Russia Is Banning Western AI. The Biggest Winner Is China.

DeepSeek already has 43% of Russia's AI market. Russia's new bill to ban Western AI would hand it the rest.

Russia's Ministry for Digital Development published proposals on March 20 to ban or restrict foreign AI tools — ChatGPT, Claude, and Gemini — if they don't comply with new rules governing what Moscow calls "traditional Russian spiritual and moral values." The rules would take effect from 2027. English-language coverage has treated this as a digital sovereignty story. It's also a China market-access story — and nobody's covering that part.

Here's what's buried in the technical detail: Chinese models like Qwen and DeepSeek are explicitly exempt.

Technology lawyer Kirill Dyakov, explaining the proposed legislation to RIA Novosti, drew the line clearly. Western AI tools fail because they transmit "user data, queries, and dialogues" outside Russia. Chinese open-source models — DeepSeek, Qwen — can be deployed on Russian government infrastructure without that data leaving the country. They pass. ChatGPT doesn't.

This isn't a carve-out for Russian sovereignty. It's a market structure that eliminates Western competitors and leaves Chinese alternatives standing.

DeepSeek's Russian foothold isn't hypothetical. A Microsoft report from January 2026 found DeepSeek had already captured 43% of Russia's AI market — ahead of its own domestic models, ahead of anything Sberbank or Yandex has built. Russia's state lender GigaChat serves around 15,000 companies. Yandex has a reasoning model. But in terms of user adoption, a Chinese tool is already the dominant foreign AI in Russia, and the new law would cement that position by clearing out its only competition.

The "sovereignty" framing is accurate as far as it goes. Russia's FZ-152 data localisation law already requires personal data on Russian citizens to be stored in Russia. The new AI proposals extend that logic to model inference: if your queries leave Russian servers, you're out. This is a genuine legal framework, not invented. The problem is that "data sovereignty" and "Chinese AI dominance" aren't mutually exclusive — in this case, they're the same outcome.

This follows a pattern Russia has run before. In February, Moscow blocked WhatsApp for its 100 million Russian users and redirected them to state-backed Max. YouTube, Facebook, and at least 13 other domains were removed from Russia's national DNS registry. Each ban cleared Western platforms. Each replacement — domestic or Chinese — filled the gap.

Digital ecosystems, once fractured, don't re-integrate. The Runet — Russia's sovereign internet infrastructure — has been under construction since 2019. The AI bill is its next layer. Russia's citizens will have access to GigaChat, YandexGPT, and whatever DeepSeek version runs on government servers. They won't have access to the tools the rest of the world is using to write, research, and build.

The bill is a draft. It will be finalised. It will probably pass. And when it does, the story won't be that Russia banned ChatGPT. It'll be that a country of 145 million people became the largest live demonstration that Chinese AI is a viable Western substitute — with government mandate, market scale, and data infrastructure to prove it.

That's not a sovereignty win. That's a dependency the Russian government is building for itself, one blocked domain at a time.