Machinepower Index / Dossiers / Poland
Rank 22–27 across the scoring rules tested. How this is tested
Poland
Diffusion Advantage — For nations that cannot build the frontier but can use it faster than anyone: import the models and win on deployment, across the state and the economy, with public consent.
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: Poland controls 1,200 H100-equivalents, wherever the sites are. 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 | 5.0 |
| Firm Power Capacity | Dataset | 4.8 | |
| Sovereign Compute | Dataset | 4.8 | |
| Hardware & Silicon Access | Dataset | 4.7 | |
| Weights | Frontier Model Output | Dataset | 3.0 |
| Alignment Capability | Assessed | 5.4 | |
| Regulatory Standing | Dataset | 8.5 | |
| Talent Density | Assessed | 5.6 | |
| Will | Policy Capacity | Dataset | 9.6 |
| Public Sector Adoption | Dataset | 7.5 | |
| Workforce Transition | Dataset | 6.8 | |
| Public Trust | Dataset | 5.6 |
Assessed marks a measurement we judged rather than took from a dataset: 3 of the 12 here.
Sources
The data-centre connection queue is estimated at several times national peak demand.
View source →One of two EU states to set up a new body to supervise the AI Act; no safety body.
View source →A cross-check for planning and permitting, which stays assessed: it scores permits and electricity connections for ordinary firms, not grid-scale connections for data centres.
View source →Largest: Helios.
View source →Stanford does not count Poland. Estimated from 2.1 million GitHub developers and Oxford Insights' diffusion pillar, averaged with an estimate from LinkedIn's AI talent share (1.01% of members), calibrated on the nations Stanford counts.
View source →Feeds governance and public-sector adoption; policy capacity is listed separately.
View source →The Index scores public trust as 1 + share ÷ 10. Survey of 23,532 adults across 32 countries, March to April 2026.
View source →Feeds workforce, rescaled across 173 economies.
View source →Fossil, nuclear, hydro and other dispatchable plant; averaged with generation for firm power.
View source →Second source for public trust, averaged with Ipsos.
View source →Half of public trust: support for building more data centres, scored as 1 + share ÷ 10 and averaged with trust in AI.
View source →Bottlenecks
These hold Poland below its Potential. Its two lowest scores cost it the most: frontier model output at 3.0 (part of Weights) and hardware & silicon access at 4.7 (part of Watts).
Assessment
Poland has two million developers and a state that already runs its own models in public services, on a grid too coal-heavy and congested to connect the data centres queuing for it.
The art of the possible
PLLuM shows the state can build for its own needs. The gain is in putting it to work across government and industry faster than richer neighbours; grid reform decides whether the compute sits in Poland.
Strategic play — Diffusion Advantage
Deploy before you build: put PLLuM and imported models to work across the state and industry, and spend scarce grid capacity on compute that serves that use.
Leverage — the nearest position. No chokepoint. Its nearest key is people: Polish engineers helped found OpenAI and run its research. People can leave, which is why it is not leverage.
Open Poland in the live Index → — every source behind every number, and the comparison tool.