SK hynix
The author is an analyst of NH Investment & Securities. He can be reached at hwdoh@nhqv.com -- Ed.
SK Hynix’s 1Q23 results came in better than feared. The memory chip market should improve in 2Q23, with prices set to increase from 2H23. The company’s technological prowess for HBM should provide additional momentum.
1Q23 results proved better than feared
We reiterate our Buy rating for SK Hynix, with a TP of W117,000. The firm recorded 1Q23 sales of W5.09tn (-34% q-q), an operating loss of W3.40tn (RR q-q), and a net loss of W2.59tn (RR q-q), with the operating loss coming in smaller than predicted.
We believe that DRAM shipments fell 20% q-q and ASP slipped 19% q-q, and NAND shipments slid 15% q-q and ASP dropped 10% q-q. The declines in shipments and ASP were below the industry averages thanks to the firm’s preemptive output cuts and inventory strategy. In 1Q23, North American hyperscalers’ data center investment continued to drop, and smartphone and PC sales remained sluggish.
Business conditions to improve in 3Q23; positive HBM momentum expected
In 2Q23, SK Hynix’s earnings will likely remain tepid, but show improvement on a q-q basis. We forecast sales of W5.18tn (+2% q-q) and an operating loss of W2.92tn (RR q-q), assuming: 1) DRAM shipment growth of +11% q-q and ASP growth of -7% q-q, and NAND shipment growth of +10% q-q and ASP growth of -8% q-q.
Thanks to clients’ aggressive inventory adjustment since 2Q22, inventories of TV and other IT sets began to decline after peaking in 1Q23. Inventories of smartphone/server-related parts and memory chips should peak in 2Q23.
The fact that the chip industry has recently decided to cut production also bodes well for future supply-demand dynamics. Given that it takes about four months for production cuts to translate into declines in supply, we forecast that a supply reduction will materialize in earnest from 3Q23, with memory price likely to rebound from then, as well.
Growing AI demand should be positive for SK Hynix. The firm’s market share and technological prowess for high-bandwidth memory (HBM), which is used for AI computing server, are known to be superior to its peers. The global HBM market is forecast to expand by 50% or more pa going forward.