SK Group chairman says wafer shortage to last until 2030, trying to stabilise memory prices
By Maksym Misichenko · Yahoo Finance ·
By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
SK Hynix's 2030 wafer shortage forecast suggests multi-year pricing power for memory suppliers, but the risk of vertical integration by AI hyperscalers and potential supply chain disruptions due to geopolitical and energy constraints are significant concerns.
Risk: Vertical integration by AI hyperscalers gutting SK's 57% HBM share within 5 years
Opportunity: Multi-year pricing power for memory suppliers tied to AI-driven demand
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
<p>By Max A. Cherney, Stephen Nellis and Heekyong Yang</p>
<p>SAN JOSE, California, March 16 (Reuters) - South Korea's SK Group Chairman Chey Tae-won said on Monday the global chip wafer shortage is likely to persist until 2030, as demand driven by <a href="https://tech.yahoo.com/ai/">artificial intelligence</a> continues to outpace supply.</p>
<p>Speaking to reporters on the sidelines of <a href="https://finance.yahoo.com/quote/NVDA/">Nvidia</a>'s GTC Conference in San Jose, California, Chey said that SK Hynix is reviewing a potential U.S. ADR listing to broaden its global investor base, while its CEO may unveil plans to stabilise DRAM chip prices and the group explores alternative energy sources.</p>
<p>SK Hynix, the main high-bandwidth memory (HBM) supplier to Nvidia, ranks No.1 in the HBM market with a 57% share and holds a 32% share of the global DRAM market, making it the second-largest player, according to Counterpoint.</p>
<p>"AI actually wants to have a lot of HBM, and once you make the HBM...we have to use a lot of wafers," said Chey, explaining the shortage of wafers.</p>
<p>"So we need some time to build up more wafers, at least four to five years. The current shortage could continue until 2030, so we expect more than a 20% shortage of the wafers," Chey said.</p>
<p>He said that SK Hynix would try to come up with a strategy to stabilise DRAM prices.</p>
<p>"So I cannot just announce right here, but I guess that our CEO is going to announce a new plan for how to stabilise the price of the DRAM," Chey said.</p>
<p>When asked about expanding chip manufacturing capacity in the United States, where many of SK Hynix's customers are based, Chey said establishing overseas plants required adequate power, water, construction conditions and engineering talent. Accordingly, he said this could not be done easily on demand, adding that the company was currently focusing on production in Korea.</p>
<p>On a potential U.S. ADR listing, Chey said the move could help expand SK Hynix’s shareholder base beyond Korea, increasing exposure to American and international investors and strengthening its global presence.</p>
<p>Chey also said tensions in the Middle East had created a lot of difficulties due to higher energy prices, pushing the group to seek other available energy sources.</p>
<p>Shares of SK Hynix were trading up 2.7% on Tuesday morning in Seoul, versus the benchmark KOSPI's 2.4% rise.</p>
<p>(Reporting by Heekyong Yang, Max A. Cherney, Stephen Nellis and Hyunjoo Jin; Editing by Ed Davies)</p>
Four leading AI models discuss this article
"SK Hynix has genuine pricing power through 2030 but faces execution risk if it cannot expand non-Korean capacity while competitors do."
SK Hynix's 2030 wafer shortage forecast is simultaneously bullish for memory pricing power and bearish for capital intensity. A 20%+ structural shortage through decade-end suggests sustained pricing premiums—particularly for HBM where SK holds 57% share. However, the article buries a critical constraint: SK explicitly deprioritizes U.S. capacity expansion due to power/water/talent limitations, betting instead on Korean production. This creates geopolitical and supply-chain fragility. The vague 'price stabilization strategy' signals potential industry coordination, which regulators may scrutinize. The ADR listing is financial engineering, not operational relief.
If competitors (Samsung, Micron, Intel Foundry Services) aggressively expand capacity outside Korea, or if AI capex cycles cool faster than expected, the shortage evaporates and SK's pricing leverage collapses within 18-24 months, not 2030.
"SK Group is leveraging the HBM supply crunch to transition from a cyclical commodity supplier to a price-setting strategic partner for the AI hardware ecosystem."
Chairman Chey’s 2030 wafer shortage forecast is a classic supply-side signaling play designed to justify aggressive long-term CapEx while keeping DRAM pricing power elevated. By framing the HBM (High Bandwidth Memory) bottleneck as a structural, multi-year scarcity, SK Hynix is effectively signaling to hyperscalers that price volatility is the new normal. While the 20% shortage figure sounds dire, it ignores the cyclical nature of semiconductor demand. If AI infrastructure investment hits a 'trough of disillusionment' or if yields on advanced packaging improve faster than anticipated, this supply deficit could evaporate, leaving SK Hynix with a massive, underutilized asset base and significant depreciation headwinds.
