Scenarios
Four stress cases over the AI-compute supply chain.
The 14 Saudi sites sit within 2,500 km of Khorramabad, the longest published range.
Reading the four layers
Range indicates geographic reach; the chart does not estimate a strike or an outage. Missile rings use published ranges from the recorded primary position. The Saudi circle is the co-location screening radius, and models no blast; dark lines join pairs with an assigned grid connection, and pale lines pairs that are only near each other. Japanese circles are illustrative stress-case radii; affected locations are chosen separately.
Taiwan is filled in red. Its rays group reached organisations by their recorded country; line width grows with the square root of their count. Endpoints are arranged for display, including countries outside this crop. A ray describes neither a physical shipment nor a dependency path. All returned country counts remain below, including Taiwan, unknown countries and countries without a drawn endpoint.
Scenarios
- Taiwan Strait4 assumed disruptions to TSMC, with live cascade results, inventory buffers and alternative fabrication capacity.Newest input June 2026
- IRGC Missile Range14 missile systems and 5 recorded launch positions; the longest range held here is 2,500 km.Newest input March 2026
- Saudi Grid2 pairs of sites share a named substation, 6 share a grid zone, and 4 are only within 200 km.Newest input August 2026
- Japan EarthquakeThe M9 Nankai Trough case selects 4 locations: Kioxia Yokkaichi, Tokyo Electron, Advantest and SCREEN.Newest input Q2 2026
The Taiwan naval blockade result has not loaded yet…
Related pages
Sources
4 sources
| Source |
|---|
| CSIS Missile Defense Project |
| USGS / HERP seismic hazard models |
| Scrutica facilities register |
| Licensed supply-chain database |
- CSIS Missile Defense Project: https://missilethreat.csis.org
- USGS / HERP seismic hazard models: https://earthquake.usgs.gov/
- Scrutica facilities register: https://scrutica.com/methodology
- Licensed supply-chain database: https://scrutica.com/methodology#licensed-substrate
How the figures were made
Scenario characterisation from published open-source data mapped against the facilities register and supply-chain topology; some cloud-region locations are approximate and some electricity connections are inferred from corridor proximity. 4 analyses, with input dates and recorded editorial reviews shown separately. Review cadences run 45–90 days.
Each analysis’s newest input against its last recorded review.
| Analysis | Newest input | Last reviewed | Next review due |
|---|---|---|---|
| Taiwan Strait | June 2026 | No recorded review | |
| IRGC Missile Range | March 2026 | 18 Jul 2026 | 1 Sep 2026, 71 days past |
| Saudi Grid | August 2026 | 18 Jul 2026 | 16 Oct 2026, 26 days past |
| Japan Earthquake | Q2 2026 | No recorded review |
Review triggers before the next scheduled pass. The 4 analyses have 9 watches; 4 are manual. Watches on facility records are checked automatically: a scan compares each watch’s country scope against facility records written since the analysis’s last recorded review and reports the watch as fired or clear. The manual watches cover changes no dataset here records (a strike, a change to the Saudi grid, a market-share revision); they are standing instructions to the reviewer and are never reported clear. An analysis with no recorded review has no date to compare against, so none of its watches can be evaluated until a review is recorded.
This page does not query facility observations. Automated watch verdicts are unavailable here; a page render is not an all-clear scan. Both the review ledger and complete watch register can be downloaded from Data.
Taiwan Strait. New or edited Taiwan facility records: capacity, operator, or status change at a leading-edge site (watched in Taiwan). New or edited facility records at the alternative foundries and packaging providers in the substitution assessment, whose capacity sets the estimated shortfall (watched in United States, South Korea, Japan, China). A tier-1/tier-2 publication of company-level leading-edge foundry share, which would supersede the 2021-vintage CSET country aggregate the concentration input currently uses (manual; no dataset here records this kind of change).
IRGC Missile Range. New or edited facility records across the Gulf, Israel and Turkey, the facilities compared against the published missile ranges (watched in United Arab Emirates, Saudi Arabia, Bahrain, Qatar, Kuwait, Oman, Israel, Turkey). A regional strike, interception episode or air-defence deployment change that requires the incident history or defence context to be reviewed (manual; no dataset here records this kind of change).
Saudi Grid. New or edited Saudi facility records, including any new data centre and the grid zone it is assigned to (watched in Saudi Arabia). A published change to Saudi transmission topology or substation assignment. A shared named substation is the strongest evidence class in the co-location table, so a reassignment changes which pairs qualify (manual; no dataset here records this kind of change).
Japan Earthquake. New or edited Japanese or Korean facility records: memory, wafer, materials or equipment sites (watched in Japan, South Korea). A revision to published DRAM or HBM market share, the figures this analysis uses for the Korean producers’ share of world memory supply (manual; no dataset here records this kind of change).