"Neu.ro delivered a complete ML development environment in MegaFon Cloud in 2 months."

Alex Osipov
megafon logo
Alexandr Osipov
Head of Cloud Platforms and Infrastructure Solutions
MegaFon is the second largest mobile operator and third largest telecommunications company in Russia, providing mobile, landline, voice, data and cloud services, consumer entertainment services and enterprise business solutions. MegaFon has >75 million subscribers, >35 million data users and >$5 billion of annual revenue.

AI/ML Cloud
Custom OEM AI Cloud
AI requires specialized infrastructure, including AI- accelerated hardware and robust MLOps.

MLOps is mission critical, but it is not core IP.

Cloud providers should consider total time and cost before attempting to build and maintain MLOPs internally.


Global Cloud Market 1

$258 billion
Cloud market (2020)
$104 billion
Software as a Service (SaaS)

Russian Cloud Market 2

$1 billion
Cloud Market (2019)
Y-o-Y Growth (2018-2019)

Hyperscale cloud providers continue to grow both top line revenue and market share by providing value added software and data services to their growing list of corporate clients.

“Demand for integrated IaaS and PaaS offerings is driving the next wave of cloud infrastructure adoption.”

– Sid Nag, Director of Research at Gartner

But demand for regional cloud providers with geographically distributed network assets and long-term client relationships remains strong. In fact, there are hundreds of telecommunications companies providing data services to corporate clients around the globe, and their combined revenue streams far outweigh the giants . Just 54 telecoms on the Forbes Global 2000 list accounted for over $3.4 trillion in assets and nearly $1.5 trillion in revenue last year. Unlike AWS and GCP, telecoms have kept their clients data secure over decades.

Recently, Apple’s release of the iPhone 12 has marked the beginning of the 5G era. Telcos own 5G.

According to Gartner’s 2020 CIO Survey, enterprises expect to double the number of AI projects in place within the next year.Technology market research firm Omdia forecast that by 2025, AI will account for as much as 50% of total public cloud services revenue.4

“AI is a huge priority. We are seeing transformation happening, with AI going from the cloud to being distributed, such as on the edge or IoT devices.”

– John Smee, VP of Engineering and Head of 5G R&D for Qualcomm


1 Gartner: Forecast: Public Cloud Services, Worldwide, 2018-2024, 2Q20 Update
2 IDC: Cloud Service Market 2020-2024 Forecast and 2019 Analysis
3 Gartner CIO Survey 2020
4 Omdia (formerly Tractica) and Deloitte

AI requires specialized infrastructure, including AI accelerated hardware and robust MLOps.

MLOps is mission critical, but it is not core IP.

Cloud providers should consider total time and cost before attempting to build and maintain MLOPs internally.

THE Business Challenge

In 2017, Megafon entered the cloud business through a $740 million acquisition. In 2020, MegaFon recognized the opportunity of AI/ML + 5G as major drivers of future cloud revenue.

To capture this revenue stream, MegaFon needed to provide its clients with specialized AI/ML infrastructure – both hardware (in the form of GPUs) and software (in the form of a ML development platform).

MegaFon faced a build or buy decision. After review of the state of the art, the company estimated that building an MVP AI Cloud platform would be a multi-year task, and getting it right could be much more. Critically, they recognized that MLOps is not core IP – it is mission critical functionality that is not strategic – they should turn to the best provider available.


For hardware, MegaFon turned to NVIDIA, the world’s leading provider of advanced computing technologies for AI/ML.

For AI/ML infrastructure, Megafon partnered with Neu.ro. Within 60 days Neu.ro deployed a custom OEM MLOps platform. Branded as the MegaFon AI Cloud, it now provides clients with all infrastructure necessary to execute successful AI Transformation strategies. In detail, the MegaFon AI Cloud is a Kubernetes-based orchestration platform with three main functions:

Resource Orchestration: leveraging all compute and storage in MegaFon cloud infrastructure, the platform orchestrates these resources for clients, and the MegaFon internal team, to run AI/ML workloads on industry-leading NVIDIA GPUs simply and transparently.

Process Orchestration: the platform integrates the best in open source and proprietary ML tools, covering the entire ML project lifecycle: from data collection to deployment and monitoring. The platform’s dedicated pipeline engine automates routine processes, focusing ML engineers’ time on launching solutions.

Access & Role Orchestration: the platform empowers organizations to benefit from collaborative development, while securing sensitive assets within company perimeters.

Complete, supported AI/ML Transformation environment Clients building AI on high-value GPUs Launch within 60 days

The result

By choosing a partnership with NVIDIA and Neu.ro, MegaFon achieved the mission-critical goal of launching AI Transformation for their B2B client base, and for their internal term, in an extremely short time frame.

They jump-started their clients’ AI/ML teams with ready-to-run ML recipes for a host of business cases in computer vision, natural language processing and predictive analytics.

They solved technical support with Neu.ro’s team of MLOps engineers. At the same time, they avoided substantial time and cost risks by outsourcing all non-strategic functions.

In doing so, MegaFon launched the infrastructure necessary to capture the high-value and growing GPU revenue stream that AI/ML represents.

Furthermore, MegaFon established a competitive advantage by providing their clients with the entiure suite of AI/ML services and support they need to de-risk their AI journeys.