Kremlin Adapting Western Chips for Military AI

Kremlin Adapting Western Chips for Military AI

Executive Summary:

• In November 2025, Russian President Vladimir Putin announced a new “Artificial Intelligence (AI) headquarters” to hasten the development of Russian-made AI, which lags years behind Western advances despite the Kremlin’s triumphant rhetoric.
• Kremlin claims of “AI sovereignty” rely on gray-market access to Western hardware—Russian-made devices, such as the Delta Sprut XL computer, use smuggled AI chips, though adaptation may occur within Russia.
• This lag in ground-up development has not stopped the Kremlin from fielding AI-capable military technology. Russia is mastering the industrial-scale application of lethal tactical automation, adapting smuggled components to create and scale a crude, ethically unconstrained, and lethal AI-powered “kill chain.”

Russian President Vladimir Putin provided new guidance on Russia’s direction in artificial intelligence (AI) at the November 2025 “Journey to the World of AI” conference in Moscow (President of Russia, November 19, 2025). The event highlighted Russia’s militaristic pivot in technology policy. The state-owned banking giant Sberbank—whose technocratic-minded chairman, German Gref, has positioned himself as Russia’s corporate AI champion—hosted the conference. Putin said that “critical dependence on foreign AI systems is unacceptable,” framing artificial intelligence as a matter of “state, technological, and value-based sovereignty” (President of Russia, November 19, 2025). To enforce this, he ordered the creation of a central headquarters to direct the AI industry.

In December 2025, Russian Prime Minister Pavel Mishustin announced that the government, in coordination with the Presidential Administration, developed a proposal to establish Putin’s new AI headquarters. The headquarters would direct, enforce, and monitor the implementation of generative AI across all industries and regions (Digital Russia, December 9, 2025). It remains unclear what will happen to the existing “Headquarters for Control over the Implementation of Instructions,” which the government established in 2023 and is affiliated with the Higher School of Economics (HSE) (The Government of Russia, March 13, 2024).

The assessment of Russia’s position in generative AI on the HSE headquarters’ website is stark. Russia lags three to five years behind the United States and the People’s Republic of China (PRC) in generative AI development (AIRU HSE, December 4, 2025). This admission from the state’s own technocrats undermines the Kremlin’s triumphalist rhetoric, including Putin’s 2017 statement that “whoever leads in AI will rule the world” (RT, September 1, 2017). The Kremlin has since struggled to convert its massive investments into leadership in AI. The reality of Russia’s lagging position in the global AI race is well captured by the 2024 Global AI Index by Tortoise Media, which ranks Russia 31st out of 83 countries in AI implementation, innovation, and investment. The United States and the PRC, which the Kremlin considers its competitors, hold the first and second ranks (Tortoise Media, September 19, 2024).

Kremlin Touts Sovereign Assembly of Imported AI Components

The central tension in Putin’s AI strategy is the mismatch between his demand for “sovereign” AI, meaning AI built and maintained independent from other countries, and the reality of Russia’s infrastructure and capacity. Some experts believe that this gap is insurmountable in the high-tech sector. While global leaders in AI operate on 3-nanometer AI chips, Russia’s own industrial roadmap hopes to master 28-nanometer production—technology from 2011—by 2030 (The Moscow Times, January 28, 2025).

The Kremlin relies on smuggled hardware to bridge its three-to-five-year lag behind the United States and PRC in AI development. In December 2025, Russian manufacturer Delta Computers unveiled the “Delta Sprut XL,” marketed as the country’s “most powerful AI platform” (Delta Computers, December 2, 2025). The manufacturer’s datasheet reveals the system is powered by Intel Xeon 6 processors and supports up to 20 NVIDIA H200 accelerators—components strictly banned from export to Russia (Delta Computers, accessed January 14). Delta Computers founder, Andrey Chernyshev, acknowledged their dependence on imported technology, stating, “unfortunately, for now, we are on foreign chips” (Natsionalnii Bankovskii Zhurnal, May 29, 2025). Chernyshev framed the company’s value not in its hardware, but in the engineering required to assemble it, saying, “often architectural changes can show very good results even with average [technology]” (Natsionalnii Bankovskii Zhurnal, May 29, 2025). Chernyshev claims the company’s innovation lies in its unique server design, which packs a higher density of accelerators onto a single board than standard Western configurations. This “sovereign architecture” allows the Kremlin to claim independence even as it relies entirely on gray-market supply chains to import its AI chips.

Moscow Pursues Sovereign AI for Ideological Control

During a December Presidential Council for Civil Society and Human Rights meeting, a consultative body for Putin, Russian entrepreneur Igor Ashmanov warned that Russian officials’ reliance on superior Western large language models (LLMs) to perform their daily tasks will inevitably leak state secrets to geopolitical adversaries (Digital Russia, December 10, 2025). Ashmanov, a prominent figure in the Russian information technology (IT) sector who evolved from a successful technocrat into one of the Kremlin’s primary ideologues for “digital sovereignty” and state control over the internet, argued that this trend erodes the security of the government and education system. Putin acknowledged the risks but framed himself as a modernizer in response, insisting that while AI adoption is irreversible, it requires strict “human” (state) control, signaling his intent to develop sovereign AI to prevent his regime’s authority from being undermined by foreign algorithms. Putin sees domestically produced and managed AI as the only way to avoid Western technological hegemony and meddling.

In December 2025, ideologue Konstantin Malofeev crystallized this vision, calling for a “Sovereign Russian AI” trained on Russian Orthodox texts, the Domostroi (a 16th-century set of Russian household rules), and the works of ultranationalist Russian philosopher Alexander Dugin. Malofeev demanded the creation of an “Internet Special Service”—akin to a digital State Security Committee (KGB)—to enforce Russia’s “cognitive sovereignty” (Tsargrad, December 10, 2025). The Kremlin wants a “sovereign” AI that rejects Western liberalism and enforces values that align with Putin’s vision of Russian civilization. This anti-Western software, however, would most effectively run on Intel and NVIDIA chips smuggled from the West, which Malofeev claims to reject.