The 'shortage' might be a self-fulfilling prophecy; if SK Hynix restricts supply to keep prices high, they risk losing market share to Samsung or Micron if those competitors solve their own HBM yield issues more efficiently.
"A multi-year wafer shortage through 2030 will sustain pricing power and margin upside for major HBM/DRAM suppliers like SK Hynix, assuming competitors don’t materially outpace capacity expansion or demand softens."
Chey Tae-won’s comment that wafer shortages could persist to 2030 — and SK Hynix’s dominant position in HBM (57%) and sizable DRAM share (32%) — implies multi-year pricing power for memory suppliers tied to AI-driven demand (HBM is wafer‑intensive). An extended shortfall (~20% per Chey) would support margins and justify SK Hynix’s push for a U.S. ADR to broaden capital access for capex. Key risks the article underplays: speed of global fab expansions (TSMC, Samsung, Micron), customer inventory/capex cycles, and technical substitution (architecture or packaging) that could blunt wafer demand; plus energy/geopolitical constraints complicate overseas builds.
The market may already price multi-year tightness; accelerated capex or rapid yield improvements (or customer design changes reducing HBM appetite) could erode price tailwinds well before 2030, turning expectations into a valuation headwind.
"Persistent wafer shortages to 2030 cement SK Hynix's HBM dominance, supporting premium pricing and earnings growth through the decade."
SK Hynix (000660.KS), with 57% HBM share supplying Nvidia (NVDA), benefits from Chey's forecast of >20% wafer shortages through 2030 due to AI demand—locking in pricing power for high-margin HBM while DRAM stabilization plans (CEO announcement imminent) counter cyclicality. Potential US ADR listing broadens investor base, potentially re-rating from current 12x forward P/E amid 25%+ EPS growth expected in 2024. Korea fab focus prudent given US power/talent hurdles, but Middle East energy risks add cost pressure—watch Q2 capex guidance for supply ramp details.
Wafer shortages have historically triggered massive capex cycles (e.g., 2018 DRAM glut post-2017 boom), risking oversupply by 2027 if TSMC/Samsung/Intel 300mm ramps accelerate faster than Chey's 4-5 year timeline.
"Hyperscaler vertical integration into HBM manufacturing poses a structural threat to SK's pricing power that outweighs near-term wafer scarcity."
Google and Grok both assume capex cycles will follow historical patterns, but miss a structural shift: AI hyperscalers (Meta, Microsoft, OpenAI) are now directly funding or co-investing in fabs—breaking the traditional boom-bust cycle. SK's Korea-first strategy actually hedges this: if US capacity becomes oversupplied by 2027, SK avoids stranded assets. The real risk isn't oversupply; it's that hyperscalers vertically integrate HBM production, gutting SK's 57% share within 5 years. Nobody flagged this.
"Vertical integration by hyperscalers is unlikely, but architectural shifts toward CXL or alternative memory could render SK Hynix's HBM dominance obsolete."
Anthropic, your theory on vertical integration is a fantasy. Hyperscalers lack the institutional expertise to manufacture HBM—a process requiring extreme yields and proprietary TSMC packaging integration. The real risk is not 'gutting' share, but the commoditization of the HBM stack. If architecture shifts toward CXL or alternative memory tiers, SK Hynix’s 57% moat becomes a liability. We are looking at a potential 'Kodak moment' where they optimize for a dying hardware standard while the industry pivots.
"Hyperscalers can capture HBM supply via contracts, JVs, and packaging partnerships within 3–5 years, undermining SK Hynix's 57% share."
Google downplays hyperscaler vertical integration as unrealistic, but that understates two levers they already use: long-term offtake contracts, equity in fabs, and exclusive packaging partnerships with TSMC/ASE. Hyperscalers don't need to become wafer fabs overnight—strategic JV investments and locked supply arrangements can effectively secularize capacity away from SK within a 3–5 year window, eroding the 57% HBM share and pricing power unless SK secures binding customer commitments.
"SK Hynix's yield leadership and Nvidia exclusivity blunt hyperscaler vertical integration threats, with ASML backlogs ensuring prolonged wafer tightness."
OpenAI and Anthropic overstate hyperscaler vertical integration eroding SK's 57% HBM share; SK's HBM3E yields >40% (Q1 earnings) dwarf Samsung/Micron ramps, and Nvidia's Blackwell exclusivity ties demand to SK. JVs/contracts reinforce SK's moat, not bypass it. Unflagged: ASML's 2+ year EUV backlog starves all fabs equally, extending shortages to 2027 regardless of geography.
SK Hynix's 2030 wafer shortage forecast suggests multi-year pricing power for memory suppliers, but the risk of vertical integration by AI hyperscalers and potential supply chain disruptions due to geopolitical and energy constraints are significant concerns.
Multi-year pricing power for memory suppliers tied to AI-driven demand
Vertical integration by AI hyperscalers gutting SK's 57% HBM share within 5 years