MWS Cloud, part of MTS Web Services (MWS), will hold its second MWS Cloud Day 2026 conference in Moscow on November 12, 2026 — in person at the Palace of Culture and via online broadcast, with free participation upon registration and confirmation. The program includes three tracks (clouds, AI, security) and 20 talks, with the key presentation featuring announcements on infrastructure, networks, containers, AI, data platform, and security prepared by the company's CTO, CPO, and CISO. The theme of the year is declared as "How to create products in the age of AI" — the discussion will cover moving AI pilots to production and the requirements for computing, data, networks, observability, and cost.

image

What happened

MWS Cloud announced that the second MWS Cloud Day 2026 conference will take place on November 12, 2026, at the Palace of Culture, 15s1 Sharikopodshipnikovskaya Street. The event will be held offline and simultaneously available via online broadcast; participation is free, but offline applications require confirmation due to limited seating. The program consists of three tracks — clouds, AI, and security — and includes 20 talks. The main presentation will be prepared by three top managers: CTO Danila Dyugurov, CPO Daniil Pivovarov, and CISO Viktor Bobylkov, and will include announcements on infrastructure, networks, containers, AI, data platform, and security. The AI track specifically features multi-tenant inference, an internal AI platform with MCP and evals, and an MTS Link case study on migration to MWS GPT Model Hub.

Context

For MWS Cloud, this is already the second conference under the Cloud Day brand, and the fact of repeating the event is as telling as the composition of its program. Russian cloud announcements have passed the phase of general talks about models: the agenda has settled on the transition from AI experiments to industrial operation, and the program features specific engineering mechanisms, not abstract visions. Among them are an alternative to the "common queue" when serving requests to models, internal AI platforms with context and skills, MCP and evals support, migration of the corporate product MTS Link to a model hub, control of unconscious consumption of cloud resources, and automated security configuration checks. The overall framework is the theme of the year, "How to create products in the age of AI": what is required for production operation of AI in terms of computing, data, networks, observability, and cost.

Why this matters for the industry

The main signal for the industry is not the event itself, but the composition of the announcements: around multi-tenant inference, a model hub, and an internal platform with MCP and evals, a platform layer for AI products is emerging in the Russian cloud. The most technically significant topic is tenant isolation on a shared computing pool: it is precisely this that determines GPU utilization, latency, and inference predictability, i.e., the key variables of any LLM service. If the announcements after the conference lead to documentation, pricing, and public API, Russian teams will have a local managed inference option to compare with their own serving stacks, and competitors will have to show comparable solutions for tenant isolation and model evaluation. The MTS Link migration case to MWS GPT Model Hub is a potential source of real production data on inference cost and quality, if speakers provide metrics, not just a description of the process. MCP support aligns with the industry trend toward standardization of context and skills: the value will need to be assessed by the quality of implementation, not just the mention of the protocol.

Why this matters for users

For engineers and team leads, this is a free and almost zero-effort way on November 12 to get a map of real production patterns: how colleagues move AI to production, where they get stuck when migrating to Model Hub, and how they calculate inference costs. You can participate in person in Moscow or via online broadcast; offline participation requires an application with confirmation due to a seating limit, so it is better to register in advance. The organizers promise to publish recordings of the talks on the landing page after the event, so those who miss the broadcast will still have access. Before the conference, it makes sense to prepare specific technical questions for the talks on multi-tenant inference and Model Hub — for example, about pricing, latency, and API availability, to get verifiable details from the presentations, not just marketing.

What is still unknown / limitations

Until November 12, everything announced remains vendor marketing statements: the sources contain no methodology, benchmarks, metrics, or datasets confirming the announcements. Inference pricing, latency and throughput metrics, SLA, and API availability of future features are unknown. It is unclear what is meant by evals: an internal harness or open benchmarks, on which datasets, and whether the tooling will be available to customers. MCP and the platform layer in the program are a stated intention, not an existing product. The value of the MTS Link case depends on whether the speakers will provide latency, cost, and quality metrics; checks will be available after November 12 via talk recordings and official MWS Cloud materials.

Sources

Author

Look at AI, editorial team