A Shortage Of NAND Flash Memory Is Coming Soon - What Caused It And What Will Be Its Impact

| About: Micron Technology (MU)
This article is now exclusive for PRO subscribers.

Summary

A shortage of NAND will most likely happen within the next two quarters.

New 3D NAND ramp-up coming on-stream may mitigate the shortage but it may take several quarters.

Investors need to be cautious because a shortage will result in NAND price increases and will also result in diminished sales.

According to NAND flash controller and solution provider Phison Electronics chairman Khein-Seng Pua:

There will be a shortage of NAND flash chips in the third quarter as most major vendors have not ramped up their 2D NAND flash chips over the past 18 months and their migration to the production of 3D parts has not been smooth."

According to Table 1 below, analysts from KDB Daewoo Securities forecast a shortfall of 5,032 million Gb in 2017:

Table 1 - NAND Supply and Demand

millions Gb

2014

2015

Change

%

2016

Change

%

2017

Change

%

SEC

21,290

32,450

52.4%

46,002

41.8%

59,973

30.4%

Toshiba (OTCPK:TOSYY)/SanDisk (SNDK)

25,458

34,040

33.7%

47,435

39.4%

58,302

22.9%

Micron (NASDAQ:MU)

11,047

14,944

35.3%

20,119

34.6%

26,075

29.6%

SK Hynix (OTC:HXSCF)

5,789

8,897

53.7%

13,095

47.2%

16,762

28.0%

Other

45

54

20.0%

66

22.2%

87

31.8%

Total Supply

63,629

90,385

42.1%

126,717

40.2%

161,199

27.2%

Smartphone/Tablet

15,971

22,665

41.9%

33,446

47.6%

45,265

35.3%

SSD

23,017

36,314

57.8%

51,422

41.6%

78,418

52.5%

USB/flash

14,344

16,523

15.2%

18,554

12.3%

20,170

8.7%

Other

8,515

11,645

36.8%

17,214

47.8%

22,378

30.0%

Total Demand

61,847

87,147

40.9%

120,636

38.4%

166,231

37.8%

Excess Supply

1,782

3,238

81.7%

6,081

87.8%

-5,032

-182.7%

If NAND supply and demand is analyzed on a month-to-month basis, the shortfall is anticipated to really kick-in in Q1 2017 with Q4 2016 almost at parity between supply and demand, as shown in Table 2 below:

Table 2 - NAND Supply and Demand

millions Gb

1Q2016

2Q2016

3Q2016

4Q2016

SEC

10,177

11,269

11,979

12,578

Toshiba/SanDisk

10,638

11,307

12,430

13,061

Micron

4,738

4,927

5,125

5,330

SK Hynix

2,803

3,105

3,423

3,765

Other

15

16

18

18

Total Supply

28,371

30,624

32,975

34,752

Smartphone/Tablet

6,890

7,559

8,883

10,114

SSD

11,177

11,848

13,857

14,541

USB/flash

4,277

4,500

4,828

4,949

Other

3,609

3,892

4,591

5,121

Total Demand

25,953

27,799

32,159

34,725

Excess Supply

2,418

2,825

816

27

Historically, since 2009, peaks in demand occur in the third quarter of each year. This happened in 2009, 2010, 2011, and 2012. The bullish Apple (NASDAQ:AAPL) investor would like to think that the shortfall was due to the introduction of Apple's iPhone that occurs in Q3 of each year. In fact, the shortage in 2012 lasted until Q2 2013, much of it due to Toshiba's decision to cut NAND production in mid-2012 by 30% to reduce inventory in the market and improve the overall balance between supply and demand. The shortfall in 2011 was due to the earthquake in Japan on March 11 that forced Toshiba to cut NAND production, and also due to a shortage of 300mm wafers to make the chips. I believe, however, that Apple was the cause of shortfalls in NAND in 2009 and 2010.

Figure 1 - Historical Supply/Demand for NAND

There were no NAND shortages in 2014 and 2015, although they were predicted, as we saw a slowdown in tablet and smartphone sales, flash cards, and USB/flash. Growth, however, was sustained in the SSD sector as they increasingly took market share from HDD sales.

The Daewoo forecast is corroborated by Phison's CEO that a shortage was coming. Perhaps he subscribes to Daewoo's outlook report, I don't know. But I suspect that he has deep insight into his supply chain and a shortage is coming, perhaps in Q3 as he suggests instead of Q1 2017 as the Daewoo forecast predicts.

