MTS

Wiki/Primary documents

Qimonda AG insolvency — January 23, 2009

Date: January 23, 2009. Filing court: Amtsgericht München (Munich district court). Capacity removed from global DRAM supply: ~10% of global bit supply at peak.

Why it matters: The single most-cited prior-cycle datapoint for "what producers remember." A major Western DRAM maker, well-capitalized, technologically credible (a Siemens / Infineon spinoff, IPO'd on NYSE Aug 2006 as QI), was gone in a single quarter when the 2008 cycle turned. Qimonda is the proximate evidence that "disciplined and well-funded" is not sufficient protection against a sharp demand miss. Every senior memory executive on every earnings call today was at a producer when Qimonda filed — this is institutional memory, not historical reading.

Primary sources:

  • Qimonda AG press release, January 23, 2009: "Qimonda Files Application for Opening of Insolvency Proceedings"
  • Qimonda 20-F filings (foreign private issuer)
  • 6-K filings announcing the insolvency
  • NYSE delisting notice, February 2009
  • Infineon 6-K, January 2009 (disclosing impairment on Qimonda stake)
  • Contemporaneous Reuters and FT coverage

What happened

  • Aug 2006: Infineon spun off its DRAM business as Qimonda AG. IPO at NYSE.
  • 2007–2008: DDR2 1Gb contract prices fell from ~$2.30 (mid-2007) to ~$0.80 (late 2008). ~65% price decline.
  • October 2008: Qimonda announced closure of its Richmond, VA fab — ahead of the bankruptcy.
  • January 23, 2009: Qimonda filed application for insolvency proceedings at Amtsgericht München.
  • 2009–2010: Dresden 300mm fab assets sold piecemeal. Some technology IP and patents acquired by Micron and others.

The shape of the failure

Qimonda was not technologically behind. It had a credible 300mm fab in Dresden, the Richmond US footprint, and access to Infineon's institutional engineering. What it could not survive was a sharp demand miss layered on top of overcapacity:

  • The industry had been adding capacity through 2006–2007 against assumptions that did not hold.
  • DDR2 contract prices fell 65% in eighteen months.
  • Qimonda lacked the balance-sheet depth of Samsung (semiconductor-division losses absorbed by other Samsung Electronics businesses) or Micron (older fabs, lower marginal cost).
  • It lacked the strategic-national protection that later kept Elpida partially supported through 2011 before its own filing.

Capacity removed

Qimonda was estimated at ~10% of global DRAM bit supply at peak. Its insolvency:

  • Removed a major Western competitor from the merchant DRAM market.
  • Accelerated the consolidation that would, three years later, eliminate Elpida (see elpida-bankruptcy.md) and leave the merchant market at three majors plus a Taiwanese tail.
  • Shifted the industry's competitive equilibrium toward the survivors, who collectively had less reason to add capacity speculatively.

Why it sits in this wiki

Two reasons:

  1. It is the precedent for the Stargate-stranded-capex failure mode. Qimonda built capacity against demand assumptions that did not materialize. The Stargate LOI (stargate-loi.md) targets ~40% of global DRAM output against assumptions about OpenAI's funding and revenue trajectory that have not yet been validated. If those assumptions fail, the capex Samsung and SK Hynix are building today against Stargate is at structural risk in the same way Qimonda's Dresden fab was.

  2. It is the institutional reason for the modern discipline doctrine. When Mehrotra says "structural, not cyclical," he is not making an analyst-style framing. He is making a statement that places Micron's capex policy on the opposite side of the Qimonda line. Capex discipline at Micron, SK Hynix, and Samsung today is the equilibrium response to having watched Qimonda die — and three years later, Elpida.

What to read next