From Big Glass to Small Panels

Key Thoughtz

• AMLCD fab investments are more fungible today than in the past.
• Undifferentiated assets create price pressure in multiple market segments.
• Producer prices decline faster than their cash costs do.
• Scale benefits decrease over time but aid companies with affiliated brands.

Read more in downloadable file(s):

AMLCD glass substrates are getting bigger while the panels are getting smaller. AMLCD producers have invested billions to build larger substrate fabs this decade. The ostensible goals have included scale benefits and the ability to produce larger displays that earn premium prices (and profits, perhaps). As plotted in the following chart, the average substrate size increased rapidly from 2004. That was the point in time when leading AMLCD producers obtained equity infusions through share or convertible bond offerings and geared up for the LCD TV market. The average thin-film transistor (TFT) substrate size increased 69% from 24.2" per edge in 2003 to 41.0" in 2007. The pace of growth slowed in 2008 as producers deferred investments but the average size reached 44.3" that year.

Weighted-average Glass Edge and Display Edge Development for AMLCD, 2000–2008

Note: WAGE is the square root of the average TFT glass area in inches. WADE is the square root of average display area.

Meanwhile, the average size of AMLCD panels increased 27% from 5.7" in 2003 to 7.3" in 2007. In 2008, the growth of average display size became only slightly larger at 7.4" per side. Slower panel size development was more notable in 2008 for large display applications (using ten diagonal inches and larger panels) because consumers sought bargain LCD TV sets and ultra-portable notebook PC. The average large panel size increased 4% to 15.1" in 2008. That was about one-half the sequential growth rates seen in 2005–2007.

Margin Squeeze

Premium prices for larger panels were part of the rationale for investing in larger substrate fabs. Unfortunately, larger substrates cost more per square inch than smaller ones do. Glass suppliers face greater yield risk and they must invest in new foundries, so costs go up. Overhead efficiencies (e.g. more output per engineer) offset some of this cost increase but the overall effect may be neutral. In addition, premium prices for 52” or larger TV panels have been harder to obtain as consumers in rich countries seek bargains and consumers in poor countries seek entry-level products.

As a result, new fabs have had less positive effect on producer profits than expected. The combined results for AU Optronics (AUO) and LG Display (LGD) illustrate such effects. If we add their reported area sold, we obtain the following chart. (Their results are added to offset the differences in their reporting currencies relative to the dollar and to represent the total industry better.) The diamonds plot sales per square foot and the squares plot cash cost (ignoring depreciation charges). Their combined sales per square foot have fallen 22% a year since 2005 but their cash cost has fallen slower at 21% a year. At those rates, producers are heading towards a zero margin condition despite increases in scale. Indeed, zero cash margin was reached in Q4’08 and in Q1’09. Fortunately, cash margins improved in Q2’09 but investing more seems to deliver less.

Combined Sales Revenue and Cash Cost for AUO and LGD (USD/ft²)

Source: AU Optronics and LG Display disclosures, 2005–2009.

Strategic Implications

Generating enough cash flow from operations to cover the cost of new fabrication plants becomes more difficult as cash margins narrow. Producers with less pricing power or less capacity (scale) than the leaders have may become unable to afford new plants without taking on large amounts of debt. Loans are hard to find today (outside China) and expensive. It therefore comes as no surprise that producers arrange project funding with brands.

What better place to situate ventures than China? That nation wants high-tech employment and infrastructure and it is eager to stimulate such development. There are several provincial networks for panel production and TV assembly, already. The time seems right for the panel industry’s center of gravity to shift from Korea/Taiwan to China. The two leading producers in Korea resemble Intel already: they retain key fabs in their native country while running back-end assembly in other countries. Taiwanese companies still face legal barriers to exporting assets or expertise but these may lessen as Taiwan strengthens its relationship with the mainland. After all, Chinese business culture is more oriented towards economic growth than towards economic value added.

In summary, the numbers often show a different picture than the ones we imagine while reading press releases. Despite exciting announcements of newer, larger substrates, panel producers have used these to make more panels for mainstream markets. Cash costs have not fallen as fast as commodity prices have, so profit margins have decreased. This condition might change when employment in Europe and the USA recovers and consumers start spending again. No one knows for certain when that will occur, but it may be several quarters from now. In the meantime, producers will be fighting for every penny of margin as more eighth and tenth generation fabs come on line.