NAND Capacity

What is responsible for the upcoming NAND shortage? According to Phison's CEO, it is due to most major vendors not ramping up their 2D NAND flash chips over the past 18 months and that their migration to the production of 3D parts as not being smooth.

Table 3 - NAND Wafer Capacity

12" eq. '000 wpm

2014

2015

Change %

2016

Change %

2017

Change %

SEC

3,981

4,380

10.0%

4,860

11.0%

5,280

8.6%

Line 11

1,440

1,320

-8.3%

1,320

0.0%

1,320

0.0%

Line 12

2,400

2,400

0.0%

2,400

0.0%

2,400

0.0%

C1 (Xi'an)

141

660

368.1%

1,140

72.7%

1,560

36.8%

Toshiba/SanDisk

5,010

5,190

3.6%

5,595

7.8%

5,710

2.1%

Fab 3

1,740

1,800

3.4%

1,860

3.3%

1,860

0.0%

Fab 4

2,430

2,400

-1.2%

2,400

0.0%

2,400

0.0%

Fab 5

840

990

17.9%

1,335

34.8%

1,450

8.6%

Micron

2,655

2,850

7.3%

3,000

5.3%

3,240

8.0%

Manassas

420

420

0.0%

420

0.0%

420

0.0%

Lehi (IMFT)

840

840

0.0%

840

0.0%

840

0.0%

Singapore (IMFT)

900

930

3.3%

1,080

16.1%

1,320

22.2%

Singapore (TechSemi)

495

660

33.3%

660

0.0%

660

0.0%

SK Hynix

2,115

2,160

2.1%

2,475

14.6%

2,640

6.7%

M11

1,560

1,440

-7.7%

1,440

0.0%

1,440

0.0%

M12

555

720

29.7%

1,035

43.8%

1,200

15.9%

Other

36

36

0.0%

36

0.0%

40

11.1%

Total NAND

capacity

13,797

14,616

5.9%

15,966

9.2%

16,910

5.9%

Let's take a closer look.

Samsung's (OTC:SSNLF) Xi'an fab, which cost $14 billion and took 20 months to complete, began operations on May 9, 2014. Referring to Table 3 above, we see the large growth in capacity at the fab in 2015 and continuing into 2017. Samsung's 8.6% increases were not influenced by its Line 11 or Line 12.

Amit Daryanani, an analyst at RBC Capital Markets, noted in March 2016 that:

"NAND chips in the Xi'an plant by the end of 2015 will have approximately a 30,000 capacity at 48 layers and 60 to 70,000 at 32 layers."

On the flip side, Daryanani did note that NAND conversion will lower the average selling price (ASP), "thereby likely throwing some suppliers into a distress situation and leading to a sharper cut in spending at some point."

Capacity increases were negligible for Toshiba/SanDisk. However, Toshiba announced on March 18, 2016 in its FY2016 Business Plan that it was restructuring to enhance cost competitiveness by accelerating development of BiCS and expand SSD business by:

• Investing in equipment for the 3D process

• Completing the New Fab 2 Building (1Q/FY2016)

• Developing a site for new production facility (FY2016)

• Constructing a new production facility (FY2017)

Toshiba plans to invest around ¥360 billion ($3.233 billion) in its Fab 2 in FY2016 to FY2018. BiCS production at Fab 2 is set for the first half of FY2016 (from April to October 2016). NAND production costs are going to be reduced through the use of Canon's (NYSE:CAJ) Nano Imprint Lithography, which I plan to expand on in an upcoming Seeking Alpha article this week entitled "The Many Faces of Lithography."

Toshiba's Fab 5 accounted for more than 60% of the manufacturer's total NAND Flash output in the fourth quarter of 2015.

Intel (NASDAQ:INTC) and Micron co-designed the 3-D NAND chips. The two companies have a joint development program for developing non-volatile memory technologies and continue to jointly invest in the IM Flash Technologies joint venture, which has a 300mm non-volatile memory fab in Lehi, Utah.

Micron's newest fab in Singapore, 10X, is set to start mass production of 3D-NAND Flash in the second half of 2016, and shipments, according to Table 3, are expected to increase YoY 16.1% in 2016 and 22.2% in 2017. Product focus are USBs and client SSDs.

