Live
🔥 OpenBMB/ChatDevGitHub Trending🔥 microsoft/agent-lightningGitHub Trending🔥 apache/supersetGitHub Trending🔥 shanraisshan/claude-code-best-practiceGitHub TrendingA-SelecT: Automatic Timestep Selection for Diffusion Transformer Representation LearningarXivGUIDE: Resolving Domain Bias in GUI Agents through Real-Time Web Video Retrieval and Plug-and-Play AnnotationarXivSommelier: Scalable Open Multi-turn Audio Pre-processing for Full-duplex Speech Language ModelsarXivCANGuard: A Spatio-Temporal CNN-GRU-Attention Hybrid Architecture for Intrusion Detection in In-Vehicle CAN NetworksarXivDesignWeaver: Dimensional Scaffolding for Text-to-Image Product DesignarXivA Lightweight, Transferable, and Self-Adaptive Framework for Intelligent DC Arc-Fault Detection in Photovoltaic SystemsarXivConsistency Amplifies: How Behavioral Variance Shapes Agent AccuracyarXivStabilizing Rubric Integration Training via Decoupled Advantage NormalizationarXivSemi-Automated Knowledge Engineering and Process Mapping for Total Airport ManagementarXivAIRA_2: Overcoming Bottlenecks in AI Research AgentsarXivBeSafe-Bench: Unveiling Behavioral Safety Risks of Situated Agents in Functional EnvironmentsarXiv🔥 OpenBMB/ChatDevGitHub Trending🔥 microsoft/agent-lightningGitHub Trending🔥 apache/supersetGitHub Trending🔥 shanraisshan/claude-code-best-practiceGitHub TrendingA-SelecT: Automatic Timestep Selection for Diffusion Transformer Representation LearningarXivGUIDE: Resolving Domain Bias in GUI Agents through Real-Time Web Video Retrieval and Plug-and-Play AnnotationarXivSommelier: Scalable Open Multi-turn Audio Pre-processing for Full-duplex Speech Language ModelsarXivCANGuard: A Spatio-Temporal CNN-GRU-Attention Hybrid Architecture for Intrusion Detection in In-Vehicle CAN NetworksarXivDesignWeaver: Dimensional Scaffolding for Text-to-Image Product DesignarXivA Lightweight, Transferable, and Self-Adaptive Framework for Intelligent DC Arc-Fault Detection in Photovoltaic SystemsarXivConsistency Amplifies: How Behavioral Variance Shapes Agent AccuracyarXivStabilizing Rubric Integration Training via Decoupled Advantage NormalizationarXivSemi-Automated Knowledge Engineering and Process Mapping for Total Airport ManagementarXivAIRA_2: Overcoming Bottlenecks in AI Research AgentsarXivBeSafe-Bench: Unveiling Behavioral Safety Risks of Situated Agents in Functional EnvironmentsarXiv

Memory chip giant SK hynix could help end ‘RAMmageddon’ with blockbuster US IPO

TechCrunch AIby Kate ParkMarch 27, 20264 min read2 views
Source Quiz

SK hynix’s potential U.S. listing could raise $10-$14 billion to help it build more capacity, encourage others to follow, and end the 'RAMmageddon' memory shortage.

SK hynix, a South Korean memory chip giant already listed on the KOSPI, is laying the groundwork for a potential U.S. listing that could reportedly raise an estimated $10 billion to $14 billion.

The company announced this week that it has confidentially filed a Form F-1 with the listing, targeting the second half of 2026.

But the real question isn’t just how much it can raise: it’s whether a U.S. listing could increase its trading value as one of the most critical players in the AI chip supply chain.

Despite its critical role in high-bandwidth memory (HBM), a key component powering AI systems from companies like Nvidia, the stock has historically traded at a discount to global peers, according to a Seoul-based semiconductor analyst. It’s got a market cap of around $440 billion, but its valuation multiples remain below those of U.S.-listed semiconductor firms, raising questions about whether geography, rather than fundamentals, is partly driving the gap.

The move is widely seen as an effort to increase its valuation to match global peers like Micron.

