SK Hynix Pulls Chips Lower; Alphabet, Meta Eye AI Spending

The AI trade hit a speed bump on Monday as SK Hynix’s Nasdaq-listed shares shed nearly 8%, pulling memory stocks and chip names broadly lower and reviving valuation concerns that had been quieter since last week’s record debut.

Key points

• SK Hynix US-listed shares fell nearly 8% on Monday as investors booked profits following last week’s blockbuster Nasdaq listing, triggering declines in Seoul and across the broader global semiconductor space.
• Micron (MU) dropped 5.1% in premarket trading to $929, caught in the same memory-sector retreat.
• Alphabet (GOOG) reports Q2 earnings on July 22; shares are down roughly 7% heading in, but analysts preview no letup in AI spending commitments from the company.
• Meta is putting its in-house AI chip into production in September as it targets doubling its compute capacity.
• TSMC will add two advanced chip packaging plants in Chiayi Science Park, Taiwan, according to a government minister.

What’s behind Monday’s chip selloff?

The drop in SK Hynix shares looks like a straightforward case of post-IPO profit-taking rather than any fundamental deterioration. The stock had surged on its debut last week as investors piled into the AI memory story. Monday’s reversal simply unwound some of that enthusiasm, but the ripple effect was wide enough to drag Micron and other semiconductor names lower alongside it.

The broader concern surfacing in coverage is valuation sustainability. AI-linked chip stocks have run hard in 2026, and any stumble in a high-profile name can quickly become a stress test for the whole group. For investors watching Micron specifically, the premarket decline to $929 extends a period of volatility around memory pricing expectations tied directly to AI server demand.

Does Alphabet’s Q2 preview change anything?

Alphabet heads into its July 22 report carrying a 7% share-price deficit for the year, making the print and the forward outlook more consequential than usual. The preview from Seeking Alpha centers on one key question: whether Google will signal any moderation in its AI infrastructure spending pace, or double down again. The expectation, based on available analysis, is the latter. Google Cloud’s growth trajectory and continued AI product integration are the variables most likely to move the stock.

For the sector, Alphabet’s commentary on capex will matter beyond GOOG itself. If the company reaffirms aggressive data center spending, that is a direct read-through for Nvidia, TSMC, and the memory suppliers caught up in today’s selloff.

Meta’s chip push and what doubling compute actually means

Meta’s internal chip timeline got more concrete. The company’s data center chip is headed to production in September, part of a broader plan to double its computing capacity. Meta is also exploring selling excess capacity to outside customers, which would move it closer to the hyperscaler model that Amazon Web Services and Google Cloud have long operated.

The strategic read for investors: Meta reducing its reliance on third-party silicon is a long-term headwind for merchant chip vendors at the margin, though it takes years for custom silicon programs to meaningfully displace Nvidia or AMD at scale. In the near term, Meta’s compute buildout is itself a demand driver for TSMC’s packaging capacity, which is exactly why the Chiayi plant expansion announcement is relevant context. TSMC adding two advanced packaging facilities in Taiwan points to sustained order visibility from hyperscalers, even on a day when the market is selling the sector.

TSMC’s packaging expansion and South Korea’s fiscal tailwind

On the supply side, TSMC’s decision to expand advanced chip packaging in Chiayi underscores that the bottleneck in AI compute has shifted meaningfully toward packaging, not just raw wafer production. CoWoS and similar advanced packaging technologies are required for the HBM memory stacks that go into AI accelerators, so this capacity addition is structurally significant.

Separately, South Korea announced a record 2027 budget exceeding 800 trillion won (roughly $530 billion), with stronger tax revenues from the AI chip industry cited as a key enabler. The government plans to direct spending toward chips, AI data centers, and physical AI technologies. That fiscal backdrop is notable for SK Hynix investors: the domestic policy environment remains firmly supportive of the memory sector, even as the stock cools off from its debut highs.

Nothing published here is investment advice. Always do your own research before making any investment decisions.