Kremlin Integrating AI Into Military

In late December 2025, the Kremlin finalized the 2027–2036 State Armaments Program, which envisions the broad integration of AI into strategic nuclear forces, space capabilities, air defense, communications, electronic warfare (EW), command systems, unmanned robotics, and some weapons systems (TASS, December 26, 2025). The Kremlin asserts that dependence on foreign AI in its military is unacceptable yet unavoidable in the short term. Moscow is decentralizing its military innovation, attempting to modernize its military-industrial complex (The Institute for Northern Eurasia Transformation, November 2025). The Kremlin aims to bridge the widening technological gap with the West through the ruthless operationalization of smuggled hardware, rather than through domestic invention.

Russian Defense Minister Andrei Belousov is managing the modernization of the Russian military. Former Chief of General Staff Yuri Baluyevsky and defense analyst Ruslan Pukhov laid out the need to integrate AI into the Russian military in a January article titled “Digital War—New Reality” (Russia in Global Affairs, January 1). Baluyevsky and Pukhov argue that the modern battlefield has become a “total extermination zone” due to the transparency created by ubiquitous surveillance. They conclude that the tank is now more of a target than a weapon, and the future belongs to “cheap, expendable” robotic systems. Belousov has operationalized this doctrine by establishing the “Center Rubicon” drone unit. Unlike the clumsy bureaucracy of the regular military forces, Rubicon operates like a startup, integrating commercial technology directly into the kill chain (Radio Free Europe/Radio Liberty, September 17, 2025). Rubicon’s operational success lies in shifting targeting priorities from destroying vehicles to “hunting the hunter”—locating and killing drone operators. Using electronic intelligence (ELINT) and AI-enabled data fusion to triangulate control signals, Russian artillery strikes the human nodes of the Ukrainian drone network.

The focus of Russian military AI integration is on “machine vision” to improve targeting and neutralize EW jamming. The Russian V2U reconnaissance drone exemplifies this, using an NVIDIA Jetson Orin module and mixed foreign components from companies such as Intel and Sony to navigate terrain without GPS and coordinate swarm actions (Militarnyi, June 9, 2025). According to a defense industry source, Russian engineers have successfully adapted standard radio-controlled drones to bypass EW. The source claimed that all new models operating in the combat zone now feature “AI-powered automatic target locking,” enabling the devices to “guide themselves to the target without operator participation,” making them immune to EW countermeasures (TASS, August 28, 2025). Even legacy air defense systems such as the Tor-M2U have been updated with AI algorithms to track and intercept complex targets, such as HIMARS rockets (Sovershenno Sekretno, December 5, 2024).

These systems are defined by their use of adaptive, commercially available technology rather than Russian-made breakthroughs. Despite offering a significant tactical edge by bypassing electronic warfare defenses, their hardware relies heavily on imported components—specifically commercial parts from the PRC, such as laser rangefinders, and Western dual-use processors from companies such as NVIDIA and Xilinx. The prevailing Russian military technology doctrine is one of modification: retrofitting inexpensive airframes with commercial AI visual capabilities, a sharp contrast to the ground-up development of high-end components and architectures seen in the United States, some PRC programs, and some Ukrainian startups.

Russia claims to be embedding AI into its platforms to automate complex decision-making, a trend that is alarmingly extending to nuclear capabilities. The Su-57 fifth-generation fighter features the IUS-57 information-control system, described as a “virtual co-pilot” that manages flight systems and weaponry, allowing the human pilot to focus solely on tactics (Telegram/@RussianArms, May 24, 2025). In 2025, the S–70 Okhotnik heavy drone, operating as a “loyal wingman” to the Su-57 fighter, was also equipped with an AI system (Russia in Global Politics, November 1, 2025)

The 2027–2036 State Armaments Program explicitly prioritizes the integration of AI into “strategic nuclear forces” and “unmanned robotic systems,” including autonomous strategic delivery vehicles such as the Poseidon nuclear-armed torpedo (Russia in Global Politics, November 1, 2025). This policy raises significant concerns regarding experiments with AI in nuclear command and control.

Russia is keen to integrate AI into the military to lessen the scramble for talent to replace the technology specialists who fled the country following the full-scale invasion of Ukraine in 2022. In late November 2025, the Russian Ministry of Defense held the first “All-Army AI Festival” at the Tver Suvorov Military School. The event showcased cadets’ projects, including “Zastava,” an AI system designed to parse Telegram channels for “terrorist threats.” Moscow’s AI push is not just propaganda; it is a long-term survival strategy. The Kremlin is attempting to build a loyal pipeline of military-technical experts from adolescence, militarizing the next generation of engineers to support its aggressive foreign policy and suppress dissent at home (Rambler/Zvezda TV, December 1, 2025)

Conclusion

Moscow is creating an AI “kill chain” that is crude, ethically unconstrained, and lethal in spite of the constraints on Russia’s technology sector. Putin’s speech at “Journey to the World of AI” was an admission that the Russian market has failed to deliver the AI sovereignty that the Kremlin demands. The state is now taking the wheel via the new “AI Headquarters.” Russia is losing the race for AI, limited by sanctions and a reliance on smuggled hardware that it cannot scale. This lag, however, has not prevented Moscow from fielding AI-enabled systems that are militarily relevant. Through entities such as Center Rubicon, Russia is mastering the industrial-scale application of tactical automation, adapting smuggled components under Belousov’s technocratic management.