Intel is converting its Dalian fab in China from logic IC to 3D-NAND production, with small batch runs expected in Q4 2016. The company will begin testing its 3D XPoint in Q4 2016.

SK Hynix' M14 fab is in the second phase of expansion and will be in operation in the second half of 2016 to meet 3D-NAND Flash demand. TLC flash, the cheapest form of flash to manufacture, represented 40% of NAND flash output in Q4 2015.

Where will NAND chips go?

According to Table 1, growth in smartphones/tablets have moderated and USB/flash demand has plummeted due to the drop in the PC market and to a growing shift of storage workloads to the cloud. SSD will be the biggest draw on NAND demand, increasing 41.6% in 2016 and 52.2% in 2017.

HDD shipments dropped to 17% in 2015 to 469 million units and will drop below 400 million units in 2019, according to the report "The Hard Disk Drive (HDD) And Solid State Drive (SSD) Industries: Market Analysis and Processing Trends," published by The Information Network.

According to the report:

HDD shipments peaked at 651 million units in 2010, and shipments in 2015 were 72% of maximum. In 2019, shipments of under 400 million units will represent 60% of the maximum shipments from 2010.

The supply chain to the HDD manufacturers is also under pressure. For example, glass disk leader Hoya, with a capacity of 305 million disks, is operating at just above 60% capacity. Total glass disk capacity is 800 million disks, but only 360 million disks were incorporated in 2.5-inch HDDs in 2015.

In the HDD sector, all three manufacturers - Western Digital, Seagate, and Toshiba saw shipments drop in 2015 at least 18% compared to 2014. Western Digital (NASDAQ:WDC) saw its lead over Seagate (NASDAQ:STX) decrease to a share of 43.6% in 2015 from 44.2% in 2014. Demand for storage is growing and HDDs are being replaced by SSDs. As prices of SSDs drop, the substitution rate of HDDs by SSDs will accelerate."

While weak demand for PCs, tablets, and smartphones is impacting the semiconductor industry, one of the bright lights in the NAND industry is the increased content per system each year, as shown below in Table 4. But it is a curse as well as a blessing. The high demand is causing a potential supply shortage, necessitating the need for manufacturers to spend billions of dollars, not only in capacity builds but in research and development expenses to stay ahead of the curve and competition. Micron's new Singapore fab cost $4 billion, yes billion, yet that pales in comparison to Samsung's $14 billion Xi'an.

Table 4 - Handset Shipments and NAND Content

2014

2015

2016

2017

CAGR

Handset Shipments

(Millions of Units)

Mobile

2144

2220

2323

2361

3.3%

Feature phone

656

561

485

414

-14.2%

Smartphone

1265

1430

1607

1701

10.4%

Tablet

223

230

231

246

3.3%

SSD

94

124

142

173

22.5%

Flash Cards

660

613

558

499

-8.9%

USB/flash

312

305

295

294

-2.0%

NAND Content

(Gb/system)

Mobile

7.5

10.2

14.4

19.2

36.8%

Feature phone

0.5

0.7

1

1.1

30.1%

Smart Phone

10.4

13.7

18.7

24.3

32.7%

Tablet

11

11.8

12.8

14

8.4%

SSD

246

292.5

362.7

453.3

22.6%

Flash Cards

14.6

18.5

23

27.6

23.6%

USB/flash

15.1

17.1

19.4

21.7

12.8%

Conclusion

The NAND industry is predicted to go into a period of undersupply, probably by the end of 2016. Much of the problem is attributed to the lack of expansion in NAND facilities over the past few years, but an undercurrent is probably due to the large amount of capex and development costs that the NAND suppliers have spent on 3D NAND.

Another factor is the increasing incorporation of NAND per system, so that even though there has been a flattening in demand for systems in the past 1-2 years, NAND content has increased.

This undersupply should help manufacturers by increasing average selling prices of chips, based on traditional supply-demand factors. Smart companies will reinvest the money into capacity expansion. The length of the undersupply period will also be affected by the ramp speed in 3D production by these manufacturers.

Investors need to be cautious because a shortage will result in NAND price increases but will also result in diminished sales. In early 2013, a drop in NAND memory prices was hated because of production cutbacks by Toshiba as well as the underlying upward trend in demand for NAND memory.

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.