Postcriptum State of Devops Russia 2025 (Devops column)

I will share the final part separately. reportDima Gaevsky, my friend and colleague who is responsible for the development of our internal development platform.

The Russian software market continues to go its own, sometimes paradoxical way: on the one hand - hard external pressure, on the other - the irreversible maturation of the IT landscape. In the XL segment, the trend towards integrating AI into production pipelines has only intensified. Bigtechs have already built secure internal platforms, and now they are tagging AIOps and GenAI copilots, squeezing maximum productivity from DevOps. But it’s important to remember that even the biggest players are still catching up. Compared to the U.S. and China, budget gaps and access to modern GPUs remain an issue for at least the next two years.

Small and medium-sized companies that have experienced turbulence 2022 Years ago, they solved problems autonomously and almost everyone chose a proven OSS stack - GitLab, Ansible, ELK, Kubernetes. This was the only rational path, given the lack of mature Russian proposals and high technological uncertainty. Now that import substitution regulations have tightened (software register, quotas in public procurement, compatibility with Alt / Astra)Domestic add-ons are gradually added to this stack – from SCA plugins with GOST-crypto to GitFlic code repositories.

Security has become a separate front: a massive self-hosting GitLab without built-in patch management processes has preserved many vulnerabilities. Companies are starting to invest in SBOM and DevSecOps to cover regulatory and reputational risk. At the same time, the popularity of FinOps is growing: the cost of GPU clusters is growing faster than the ROI for experimental AI projects, and boards of directors are increasingly asking not “how much we train models”, but “how much we will save”.

Hardware constraints are palpable: top-tier NVIDIA/AMD is still under export control, China’s ASIC is a working solution, but puts a performance ceiling. This pushes XL-companies to “federal” alliances: banks and retail share pre-trained LLM, industrial enterprises share predictive service models; the state acts as an early anchor customer and subsidizes developments, but the volume of subsidies is still incomparable with global CAPEX.

The forecast for the coming years is without panic, but also without illusions. The big ones will continue to upgrade AI innovations and build FinOps offices, insuring TCO. SMB will remain on the OSS kernel, but will have to spend on DevSecOps and outsourcing SOC. Consolidation of efforts will go in two planes: horizontal coalitions of giants for model exchange and vertical “superstructure” of domestic solutions over universal OSS. As a result, the market will receive not a “Western stack against the Russian”, but a hybrid of “OSS-base + local specialized modules”, which is perhaps the most realistic scenario. 2025 – 2027 years.

#Processes #Management #Performance #Engineering #Software #SoftwareDevelopment