Applied Materials And Lam Research: GS Upgrades Speculative, Premature, And Unwarranted

SeekingAlpha

Applied Materials And Lam Research: GS Upgrades Speculative, Premature, And…

According to a Seeking Alpha news report entitled “Goldman turns bullish on semi equipment:”

Goldman Sachs, citing early signs of memory stabilization and expecting NAND and DRAM margins to bottom in Q3 and Q1 2020, respectively, upgraded:

  • Applied Materials (AMAT) from Neutral to Buy and adds the company to its Conviction Buy List. The firm increases AMAT’s price target from $48 to $56, a 17% upside.
  • Lam Research (LRCX) from Neutral to Buy with a target of $231 from $197.
  • KLA-Tencor (KLAC) Sell to Neutral with a $130 PT (was: $104).

The only way these equipment companies will benefit from “NAND and DRAM margins bottoming” is if the memory companies, primarily Micron Technology (MU), Samsung Electronics (OTC:SSNLF), and SK Hynix (OTC:HXSCL) increased capital expenditure (capex), specifically to purchase processing equipment from these companies.

This article will present my thesis that the upgrades of AMAT and LRCX were premature and speculative, ignoring significant cuts in capex in 2020 by MU and HXSCL. Further, the GS upgrade was tied to elevated WFE (wafer front end) equipment purchases from TSMC (TSM) and logic and foundry companies. However, the analyst neglected to understand that EUV lithography from ASML (ASML) would be a major beneficiary, which would curtail deposition and etch equipment from AMAT and LRCX.

Memory Capex For 2020 Pure Speculation

The GS analyst cited in the above-referenced note that he saw early signs of memory stabilization, expecting NAND and DRAM margins to bottom in Q3 and Q1 2020, respectively. Thus, MU’s earnings disclosure and guidance are important factors in the upgrade of AMAT and LRCX companies. The premise of the upgrade was an expected increase in WFE for memory companies, and GS raised its estimates of NAND capex spend from 10% to 15%, while DRAM capex spend was raised from -5% to 0% in 2020.

This estimate on 2020 capex spend, which includes buildings and equipment, is just an estimate because no memory company has explicitly stated what it plans to spend in 2020. In fact, even actual 2019 capex spend is not finalized with five months more to go in the year.

In its recent earnings call, MU CEO Sanjay Mehrotra stated:

Earlier this year, we announced a reduction in fiscal 2019 CapEx forecast from $10.5 billion plus or minus 5% at the start of the year, to approximately $9 billion now. For fiscal 2020, we plan for CapEx to be meaningfully lower than fiscal 2019. While our CapEx plans are still being finalized, we seek to balance our manufacturing investments with our free cash flow objectives.”

Note that MU’s 2020 fiscal year mentioned in the comment ends August 31, 2020, so MU, at this point, plans a lower capex for 2020 on top of a lower 2019.

Samsung Electronics will not report its CY 2Q earnings until July 31, 2019, so any capex comments we have are almost three months old. Following its 1Q earnings, the company said in a conference call that instead of expansion, it will focus on “migration.” Samsung will implement capex “according to market conditions,” although memory-related spending would “significantly fall” while continuing with longer-term investment.

That statement was for 2019, and we haven’t a shred of evidence what 2020 capex will be. But we do know that memory capex will “significantly fall.”

SK Hynix reported its CY 2Q earnings on July 25, 2019. Following its Q1 financial announcement, the company said it will maintain capex plan conservative for the upcoming months, as Cha Jin-Seok, the chief financial officer of SK Hynix noted:

For DRAM, the company will respond to demand growth for this year with technology migration without any incremental wafer capacity. As a part of this plan, we will utilize recently opened WuXi extension fab C2F to make up for the capacity loss resulting from tech migration.”

SK Hynix recently announced Q2 2019 earnings are shown in Table 1. Revenues were down 5% QoQ and 38% YoY. Importantly, Operating Profit was down 53% QoQ and 89% YoY. Based on these financials, my suspicion is that SK Hynix’s capex will be down again in 2020.

Table 1 – SK Hynix Financials – KRW (Billions)

Q2’19

Q1’19

Q2’18

QoQ

YoY

Revenue

6,452

6,773

10,371

-5%

-38%

Cost of Goods Sold

4,435

4,092

3,762

+8%

+18%

Gross Profit

2,017

2,680

6,608

-25%

-69%

SG&A Expenses

1,379

1,314

1,034

+5%

+33%

Operating Profit

638

1,366

5,574

-53%

-89%

Source: SK Hynix

The company’s CFO indeed reported following the earnings call that “capex for next year would be “reduced considerably” from this year.”

In other words, the expectation of an increase in memory capex, the foundation of the upgrade of AMAT and LRCX, has no basis of fact.

I wrote a Seeking Alpha article on June 25, 2019, discussing the capex issue, entitled “Micron: Capex Guidance Will Be Critical To Semiconductor Equipment Suppliers,” pointing out the importance of capex on a number of fronts.

Table 2 shows expected capex spend for memory companies for 2018 and 2019. I estimate MU’s 2020, which is the only data point I have based on CEO Mehrotra’s comment above that “capex be meaningfully lower than fiscal 2019.”