“SK hynix’s U.S. listing could help close a long-standing valuation gap with global peers. Despite having comparable – or in some areas stronger – production capacity than U.S.-based chipmakers, the Korean company has historically traded at a discount, partly due to its primary listing in Korea,” the analyst told TechCrunch.

The analyst also mentioned structural factors shaping the deal. “SK Square, SK hynix’s largest shareholder, which held 20.07% as of December 2025, is required to maintain a stake of at least 20% under Korea’s holding company rules.”

Techcrunch event

San Francisco, CA | October 13-15, 2026

Based on current share prices, issuing roughly 2% in new shares could raise $10 billion to $14 billion while allowing SK Square to maintain its ownership threshold, the analyst said. (Under Korea’s Fair Trade Act, holding companies must maintain minimum ownership stakes in subsidiaries, at least 20% for listed entities, to retain control.)

There’s precedent. Taiwan Semiconductor Manufacturing Company (TSMC), for example, has seen its U.S.-listed shares trade at a premium to its domestic shares at times, particularly during periods of strong AI-driven demand, suggesting that cross-listing can influence how investors price the same underlying business.

The move is already rippling across the broader Korean chip sector. Following SK hynix’s filing, some investors are now pushing Samsung Electronics to consider a similar U.S. listing. Artisan Partners, a major shareholder, said Friday that a U.S. listing (technically known as an American depositary receipt, or ADR), could help Samsung boost its valuation, too, as well as give U.S. retail investors a chance to buy its stock, according to a Bloomberg report.

A capital push to meet AI-driven demand

SK hynix’s planned ADR listing is also widely seen as a move to secure funding ahead of increased capital spending to meet the rising demand for memory from AI semiconductors.

At its annual general meeting on March 25, SK hynix CEO Noh-Jung Kwak said financial capacity will be key to sustaining growth in the AI era, adding that the company is targeting approximately $75 billion (more than 100 trillion KRW) in net cash to support long-term investments.

Soaring cost for memory, and limited supply, has been one of the bottlenecks slowing AI builds, but also impacting other industries, like consumer gamers. It’s a situation that’s been dubbed “RAMmageddon” and, if nothing in the market changes, is expected to continue until at least 2027, Nature reports.

Time will tell if that doomsday prediction holds up. The tech giants are working on solving RAMmageddon in other ways beyond increased manufacturing. For instance, Google this week introduced a tech called TurboQuant, an ultra-efficient AI memory compression algorithm. It allows AI to become vastly more efficient in using memory.

Nevertheless, the signals indicate that more memory production will be necessary as well. SK hynix is gearing up for a wave of capital-intensive projects. The company plans to invest around $400 billion by 2050 to build a semiconductor cluster in Yongin, South Korea. It is also constructing new facilities in South Korea and Indiana, with planned investments of about $25 billion and $3.3 billion, respectively, underscoring the scale of capital required.

The chipmaker said this week it will acquire advanced extreme ultraviolet (EUV) lithography scanners from ASML by 2027 in a deal worth $7.9 billion, aimed at boosting high-bandwidth memory (HBM) production for AI.

All of this would be supported by a blockbuster U.S. IPO. And that could lead other Korean chip makers to follow.

Kate Park is a reporter at TechCrunch, with a focus on technology, startups and venture capital in Asia. She previously was a financial journalist at Mergermarket covering M&A, private equity and venture capital.

View Bio

Original source

TechCrunch AI

Was this article helpful?

Sign in to highlight and annotate this article

AI
Ask AI about this article
Powered by AI News Hub · full article context loaded
Ready

Conversation starters

Ask anything about this article…

Daily AI Digest

Get the top 5 AI stories delivered to your inbox every morning.

Knowledge Map

Knowledge Map
TopicsEntitiesSource
Memory chip…SK HynixSemiconduct…IPOInvestmentAI ChipTechCrunch …

Connected Articles — Knowledge Graph

This article is connected to other articles through shared AI topics and tags.

Knowledge Graph100 articles · 342 connections
Scroll to zoom · drag to pan · click to open

Discussion

Sign in to join the discussion

No comments yet — be the first to share your thoughts!