Asia Pacific Equity Research
27 February 2024
J P M O R G A N
Technology - Semiconductors
Jay Kwon
AC
(82-2) 758-5725
jay.h.kwon@jpmorgan.com
Bloomberg JPMA KWON
J.P. Morgan Securities (Far East) Limited, Seoul
Branch
Harlan Sur
(1-415) 315-6700
harlan.sur@jpmorgan.com
J.P. Morgan Securities LLC
Sangsik Lee
(82-2) 758 5146
sangsik.lee@jpmorgan.com
J.P. Morgan Securities (Far East) Limited, Seoul
Branch
Neelay Y Kamath
(91-22) 6157 3764
neelay.kamath@jpmchase.com
J.P. Morgan India Private Limited
As we have highlighted in our GMM (link) and Korea conference takeaway notes
(link), we see HBM3E competition landscape intensifying with MU announcing
8-Hi HBM3E MP for NVDA’s H200 in 2Q24E (link) and SEC’s sampling of 12-Hi
HBM3E with 1H24 MP guidance (link). We believe both memory makers’
HBM3E timeline is broadly in-line with our expectation and see no material
changes to our current HBM TAM estimates. We note that MU’s exposure to the
overall HBM market is likely to remain limited due to negligible HBM capacity of
~20K wfpm by 4Q24 (vs. SEC/SKH 130K/110K wfpm). In terms of stock
implications, we prefer SK Hynix in the NT given its strong HBM positioning.
•
Competition in HBM3E space intensifying. MU announced volume
production of its 24GB 8-Hi HBM3E and expects it to be shipped in 2Q24E for
NVDA’s H200 GPU (link). The company further announced that its HBM3E
is produced under 1bnm technology node (vs. SEC using 1anm and SKH 1bnm
for HBM3E) and is expected to sample 36GB 12-Hi HBM3E in March-2024.
On the other hand, SEC announced that product development of 36GB 12-Hi
HBM3E is complete and guided to 1H24 MP (link) vs. SK Hynix’s HBM3E MP
schedule in late-1Q24/early-2Q24. We assume SEC’s 12-Hi HBM3E is
currently sampling for NVDA’s high-end GPU (i.e. B100).
•
Recap of HBM market TAM. We expect HBM to drive memory TAM higher
to US$12.8bn, accounting for 15% of the total DRAM market in FY24E
evolving to a 19% contribution in FY25E. In-line with NVDA’s GPU rollout
schedule (i.e. B100/H200), we expect HBM3E TAM contribution to the
overall HBM market to be 42%/58% in FY24/25E. HBM3E carries 30% or
higher price premium vs. HBM3 per bit, so that an increase in HBM3E volume
mix would help expand HBM market TAM and margin profile for memory
makers. Provided tight HBM S-D, we are not concerned about intensifying
competition as a risk to our thesis.
•
Implications for Asian memory. As expected, we see competition
intensifying in the HBM3E space with all memory makers introducing
HBM3E solutions in 1H24E. We estimate that SKH is leading the HBM3E race
with the earliest MP timeline of 8-Hi HBM3E in late-1Q24/early-2Q24 (vs.
MU in 2Q24E and SEC still under qualification). On the other hand, we
cautiously estimate 12-Hi HBM3E MP timeline to be largely similar across
memory makers (JPMe: 2Q24E). In terms of HBM market share, we view
MU’s exposure as being limited, given relatively low HBM capacity (JPMe:
20K wfpm by 4Q24E vs. SEC: 130K wfpm and SKH: 110K wfpm). We expect
Korean memory makers to continue to lead HBM market with strong market
share presence of 48%/44% split between SKH/SEC in FY24E, evolving to
47%/45% in FY25E.
•
Stock implications. Memory shares have been muted YTD (1% vs. SOX:
See page 4 for analyst certification and important disclosures, including non-US analyst disclosures.
J.P. Morgan does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that
the firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single
factor in making their investment decision.
Memory Market Update
Update on HBM3E competition landscape post news of
MU/SEC product launch; focus on HBM TAM growth
potential via mix shift
¸ü¶à×ÊÁϼÓÈë֪ʶÐÇÇò£ºË®Ä¾µ÷ÑмÍÒª ¹Ø×¢¹«ÖںţºË®Ä¾¼ÍÒª
¸ü¶àÒ»ÊÖµ÷ÑмÍÒªºÍÑб¨Êý¾Ý¼ÓV£ºshuimu9870
