From Russia’s biggest economic conference in St. Petersburg, Sberbank’s first deputy CEO Alexander Vedyakhin makes a pitch blending technology sales with geopolitics. Russia, he tells Reuters, offers something Western AI models do not: alignment with local values. GigaChat, Sberbank’s flagship large language model, is slower and less capable than leading American and Chinese systems – Vedyakhin concedes this openly – but arrives without the content filters, privacy implications, or cultural assumptions baked into systems built in San Francisco. As NewsTrackerToday underscores, the argument targets a specific anxiety running through governments across Latin America, Africa, Asia, and Oceania: that adopting Western AI means adopting Western norms at the infrastructure level of national decision-making.
The pitch lands in a specific geopolitical context. President Vladimir Putin stated last week that Russia is one of only three countries globally capable of developing homegrown AI models – positioning it alongside the United States and China as an independent pole in a race most observers treat as two-horse. Sovereign AI is a political signal, telling potential client governments they can participate in the AI transition without subordinating data governance to foreign platforms. Vedyakhin frames the target audience as states that want sovereign AI but cannot afford to build it from scratch – a description fitting many United Nations member states.
Russia’s actual position in the global AI race is complicated by its self-declared ambitions. Sberbank’s GigaChat and Yandex’s YandexGPT both lag frontier systems from Anthropic, OpenAI, Google, and leading Chinese labs. Western sanctions block access to advanced semiconductors, forcing Sberbank CEO German Gref to discuss Chinese-made chip procurement during Putin’s May visit to Beijing. Vedyakhin frames the dependency analytically: Nvidia’s dominance flows not just from hardware but from CUDA, the programming standard all major language models depend on. Owen Radner, a technology geopolitics and AI infrastructure specialist, notes that CUDA dependency makes Western sanctions structurally damaging – not merely inconvenient – for Russian AI development, a constraint NewsTrackerToday pinpoints as central to Moscow’s position in this space.
The technical direction Vedyakhin advocates is notable independent of geopolitics. He argues the industry reaches a saturation point on model parameters: users do not need billions of parameters for general problems, but compact specialized models at reasonable cost. A credit scoring model has no need to recite poetry – building that capability wastes compute. For the Global South pitch this is practical: smaller specialized models are cheaper to run and more deployable on weaker infrastructure. NewsTrackerToday ties this to a wider debate about whether the frontier race favors incumbents with massive compute, or whether compression opens space for new entrants.
Vedyakhin puts AI productivity gains in concrete terms for the Russian economy: 11 to 22 percent improvements in some sectors, with labor redistributing toward construction where physical work cannot be automated. The robotics dimension surfaces in a candid anecdote: tired of watching dancing robots at AI events, he directed his team to build something practical – an AI-powered tiling robot now in testing. He immediately qualifies the result: the robot only works on a perfectly prepared surface, which humans still need to provide – illustrating the gap between AI-assisted automation and full displacement in messy real-world environments.
The strategic logic of the Global South pivot reflects Russia’s positioning in a three-bloc technology world – the American ecosystem, the Chinese ecosystem, and the space smaller nations navigate between them. Sberbank’s offering carries a distinct value proposition: no sanctions risk, no US export controls, no content moderation aligned to Washington’s preferences. Alex Reinhardt, a geopolitical technology risk specialist, observes that for governments with sensitive use cases in defense or economic planning, those attributes matter more than benchmark rankings. As News Tracker Today investigates across its geopolitics coverage, supplying AI infrastructure to the developing world is a defining competitive front of the decade – and Sberbank’s pitch signals Russia intends to contest that ground even from acknowledged technical disadvantage.
