Fab Delays and Cancellations Reduce Our Display Equipment Spending Forecast
We have been alluding to some fab delays and cancellations in some recent articles in the DSCC Weekly Review, so in this article, we will try and quantify it. Why is this happening? This is happening because of excessive TV panel inventory in Q3’19 and Q4’19, at a time when prices are already low, manufacturers are losing money and more capacity is coming online. Capacity for 7G+ fabs has been rising at a double-digit rate throughout the first 3-quarters of 2019 due to G10.5 fabs, widening the oversupply. As a result, panel makers have been forced to reduce their utilization at a time when they are already losing money. Panel suppliers were losing money at higher utilization levels and are now losing more money with both utilization lower and prices lower. As a result, they don’t need and can’t afford to expand capacity further, widening the oversupply and their losses. We are finally at the point where market forces dictate that capex stops, despite the presence of subsidies. As shown in this Crystal Cycle chart I created 13 years ago, when prices fall and companies lose money, there should be under-investment and little capacity added which will create a shortage. The “under-investment” stage is where we are moving to, so that a shortage is created and prices will rise. Once prices get higher and manufacturers become increasingly profitable, they look to make more money by adding capacity, but because prices are high, it causes demand to slow which leads to another surplus and prices fall repeating the cycle. I would say we have had an extended over-investment cycle in LCDs from 2016-2019 due to China’s subsidies, which should result in a longer under-investment cycle. However, because Korean manufacturers have decided to shut down a significant amount of existing capacity as well, rather than just delay future investments, it could result in a shorter under-investment cycle. This is complicated by the availability of subsidies which may decline significantly after 2021. The lack of subsidies combined with Korea shutting down LCD capacity would restrict supply growth, result in shortages, more price increases and perhaps market-driven as opposed to subsidy driven investments, which would be a healthier situation in the long run. In any case, let’s move on to the new equipment spending forecast.
1997 – 2005 Crystal Cycle Example
Source: Ross Young’s presentation at DisplaySearch 2006 US FPD Conference
In 2019, we see HKC H2 downsizing its capacity from 150K all installed in 2019, to 120K. This will reduce our 2019 equipment spending forecast by $600M or 4%, producing a 29% Y/Y decline in 2019.
In 2020, we see more downsizing and delays.
- First, we expect Sharp to downsize its G10.5 fab in Guangzhou from 90K to 45K - 60K. We previously saw 60K installed in 2019 and 30K in 2020. If only the last 30K is cancelled, it will reduce our 2020 forecast by $1.6B. If it ends up being a 45K reduction, we will need to reduce 2019 by another $600M.
- Second, we need to cancel the Infintech 60K G5.5 a-Si LCD forecast as they have failed repeatedly to get funding and multiple equipment suppliers have informed us this investment will no longer happen. This reduces our forecast by another $1.0B.
- Third, BOE B17 Phase 4 was delayed by 6 months according to one supplier, which affects 15K of our forecast. This reduces 2020 by ~$620M and increased 2021 by $620M. However, another supplier has told us they don’t see this delay yet. It may only be applied to some tools, but we will make the change for now.
- Fourth, HKC H4 is downsizing from 150K to 60K according to one supplier we talked to, all in 2020. Another supplier we talked to believed it would be reduced to 90K – 100K, with 50K canceled. For now, we will assume 90K is canceled, reducing our forecast by $1.8B.
The sum of all these changes brings equipment spending down by 25% from $20.3B to $15.2B. This means that 2020 is only up 2% over 2019. However, there is one big potential change and that is SDC’s Q-1 line. We had originally thought that all the equipment would be installed in 2H’20, it now appears that at least some of the backplane equipment won’t get installed until 2021. On the other hand, we are learning that they are buying more backplane equipment than originally thought, 16 litho tools instead of 12, as well as a significant number of PECVD tools. The litho tools will all be supplied within 2020. So, we will leave this fab unchanged for now, but it could end up raising both 2020 a little and 2021 more when everything eventually shakes out.
Speaking of 2021, we see fab delays and cancellations that year as well.
- We have cancelled the Visionox V2-2 investment based on discussions with multiple equipment suppliers. One supplier indicated that there are no more subsidies for this fab. A couple of other suppliers did not expect a cancellation. This investment will reduce the spending forecast by $1.3B.
- We see CSOT T7 declining from 90K to 60K with 15K coming out of our 2021 forecast and 15K coming from 2022. These were expected to be oxide lines and more capital intensive at close to $900M for the first 15K and nearly $800M for the last 15K.
- We do add 15K at BOE B17 which boosts 2021 by over $600M.
As a result, we expect to reduce 2021 by 15% to $8.9B which will produce a 41% decline vs. 2020. However, we will likely increase 2021 more than that due to some Q-1 investments spilling over. 2022 is reduced by 8% as a result of the CSOT downsize to $8.6B, which would be down 4% vs. 2021.
As capacity tightens and prices rise, we do expect some of these downsized fabs to eventually get built. If capacity is tight enough, they may even get built without the need for subsidies. They are likely to be installed from 2022 on.
For details on spending at every fab from 2016-2024 by tool type including market share for most segments, please see our Quarterly Display Capex and Equipment Report.
Existing vs. New Preliminary Display Equipment Spending Forecast
Source: DSCC’s Quarterly Display Capex and Equipment Spending Forecast