How we score AI bottlenecks: the 5 factors behind every tightness score
Every one of our 56 bottlenecks has a tightness score from 0 to 100. It is an editorial judgement, but it follows the same five questions every time.
1How concentrated is supply?
One or two suppliers means less room to add capacity and more pricing power.
2How long does it take to add capacity?
Lead times range from months (some chips) to more than five years (turbines, grid connections).
3How much capital does new supply need?
Fabs and grids cost billions, which slows how fast competitors can enter.
4How exposed is it to geopolitics?
Export controls and single-country supply raise the risk of disruption.
5Do suppliers have pricing power?
Rising prices and sold-out order books are the clearest evidence of a real shortage.
Why a score at all
Comparing a memory shortage with a turbine shortage is hard without a common scale. A single 0–100 score makes it possible to rank 56 bottlenecks, spot which are getting tighter and see whether the market has priced them in.
How scores change
We revisit scores when lead times, capacity plans, prices or policies change, and every bottleneck page names the evidence that would lower its score. Tightening, stable and easing trends show which way each one is moving.
The method page also lists every data source used on the site.
Frequently asked questions
Is the tightness score market data?
No. It is our editorial judgement, combining these five factors with company commentary and industry data.
What does 'easing' mean?
Supply is catching up, which usually shifts bargaining power to buyers.
Can a score be wrong?
Yes. Every bottleneck page lists what would prove the call wrong, and we change scores when the evidence changes.
Get the daily brief
Everything that happened across the AI supply chain, in a 3-minute email every weekday. Free.
More guides
Figures quoted are as of the publish date; the linked pages show live data. Educational research by Singularity Research, not investment advice.