Samsung and SK Hynix unveiled a combined investment plan that could top 1,000 trillion won, or roughly $650 billion, over the next decade, backed by President Lee Jae Myung’s push for balanced regional growth across the country.
Despite the staggering headline figure, both stocks fell, Samsung dropped 5.3%, and SK Hynix slid 3.4%, in a reaction that says as much about investor anxiety as it does about the spending plan itself.
Complicating matters further, the same two companies are now defendants in a new lawsuit filed in California accusing them, alongside Micron, of deliberately engineering the global memory chip shortage that has sent DRAM prices soaring over the past several years.
The $650 Billion Breakdown
According to details shared by BullTheory, the bulk of the new investment centers on Samsung and SK Hynix committing 800 trillion won, or roughly $518 billion, toward building four new semiconductor fabrication plants. Alongside that, 81 trillion won, around $52.45 billion, is earmarked specifically for high-bandwidth memory packaging facilities, the kind of advanced memory infrastructure that AI chipmakers increasingly depend on. A further 30 trillion won, approximately $19.42 billion, is allocated over 15 years toward next-generation memory technology.
The broader ambition behind these numbers is significant. Seoul is targeting a doubling of the country’s DRAM production capacity within five years, and the plan includes 18.4 gigawatts of AI data center capacity by 2035, with an initial 8.4 gigawatts targeted by 2029 through a coalition involving SK Group, GS Group, and Naver. Samsung Chairman Lee Jae-yong framed the urgency plainly, stating that current capacity simply cannot meet existing demand, with additional investment dollars also flowing into robotics, biotech, batteries, and chip substrates across multiple regions of the country.
Why Investors Sold the News Anyway
The market’s skepticism comes down to timing and risk absorption. A single semiconductor fabrication plant alone costs at least 60 trillion won, or about $40 billion, and takes years to construct from groundbreaking to operational output. That means the actual supply increase this plan promises won’t materialize for years, even as investors are left absorbing the financing costs immediately.
That mismatch between long-term payoff and near-term cost is part of why the Korea Exchange pulled a planned options launch tied to these stocks. The market appears to be pricing in execution and funding risk well ahead of the headline investment number, a concern made sharper by the fact that South Korea’s broader market was already down roughly 10% over the past week amid wider AI valuation fears across global markets.
A Lawsuit Accuses the Big Three of Coordinated Supply Cuts
While Seoul was unveiling its spending ambitions, a separate and considerably more pointed story was unfolding in a California courtroom. A lawsuit filed June 25 accuses Samsung, SK Hynix, and Micron of using their strategic pivot toward AI memory chips as cover to deliberately cut production of standard DRAM, the everyday memory found in laptops and phones that most consumers never think twice about.
The numbers cited in the filing are striking. DRAM prices have climbed roughly 500% to 700% over the past four years. The lawsuit also points to Micron’s decision to shut down Crucial, its consumer DRAM brand, at what the filing describes as the most profitable price point in the brand’s history, a move the lawsuit characterizes as economically irrational unless it reflects deliberate coordination between the companies rather than independent business decisions. The complaint draws a direct line to consumer harm as well, citing Apple’s recent price increases on iPads and Macs as evidence that the financial damage from rising memory costs is already reaching ordinary buyers.
A Cartel With a Documented History
What makes this lawsuit particularly difficult for the defendants to wave away is that this isn’t the first time these companies have faced this exact accusation. Between 1998 and 2002, Samsung, Hynix, Micron, Infineon, and Elpida operated what U.S. federal prosecutors confirmed was an actual price-fixing cartel. Samsung ultimately paid a $300 million criminal fine, Hynix paid $185 million, and Infineon paid $160 million, with several executives serving real prison sentences ranging from four to fourteen months.
The new lawsuit goes a step further, alleging that Samsung and SK Hynix later rehired and promoted some of the same executives who were convicted in that earlier price-fixing case into senior roles within their organizations. Combined, the three companies control the overwhelming majority of global DRAM supply today, and building a single new DRAM factory from scratch costs between $15 billion and $20 billion while taking years to come online, a barrier to entry steep enough to make it nearly impossible for new competitors to break into the market and undercut existing pricing. That structural reality sits at the center of what the lawsuit is targeting: three companies with effective control over a market the entire world depends on, already convicted once before, now facing the same accusation again while prices keep climbing and buyers have nowhere else to turn.
No Relief in Sight, According to Jefferies
Wall Street isn’t projecting any near-term reprieve either. Jefferies forecasts DRAM prices climbing another 40% to 50% next quarter, followed by a further 30% to 40% increase the quarter after that, a trajectory that could see prices roughly double by the end of the year. The firm projects an additional 40% to 45% increase in 2027, with no real normalization in the memory chip market expected until 2028.
The Bigger Warning From Central Bankers
Zooming out from the chip-specific drama, central bankers are now openly worried that the broader AI investment boom could be seeding the next major financial shock. Rohan Paul highlighted one of the sharpest warnings yet, issued by the Bank for International Settlements, regarding the debt accumulating behind the AI buildout. The institution’s core concern isn’t AI technology itself, it’s the leveraged supply chain being constructed around revenue streams that haven’t yet proven durable.
The mechanics of that risk are specific. AI demand has pushed hyperscalers to spend heavily on chips, data centers, and power capacity, and that spending has propped up growth, trade flows, and easy financial conditions while equity investors price in years of sustained high earnings growth. Debt has reshaped the structure of this boom in a meaningful way: hyperscaler bond issuance topped $100 billion in 2025 alone, while off-balance-sheet vehicles have shifted data-center obligations toward private credit funds, insurers, and other non-bank lenders. Circular financing compounds the risk further, since chipmakers, hyperscalers, AI labs, and compute providers can fund each other while simultaneously booking future sales from one another, a dynamic that makes genuine underlying demand much harder to read clearly from the outside.
If a capital expenditure slowdown materializes, the BIS warns it could hit suppliers first, then spread into credit markets, then eventually reach households, a real concern given that U.S. stocks make up roughly 64% of the MSCI Global index, with household equity exposure currently higher than in past economic cycles. Private credit adds another layer of systemic risk on top of that, with direct lenders having quadrupled their AI and IT exposure over the past five years to roughly 15% of total portfolios, even as some retail-facing funds already face redemption pressure. The BIS is careful to note that AI can still deliver genuine, lasting productivity gains. The concern is narrower and more specific: the financing structure built around the technology now assumes those gains arrive fast enough to support an enormous and rapidly growing base of fixed costs.
Disclosure: This is not trading or investment advice. Always do your research before buying any cryptocurrency or investing in any services. Follow us on X @nulltxnews
Be the first to comment