 1600 GB/s per node.

Where is it located?

Both systems sit in a Sberbank data centre at the Skolkovo Innovation Center in Moscow. Christofari occupies one machine room and was built in less than a year. It is certified in Russia for work involving personal data.

How powerful is it?

Christofari delivers about 6.7 petaflops of effective performance and Christofari Neo 11.95 petaflops. At launch Christofari ranked 40th in the world and first in Russia. Christofari Neo placed 7th in the HPL-AI rating in November 2021.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "Christofari", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryNeocloud rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.

It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.