The title of this post was the single line summary of Dan Niles' quarterly outlook for the semiconductor industry. Dan is the founder and portfolio manager of AlphaOne NexGen Technology Fund. Each quarter GSA has a conference call with him where he presents what he is seeing. He starts from big picture stuff like unemployment and interest rates and works down to the outlook for the semiconductor industry.

The title of his presentation this quarter was:

Tariff Pull-Ins, Demand Fal-Off, and Excess Inventory—Not the Triple Crown You Want

Note that everything below is what Dan said unless it is [in square brackets] when it is my additional commentary.

General Economic Conditions

In the second half of this post, I will go into more detail on what Dan said about the semiconductor industry and the end-markets that it serves. But here is the big picture in a few bullets (this is 35 slides condensed into half that many lines).

• "We believe that the US and Global economy slows but a recession is not likely in 2019. But we are not convinced that the S&P500 has seen lows for the correction."
• US unemployment near 50-year lows. Job losses near 50-year lows.
• US home inventories are at extremely low levels which is cushioning against mortgage rates being at 7-year highs.
• Commodity prices low and stable. Global leading indicators already rolling over (turning bearish)
• Central Banks helped create over $13T in sovereign debt, and $100B in corporate debt, with negative yields. Rates are now rising. This bond market bubble was created over 35 years and its deflation could be painful [bonds go up when interest rates go down, and we have had 35 years of declining interest rates...until now]. The graph below shows the US 10-year bond yield since 1962.
• Corporations have levered up at the wrong time (taking on debt, raising dividends, share buybacks). "We worry that corporate debt is the new subprime threat in 2019."
• Shiller PE (price/earnings) near 1929 levels (but they were higher still in 2000)

Semiconductor Market

One uncertainty is that tariffs pulled some demand from 2019 into 2018, which means there is excess inventory to be worked through. End-demand is below expectations with both smartphone and PC markets shrinking even before considering Apple's recent announcement about China. Semi sales were up 20% year on year in 2018 [driven a lot by memory pricing] but set to slow sharply. About 90% of semiconductor companies that have reported lowered their numbers [and TSMC reported after Dan's presentation and lowered theirs].

Poor end market unit growth versus strong semiconductor sales reconciling:

• PC demand now negative year-on-year for 7 years. Dan expected it to be up this year, but that didn't happen.
• Smartphones grew 28-58% in 2011-14 but down year-on-year for 2018 with ASPs and profit contraction likely in 2019. There is likely to be an additional trough due to the transition to 5G, but 5G not yet ready.
• Transition to self-driving cars drives huge semiconductor content gains, but not enough (yet) to offset PC/SP.
• 3B connected devices in 2013, 5B in 2015 with about 20% growth through 2020.
• 1 trillion cumulative IoT devices shipped by 2035

The inventory situation is not pretty:

• Apple inventories up 31% year-on-year, with TSMC and HonHai [FoxConn] up about 40%. Apple pre-announced negatively for the first time since 2002.
• PC OEMs not bad.
• Tight memory supply is now excess supply until mid-2019. Hynix inventory up 49%, Micron 20%.
• Communications inventory seems good.
• Automotive inventory is slightly elevated with poor demand. China automotive declined for the first time since records began in 1990 [and China is the biggest automotive market, bigger than the US, Japan, and Germany combined]

Issues in smartphones and memory will have a large impact on semiconductor companies:

• Smartphone units are now shrinking along with PCs and combined are 35% of demand.
• China is 15% of global GDP but 25% of semi consumption with 25% of demand for PCs and handsets, 30% of auto, 20% of infrastructure. The overall economic situation in China is uncertain. Things will get really ugly in Q1, especially after Chinese New Year [February 5th].
• Excess in memory markets will have an outsized impact on total semi revenues and capex [equipment] spend.

Capital expenditures in semiconductor have been high for the last couple of years, and that will show up in supply starting this year. Even though capex is being cut back now, that won't help until 2020 since the 2017/8 capital expenditure is showing up this year.

Capex. That supply from a year ago shows up in supply. So even though cutting back hard now won’t help until 2020, you will get 2017/8 spending showing up in 2019. See graph below.

Conclusion

First, some uncertainties that we have to wait and see how they play out:

• 2017 capital expenditure was up 38%, and then up 10% again in 2018. That will cause oversupply problems of some sort in 2019.
• How much inventory is in Apple and its supply chain? It was already high before they pre-announced negatively early in January.
• Dan is now hearing of inventory digestion at the largest hyperscale vendors as their capex slows.
• Automotive is not a large memory end-market, but 10 big global OEMs [car companies] have built up inventory, which would not normally be a big problem except China units [cars] down 3% year-on-year for the first decline in 18 years of records
• Lead times started stretching at the start of 2017, which helped drive a buildup of inventory. Leadtimes declined in 2018, but then concern over tariffs in late 2018 led to further inventory buildup despite slowing demand.
• "Made in China 2025" and the backlash from the US (eg ZTE [and Huawei]) is a Sputnik moment for China. This is a big problem for the US semiconductor industry going forward, much like what Japan did to memory in the 1980s and Korea in the 1990s.
• The longest ever DRAM rally (8 quarters) ended in Q4 as the spot versus contract discount rose to a double-digit percentage.

Though the global economy is in better shape due to recent Fed action and China stimulus, a correction is likely through mid-2019 for inventories to be worked through. Severity could be worse than currently anticipated, especially in the lull post-Chinese New Year.

The overall conclusion is the title of this post: the first half of 2019 is likely to be really bad.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.