In early April, on a stage in the southwestern outskirts of Moscow, a moderator at Russia’s annual Data Fusion conference wanted to know: what is the most important thing for Russia to get right in its quest to develop an AI ecosystem?
The six men on the stage before her represented Russia’s second-largest bank, the state nuclear power company, and the Ministry of Digital Development. Instead, she started with the only person joining via video link.
“Katerina Vladimirovna,” she said, referring to the pale face, whose credential at the conference was managing director of a small research and development foundation, by her patronymic. “Your answer, please.”
“Talent is everything,” replied Vladimir Putin’s younger daughter, whose full name is Katerina Vladimirovna Tikhonova, knowingly or not echoing a 1935 address by Joseph Stalin. “Everything else is a consequence of talent.” The panelists were quick to agree. And yet, there are reasons to doubt that the talent that Russia is capable of developing is sufficient to overcome Russia’s structural weaknesses in AI.
In recent months, Russian authorities and institutions have made a concerted push to develop homegrown AI talent. Vladimir Putin has established a Presidential Commission on AI and changed national curricula to emphasize the technology. Moscow State University, the nation’s most prestigious university, has established a new AI faculty, alongside an AI institute headed by Putin’s daughter. These moves seek to address the brain drain of top technical talent following the invasion of Ukraine by playing to a traditional Russian strength—upskilling members of a population of some 140 million people, which has historically seen success in the mathematical sciences. However, these moves do little to address Russia’s greatest weakness in AI: scarce access to indispensable hardware, due to limited domestic production capacity and stringent sanctions.
‘Talent is everything’
In April, Russia’s main TV news channel depicted Moscow State University gilded in futuristic laser lines as the presenter announced a new AI faculty, due to welcome its first cohort of 72 students in September. The exclusive course, which is financially supported by oligarch Oleg Deripaska, a Putin associate, spares no expense. More than half of the places are sponsored, waiving the $7,000 course fee, and the faculty is granted access to one of the nation’s most powerful supercomputers, unveiled in 2023. The faculty completes a “unified ecosystem” comprising an AI institute, headed by Putin’s daughter herself, which opened in 2020; a research center established in 2025; and now an educational body to train the next generation of experts.
Tikhonova’s post at the heart of the ecosystem is likely nepotistic, says Katheryna Bondar, a senior fellow at the Center for Strategic and International Studies. Tikhonova, who in a past life was an international rock ’n’ roll dancer, has a Ph.D. in mathematics, but has not published research in AI. However, the growth of the AI industry around her institute could be a sign of the growing salience of AI in the Russian president’s orbit.
The new AI faculty is part of a broader effort. In March, AI was added to the national informatics olympiad, held annually since 1989, as the country aims to increase its output of AI specialists from roughly 3,000 in 2022 to 15,500 by 2030. “What really surprises me is how comprehensive their thinking and approach is,” says Bondar.
Brain drain
Talent is widely considered to be one of the key inputs to AI progress, but it has been an issue for Russia thus far in the AI boom. After Russia invaded Ukraine in early 2022, roughly a quarter of Russian software developers’ GitHub profiles changed, ceasing to share their location or showing that they had left the country. “I didn’t want to be part of this,” says Dima Dobrynin, who led an autonomous driving project at Yandex, Russia’s answer to Google, before leaving in the weeks following the invasion. Many of his friends with technical backgrounds departed, too.
The departure of some of Russia’s brightest computer scientists has come at an inopportune moment for the country. The war in Ukraine has shown the importance of modern data and AI infrastructure on the battlefield. “Russians were always complaining that Ukrainians get this super sophisticated software,” says Bondar. In February, Putin established a Presidential Commission on AI to help establish state policy on AI. In case there was any doubt about where the technology being developed by private Russian companies is headed, the Commission features the Minister of Defense and Director of the FSB alongside representatives from some of the nation’s most tech-savvy companies. The distinction between civil and military technological development in Russia is blurred, says Bondar.
Sovereignty and isolation
At the heart of many of Russia’s AI efforts is a desire for sovereignty—lately a buzzword in international AI discourse. The platonic ideal is to have AI models developed by domestic researchers, on domestic hardware, capturing domestic values, thereby ensuring the technology’s control by the government and independence of external interference. “For Russia, this is a question of state, technological, and, one could say, value sovereignty,” said Vladimir Putin at a November AI conference.
The desire for control is evident in Putin’s close associates at the top, and the restrictions in access to resources. “No one has access to MSU-270 [the Moscow State University supercomputer] except for a narrow circle,” a source told T-invariant, an émigré Russian media outlet. The recently announced AI faculty promises to produce world-class AI specialists, by giving them access to this coveted hardware—but this exceptionalism is also a challenge. “You cannot build world-class anything in isolation,” says Dobrynin.
Russia’s determination to maintain a tight grip on AI development has hampered its own efforts to advance the technology. A March draft bill mandated that Russian AI models respect “Russian spiritual and moral values.” Some industry leaders pushed back, pointing out that Russia lacks the data and computing power to train such models, and that the bill’s demand that AI models prioritize “the spiritual over the material” is not a robust legal definition around which companies can build their technology.
The chips problem
However, Russia’s talk of growing its own AI talent pipeline masks a greater weakness in the nation’s sovereign AI efforts. The country’s ability to manufacture the specialized computing hardware needed to train and run AI models lags China and the U.S. by decades, says Samuel Bendett, an advisor at the Center for Naval Analyse
So far, Russia has relied on a stock of gray-market American chips—official exports have stopped since the war began—which it reportedly used to develop the supercomputer used by the new AI faculty. However, a recent crackdown on chip smuggling and proposed U.S. legislation to track the locations of American chips, to prevent them ending up in undesirable territories, may throttle that supply.
Historically, Russia has been stronger in software than hardware development, according to Bendett. This lack of domestic hardware expertise, along with the complexity of global semiconductor supply chains, complicates Russia’s attempt to indigenize AI: the new initiatives at Moscow State University emphasize software development in AI, but don’t train electronic engineers that could cultivate the hardware to run their algorithms.
Russia hopes that China may prove to be a durable supplier. Russia’s battlefield data, accumulated over the course of the war in Ukraine, is a valuable asset for China, which has not fought a ground war in decades, as it seeks to train its own AI models with military applications, says Bendett.
But there are reasons to be skeptical of this approach. “China barely produces sufficient AI chips for their own demand,” Lennart Heim, an AI and semiconductor policy expert, told TIME. Russia likely falls lower on China’s priority list of customers than countries where Beijing is vying for influence with the U.S.
Amid the enthusiasm about producing homegrown talent on the Data Fusion stage, AI chips—and their scarcity in the Russian AI industry—were barely an afterthought.
English (US) · 