Siberian Banks Pitch “Home‑Made” AI to Developing Nations

Russia’s largest bank is offering its own artificial‑intelligence tools to countries in Latin America, Africa, Asia and Oceania that worry about privacy and cultural fit in Western products. The bank’s chief AI officer said the system is not as fast or large as top U. S. models, but it follows local values and can be built with fewer resources. He explained that many governments want a “sovereign” system they can control, yet lack the money to develop one from scratch. The bank’s flagship model, built in partnership with a domestic software giant, is still catching up to leaders like Anthropic and DeepSeek. According to the executive, the next step is to shrink models so they use less data and power while still solving specific tasks such as credit scoring or language translation. “People need practical tools, not billions of parameters, ” he added.Russia’s president recently highlighted that only a handful of countries can produce their own large‑language models, and the bank is positioning itself to meet that niche. The CEO noted that Western sanctions have forced the company to source chips from China, which power its main model. He also pointed out that a single piece of software—CUDA—has become the de‑facto standard for many AI systems, making it hard for new hardware to compete. The executive estimated that AI could raise productivity by up to twenty‑two percent in certain sectors of the Russian economy. He also joked that instead of fancy robot demos, his team is building a tiling machine to help with construction. The prototype works only on surfaces that have already been prepared by humans, he said. While the model is still early in development, the bank’s approach reflects a broader trend: emerging economies are looking for affordable, culturally relevant AI solutions that do not rely on Western platforms. By focusing on local content and efficient design, the Russian bank hopes to offer a viable alternative that respects privacy concerns.