The Chip Policy Is Building Fabs That Can't Fix the Shortage

Micron's new fabrication plant in Virginia will produce advanced DRAM for defense, aerospace, automotive, and medical applications. It will not produce a single chip for a smartphone or a laptop. The company made this explicit when it announced the facility.

the most advanced memory technology ever produced in the United States — Micron Technology

The Virginia fab is not an outlier. It is the rule. Tower Semiconductor's $3 billion expansion in Japan, backed by a billion dollars in Japanese government grants, is building capacity for silicon photonics and advanced packaging aimed at AI and data-center demand.

The expansion is designed to support the rapidly growing long-term customer demand, substantially increasing the company's manufacturing capacity and extending its technology leadership. — Tower Semiconductor

Samsung and SK Hynix are investing hundreds of billions of dollars to expand chip capacity in South Korea under a government plan worth at least $880 billion — and all of it is for high-bandwidth memory and high-end DRAM, the grades that feed AI accelerators, not the commodity memory that goes into consumer devices ^[1]. Micron's entire 2026 HBM capacity is sold out under long-term fixed-price contracts; the company can meet only half to two-thirds of customer demand and is actively exiting the consumer market to focus on data center ^[2][3].

We expect tight conditions to persist beyond calendar 2027 as a result of AI-driven demand across all segments coupled with structural supply constraints. — Sanjay Mehrotra

The pattern is consistent across every major allied-hub fab now under construction. Billions of dollars in alignment capital — the policy framework designed to secure semiconductor supply chains within US-allied geographies — are flowing toward exactly the grade of memory that does nothing for the shortage policymakers are being warned about. The warnings are getting louder. In early June, nine US trade groups representing the automotive, retail, and medical-device industries wrote to the Treasury and Commerce secretaries warning that AI data-center demand is causing a memory shortage that now threatens consumer electronics, vehicles, and medical devices ^[4].

an urgent imbalance in the market for memory chips could lead to significant and sustained near-term price increases for American households and disrupt critical U.S. supply chains. — Alliance for Automotive Innovation

Tim Cook put it more vividly. The Apple CEO called the memory crisis a hundred-year flood and said he has never seen anything like it in over forty years ^[5].

I’ve never seen anything like it in any area in over 40 years. — Tim Cook

Apple has already raised hardware prices by as much as fifty percent across Macs, iPads, and the Vision Pro, with Cook calling the increases unavoidable ^[6]. Microsoft says console memory costs have more than doubled, with another doubling expected by fall 2027 ^[5]. Global PC shipments fell between 3.6 and 4.9 percent in the second quarter — the first decline in two years — yet revenue rose, because vendors are pushing through price increases faster than demand is dropping ^[7].

The real story here is the disconnect between units and dollars: shipments are falling, but revenue is climbing because vendors are pushing through price increases faster than demand is dropping. — Jitesh Ubrani

The shortage is now severe enough to override the alignment logic itself. Apple has begun testing memory chips from ChangXin Memory Technologies, a blacklisted Chinese firm, for devices sold in China ^[8]. Tim Cook is personally lobbying the Trump administration for a procurement waiver.

I think we should look at all supply. — Tim Cook

The administration, for its part, has quietly shifted its posture. After briefly blacklisting a set of Chinese tech companies and then withdrawing the list, the White House has moved toward what it calls managed trade focused on non-sensitive goods ^[9].

We want our trade to be more balanced. — Jamieson Greer

The policy channels capital toward AI capacity in friendly geographies. The shortage it cannot relieve forces the market to bypass those same allies — reaching past them to the Chinese suppliers the framework was designed to make optional. The competitive winner in this dislocation is not the aligned hub. It is the vertically integrated incumbent. Samsung makes both the scarce memory chips and the phones that need them. In the first quarter of 2026, the company reported a record operating profit of 57.2 trillion won — a 753 percent increase over the prior year — and nearly all of it came from semiconductors. The semiconductor division alone generated 53.7 trillion won in profit, a forty-nine-fold increase. Over the same period, Samsung's mobile division saw its operating profit fall nearly 35 percent, crushed by the rising cost of the very components its sister division was selling at record margins ^[10].

Our supply falls far short of customer demand. — Samsung Electronics

The company is not immune to the scarcity premium. It raised US prices across its Galaxy phone and tablet lineup in April — the Tab S11 up a hundred dollars, the top-tier Ultra model up two hundred and eighty ^[11]. The advantage is not escaping the squeeze. It is surviving it. That is more than pure-play competitors can say. Global smartphone shipments fell 11 percent in the second quarter to their lowest level since 2013 ^[12]. OnePlus has halted operations in the US and Europe. Xiaomi, Oppo, and Vivo all saw steep declines. Samsung, by contrast, reclaimed the top global smartphone position with a 24 percent share, and Apple held its ground ^[12].

Higher component prices lifted production costs and weighed on consumer demand. — Counterpoint Research

Samsung's vertical integration turns what is an existential threat for pure-play phone makers into a transfer price — a cost that moves from one division's ledger to another's, painful but not fatal. The company has also diversified its memory dominance beyond smartphones and data centers, overtaking Micron as the world's largest automotive memory supplier with a 40 percent market share, locking in long-cycle contracts that outlast any single consumer-electronics cycle ^[13]. And yet the market is not convinced the story holds. Samsung reported a record second-quarter operating profit of 89.4 trillion won — roughly $58.4 billion, a nineteen-fold increase over the prior year — and its shares promptly fell 10 percent, triggering a broader sell-off that pushed the KOSPI index into a bear market, down 20 percent from its June peak ^[1][14]. Morgan Stanley warned that the memory upcycle may be cresting ^[1].

If we delivered a bad quarter, it is evidence there's an AI bubble. If we delivered a great quarter, we are fueling the AI bubble. — Jensen Huang

S&P Global flagged a deeper vulnerability: Korean corporate earnings are dangerously concentrated in semiconductors, while most other sectors are either growing modestly or remain under external pressure ^[15]. Jensen Huang captured the paradox with characteristic bluntness.

Memory is still a cyclical industry – although different this time – and is approaching peak rate of change in (1) pricing YoY, (2) inventory, (3) the earnings revision breadth. — Morgan Stanley

The alignment logic and the market logic are pulling in opposite directions. The policy channels billions toward AI-grade capacity in allied geographies, but the consumer shortage it cannot touch is forcing the market to reach past those allies. Samsung, the vertically integrated incumbent, internalizes the contradiction — its semiconductor division subsidizes its phone division, and it survives a squeeze that kills competitors who must buy chips at market prices. But the market's own verdict on how long the scarcity lasts is not yet settled. The executives insist the shortage is structural: Samsung's Kim Jaejune warns the HBM supply-demand gap will widen further in 2027, Intel's Lip-Bu Tan says relief may not arrive until 2028, and SK Group's Chey Tae-won says capacity will not meet demand until 2030 ^[16]. Investors, pricing the KOSPI into a bear market despite record profits, are betting on something shorter. One of them is wrong.