Machinepower Index / Dossiers / Russia
Rank 25–50 across the scoring rules tested. How this is tested
Russia
Siege Autarky — For nations cut off from allied hardware and markets: build whatever the available silicon allows, accept a slower path, and treat sovereignty as a necessity rather than a choice.
How the Score, Potential and bottleneck cost are worked out: methodology.
The three layers
Energy, grids, data centres and silicon — the physical capacity to run models at scale.
Compute tracked by Epoch AI: Russia controls 1,400 H100-equivalents, wherever the sites are; has 400 on its soil with the owner not reported. Hosted compute is shown, not scored.
Models, alignment capability, standards influence and the talent that produces them.
The capacity to wire AI into the state and the economy while holding public consent.
Where the score comes from
| Layer | Sub-dimension | Basis | Score |
|---|---|---|---|
| Watts | Planning & Permitting | Assessed | 7.6 |
| Firm Power Capacity | Dataset | 7.1 | |
| Sovereign Compute | Dataset | 4.5 | |
| Hardware & Silicon Access | Dataset | 1.5 | |
| Weights | Frontier Model Output | Assessed | 4.3 |
| Alignment Capability | Assessed | 6.0 | |
| Regulatory Standing | Dataset | 7.4 | |
| Talent Density | Assessed | 5.5 | |
| Will | Policy Capacity | Dataset | 6.6 |
| Public Sector Adoption | Dataset | 7.0 | |
| Workforce Transition | Dataset | 8.0 | |
| Public Trust | Assessed | 6.6 |
Assessed marks a measurement we judged rather than took from a dataset: 5 of the 12 here.
Sources
SaluteDevices, Sber's technology arm, documents the GigaChat family in a June 2025 technical paper — a domestic Cyrillic-centric model stack serving the regional market.
View source →Sanctions continue to block legal high-density GPU imports, forcing reliance on gray-market supply and pre-embargo hardware, capping training-cluster scale.
View source →Russia's national AI strategy runs to 2030 with implementation concentrated in state-backed champions (Sber, Yandex), prioritising domestic substitution over international integration.
View source →Used for the workforce measurement. The IMF combines schooling, pay, labour mobility and active labour-market policies; the Index rescales it so the weakest of 173 economies scores 0 and the strongest 10.
View source →Fossil, nuclear, hydro and other dispatchable plant; averaged with generation for firm power.
View source →Bottlenecks
These hold Russia below its Potential. Its two lowest scores cost it the most: hardware & silicon access at 1.5 (part of Watts) and frontier model output at 4.3 (part of Weights).
Assessment
Forced into isolation, Russia relies heavily on domestic model weights coached by local search/banking giants (Yandex and Sber) and locally generated energy. However, severe sanctions lock it out of global advanced foundry/GPU supply corridors, while massive talent emigration severely caps its intelligence potential. Its sovereignty is highly insulated but structurally resource-constrained.
The art of the possible
Russia's realistic upside is resilience rather than parity: domestic models serving the Russian-language world, and compute built on cheap energy within sanction constraints. Any thaw in chip access would change its trajectory faster than any domestic lever.
Strategic play — Siege Autarky
Settle for resilience rather than parity: domestic models for the Russian-language world, compute built on cheap energy inside the sanctions, and the negative leverage of denial and disruption. Any thaw in chip access would change its position faster than anything it can build alone.
Leverage — the nearest position. Sanctioned out of the chain, with no chokepoint left inside it. What remains is negative leverage — denial, disruption, and the mirror copy of the same battlefield data Ukraine now licenses to the West.
Open Russia in the live Index → — every source behind every number, and the comparison tool.