← AI in the supply chainC4 · LESSON 6 OF 7

C4, Lesson 6: Supplier Risk and Disruption: Seeing the Weak Link Before It Breaks

Tags: Sourced / General / Our view / Illustrative. About 13 minutes.

1. The story: the Thailand floods and the hard drive (October 2011)

In its quarterly filing, Western Digital said it suspended production at all of its Thailand manufacturing facilities during the week of October 10, 2011, after severe flooding. The flooded sites included its magnetic head slider fabrication plants, which supplied a "substantial majority" of its head needs, and its hard drive assembly sites. About 60% of the hard drives it made in the September 2011 quarter came from Thailand. It also reported shortages of parts from other suppliers in flooded Thai industrial parks. [Sourced: Western Digital quarterly report, subsequent event note, SEC filing, https://www.sec.gov/Archives/edgar/data/106040/000095012311093040/R17.htm] Its annual report says production restarted in the March quarter of 2012, slider production was extended into Malaysia, and the flooding cost $214 million in net charges in fiscal 2012 ($119M asset impairments, $61M recovery, $28M inventory write-downs, $27M wages during shutdown, less $21M insurance and reimbursements). [Sourced: Western Digital 2012 Annual Report, p. 44, https://materials.proxyvote.com/Approved/958102/20120917/AR_142253/html/44.html] Across the industry, analyst firm IHS iSuppli reported hard drive shipments fell 29% in the fourth quarter of 2011 and average selling prices rose 28%, from $51 to $66; it forecast that prices would stay above pre-flood levels until 2014, partly because of industry consolidation. That was a forecast, not an outcome: by April 2013 ExtremeTech, using whole-market retail price averages from Dynamite Data, reported hard drive prices back at pre-flood levels. [Sourced, secondary: IHS iSuppli figures as reported by Computerworld, June 6, 2012, https://www.computerworld.com/article/1540261/hard-drive-prices-to-remain-high-until-2014.html ; outcome: ExtremeTech, Apr 19, 2013, https://www.extremetech.com/computing/153879-storage-pricewatch-hdds-back-to-pre-flood-prices-ssds-grow-as-gb-holds-steady] Notice what the filings do not show: a company that simply lost. Western Digital's fiscal 2012 revenue rose 31% to $12.5 billion, helped by an acquisition (HGST) and higher prices; the annual report gives average hard drive price up from $45 to $62. A disruption hurts buyers and unprepared sellers; it can help whoever still has supply. [Sourced: same annual report; our reading] Buyers felt it as price, not just delay. PC makers signed long-term agreements locking in supply at prices IHS said were about 20% above pre-flood levels. [Sourced, secondary: IHS iSuppli via Computerworld, June 2012]

2. The one idea

Your supply chain is only as strong as its least visible dependency. The risk is rarely your direct supplier; it is the supplier's supplier, and the single place many suppliers share (here, a few flooded industrial parks). AI helps by scanning more of the map and more news, but deciding how much insurance to buy is a business choice. [General]

3. The kitchen-table version

You sell $12 water bottles. One factory makes 100% of them. If its roof floods, you have nothing to sell for weeks. A second factory costs more per bottle. The question is how much of that extra cost is worth paying to sleep at night. [General]

4. What AI actually does here

  • Mapping: reading contracts, shipping records and public filings to guess who supplies your suppliers.
  • Monitoring: scanning news, weather and port data for events near those sites, and ranking alerts.
  • Scoring: giving each supplier a risk score from finances, location, delivery history.
  • What it cannot do: know a fact nobody recorded. If a key part comes from an unlisted sub-supplier, no model sees it unless data exists. Always ask a tool what data it is using. [Our view]

5. Worked example: is a second supplier worth it?

All inputs [Illustrative], computed by script. A small store sells 1,000 bottles a week at $5 margin each.