Table 2 – Capex of Companies

2018

2019

% Change

2020

% Change

Samsung

10,461

6,108

-41.6%

SK Hynix

14,588

10,828

-25.8%

Micron

9,085

9,097

0.1%

7,606

-16.4%

Toshiba/WDC

5,594

4,591

-17.9%

Renesas

575

319

-44.6%

Nanya

634

271

-57.3%

40,937

31,214

-23.8%

Source: The Information Network

Non-memory Headwinds for AMAT and LRCX

The GS upgrade of AMAT and LRCX includes chips other than memory, and the report notes that the company maintains its WFE growth of 10% in logic/foundry and moved up its foundry WFE growth from -5% to 0% in 2020.

Outside of memory, GS forecasts capex spend to be stable in logic/foundry, maintaining its 2020 advanced logic/foundry spend, primarily for TSMC, Intel (INTC), and Samsung’s foundry business in 2020 growing just 4%.

These three semiconductor companies had revenues of $34.9 billion of the total global semiconductor revenues of $68.5 billion in Q1 2019. Thus, with a 51% share of the market and capex spend of just 4% growth in 2020, benefits to AMAT and LRCX should be minimal.

Technology trends by these three companies will also be a headwind to AMAT and LRCX; namely EUV lithography. I discussed in several Seeking Alpha articles EUV lithography as a headwind for these companies. The most recent was published on March 19, 2019, entitled “Canon’s Nanoimprint Lithography: A Chink In ASML Holding’s Armor” I noted:

“To process a semiconductor device at 7nm dimensions requires just one exposure step with EUV versus multiple exposures using ArF immersion. The latter uses successive litho-etch-litho-etch (LELE) steps in a double patterning process or litho-etch-litho-etch-litho-etch (LELELE) in a triple patterning process. An LELE process costs about 2.5 times a single exposure patterning process, while an LELELE costs about 3.5 times that of a single exposure. Multiple patterning extends IC scaling, but the increased complexity of using more process steps at each node translates into time and cost.”

According to our report entitled “Sub-100nm Lithography: Market Analysis and Strategic Issues,” these three companies, which hold a 50% share of the semiconductor market (revenues), already have installed 27 EUV lithography systems out of 32 sold by ASML. Through 2018, these 27 represent 84% of all systems. Past and future EUV lithography systems are listed in Table 3.

Table 3 – EUV Lithography System Installs

2016

2017

2018

2019E

2020E

Intel

0

3

4

3

8

Samsung LSI

1

3

6

4

7

TSMC

2

3

5

18

12

Source: The Information Network

According to TSMC’s SVP and CFO Lora Ho:

Our N7+, which adapts EUV for a few critical layers, has already entered volume production. We expect our customers’ end products using N7+ will be in the market in high volume this quarter. We expect strong demand to continue into next year.”

Investor Takeaway

Goldman Sachs, citing early signs of memory stabilization, expect NAND and DRAM margins to bottom in Q3 and Q1 2020, respectively. The analyst used this argument to upgrade AMAT and LRCX. For AMAT and LRCX to benefit from NAND and DRAM margins to bottom, memory companies would need to increase spending of equipment.

There are two issues:

  1. Are DRAM and NAND margins bottoming?
  2. Have memory companies announced an increase in capex spend for equipment?

Table 1 above shows that SK Hynix’s margin dropped 53% QoQ and 89% YoY. This is not a bottoming of margins.

Table 4 shows financial data for Micron Technology and notes that for FQ3-19, the most recent quarter reported last month, gross margin was 39% versus 50% the previous quarter and 61% a year earlier. QoQ margin decreased 55% and YoY decreased 40%. Again, as with SK Hynix, this is not a bottoming of margins.

Table 4 – Micron Technology Financials

FQ3-19

% of Revenue

FQ2-19

% of Revenue

FQ3-18

% of Revenue

Revenue ($M)

$4,788

100%

$5,835

100%

$7,797

100%

Gross margin

1,884

39%

2,928

50%

4,750

61%

QoQ

-55%

YoY

-40%

Source: Micron Technology

The other factor leading to the GS upgrade was positive news coming from TSMC. Indeed, TSMC reported a stellar quarter, but concomitant with financials is disclosure by TSMC that its EUV technology is gaining traction and moving into high-volume production. The company is also moving to EUV evaluation for 5nm, showing that going forward, deposition and etch equipment from AMAT and LRCX would be minimized.

Further, TSMC, Intel, and Samsung (for its logic foundry) purchased 27 EUV systems between 2016 and 2018. At $100-130 million each, these companies combined did not spend more than $3 billion to have them sit idle. This EUV trend presents a significant headwind to AMAT and LRCX.

To summarize, the notion of bottoming of memory margins is not apparent at this time, and tying that premise to increased equipment spending benefiting AMAT and LRCX is premature and speculative. To date, MU and SK Hynix have given only a hint of capex spend for 2020.

Also, the GS analyst failed to consider that TSMC and others are moving to EUV, which minimizes the need for deposition and etch equipment from AMAT and LRCX.

I will present an analysis of KLAC, the third company upgraded, along with its peers, after the company’s earnings call on August 5, 2019.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Source

admin