  • Risk: assume a 5% chance per year that the single factory is out for 12 weeks.
  • Single source: a 12-week outage loses 12,000 bottles, $60,000 of margin. Expected loss per year: 5% x $60,000 = $3,000.
  • Split 70/30 between two factories: the second factory can cover half of the first's volume after 4 weeks. Lost bottles: 5,600, or $28,000. Expected loss: $1,400. Expected saving: $1,600 a year.
  • Cost of splitting: if the second factory charges 5% more on its 30% of volume at a $6 unit cost, that is 0.3 x 1,000 x 52 x $6 x 5% = $4,680 a year. So on average, the second supplier does not pay for itself here. The honest case for it is the bad year: a $60,000 loss may be survivable or may close a small shop. Companies buy insurance against the tail, not the average. That is a decision, not a calculation. Change the inputs (probability, outage length, margin) and see where it flips. [Our view]

6. Choices you have (and what each costs)

  • Second source: cost of a price premium and more management; speeds recovery.
  • Safety stock: in the example, 4 weeks of the main supplier's volume at $6 is $16,800 tied up; at an assumed 25% yearly carrying cost, $4,200 a year. [Illustrative]
  • Move production closer or spread places: slower and costly; think about shared locations, not just company names. The WD story is about one region.
  • Contracts and insurance: Western Digital had property and business interruption cover and said its final losses might exceed its limits. [Sourced: 2012 Annual Report]
  • Do nothing: a legitimate choice if the tail is survivable; make it on purpose.

7. What it costs and what it breaks

  • Data work and mapping effort; many suppliers will not share their own suppliers.
  • Alert fatigue: if the tool flags everything, people stop reading.
  • False comfort: a risk score is a guess from available data.
  • Concentration can come back by the side door (one port, one region, one component).

8. Monday checklist

  1. List your top 10 products and who supplies each.
  2. For each, ask: if this supplier vanished for 12 weeks, what happens?
  3. Ask your top 3 suppliers where their critical parts and plants are.
  4. Pick one product and price a second source and a safety stock; compare to the loss.
  5. Write down how you would learn of a problem at a supplier (who calls whom).

9. Try it yourself (no code, 30 minutes)

Draw the supply of one product on paper: you, your supplier, their supplier if you know it, and their location. Mark any place that appears twice. That is your weak link.

10. Self-check

  1. Why did the Thailand floods hit companies that were not in Thailand? Shared parts and sliders; the WD filing says other plants outside Thailand faced slider shortages.
  2. What share of Western Digital's September 2011 quarter hard drives came from Thailand? About 60%.
  3. Did the floods mean every drive maker lost money? No; prices rose and WD's fiscal 2012 revenue rose, though it booked $214M in net flood charges.
  4. In the example, why is the second supplier not obviously worth it? Expected saving $1,600 a year versus $4,680 premium; it is about the bad year.
  5. Name one thing AI cannot see. A dependency nobody recorded.

11. Where this is going

Better data sharing and scanning may shorten the time to spot a problem. Concentration (a few makers, a few places) will stay the thing to watch. [Our view]

Sources

Review flags

  • IHS iSuppli figures are second-hand (press coverage of a paid report); I did not see the report.
  • RESOLVED (Oct 7, 2026): ExtremeTech, Apr 19, 2013 (read) reports retail HDD prices back at or slightly below pre-flood levels 18 months after the floods (Dynamite Data whole-market averages). The earlier "makers' average prices stayed higher" clause could not be sourced and was removed from the lesson. IHS iSuppli's own report was not seen; the lesson cites it via Computerworld and Tech Report. Seagate's experience and the Thai-flood revenue claims for WD were not rechecked; the lesson does not claim the floods caused WD's rise.
  • WD's revenue rise in fiscal 2012 is partly the HGST acquisition; I do not claim the floods caused it.
  • All example probabilities, volumes and costs are invented to show the method.
  • Nothing here says how much supplier-risk scoring tools actually reduce losses; I have no neutral source for that.

Claim ledger (C4 Lesson 6)

#ClaimSourceVerifiedStatusRe-check
1WD suspended Thailand production week of Oct 10, 2011; ~60% of Sept-qtr drives from ThailandWD 10-Q note (SEC)2026-10-06Readnone
2Production restarted March qtr 2012; $214M net flood charges; breakdownWD 2012 annual report p.442026-10-06Readnone
3WD FY2012 revenue +31% to $12.5B; ASP $45 to $62Same2026-10-06Readnone
4Industry shipments -29% Q4 2011, ASP $51 to $66 (+28%); LTAs ~20% above pre-floodIHS iSuppli via Computerworld2026-10-06Secondary, report not seennone
5Outage, margin, supplier and carrying-cost arithmeticOur computation2026-10-06Illustrative, script-checkednone