Semiconductor Sell-off: Why Chip Prices Are Rising as Chip Stocks Fall

Semiconductor markets are sending investors two apparently contradictory signals. The price of many memory chips is rising as artificial-intelligence infrastructure absorbs increasingly large volumes of high-performance components. Yet semiconductor shares have suffered a sharp sell-off, led by some of the companies benefiting most from that demand. This is the central tension behind “chipflation”: tighter semiconductor supply and higher component costs occurring alongside falling stock-market valuations.

What triggered the semiconductor sell-off?

The latest decline intensified in South Korea, where Samsung Electronics and SK hynix sit at the heart of the global memory industry.

On Wednesday, 29 July, South Korea’s KOSPI fell 7%, taking its decline for July to 33%. The sell-off spread to US technology shares, with the Nasdaq losing 1.7% and the Philadelphia Semiconductor Index falling 5.3%. The chip index was nearly 30% below its 22 June peak.

Markets remained volatile on Thursday morning. Reuters reported that the wider Asia-Pacific index was down 0.6%, while the KOSPI was heading for a weekly decline of around 15%. South Korean authorities have introduced stabilisation measures after leveraged single-stock exchange-traded funds contributed to unusually severe market moves.

The decline therefore reflects more than a change in the outlook for chip demand. It also involves crowded positioning, leverage and investors reassessing how much future AI growth was already reflected in share prices.

What does chipflation mean?

Chipflation describes sustained increases in semiconductor and electronic-component costs.

The pressure is particularly visible in memory. High-bandwidth memory, or HBM, is essential for moving data rapidly between processors in AI servers. Producing it is complex and consumes more manufacturing capacity than conventional memory.

Micron has explained that growing HBM production restricts the supply available for other DRAM products. The company said in June that DRAM and NAND supply-demand conditions were expected to remain tight beyond 2026, partly because of cleanroom constraints, long construction lead times and the increasing manufacturing requirements of newer HBM generations.

That helps explain why memory prices can rise even when semiconductor shares fall. The physical market is responding to constrained capacity and strong demand. Equity markets, by contrast, are assessing valuations, future investment costs, competition and the eventual return on AI spending.

Record earnings have not eliminated investor anxiety

Samsung’s results, released on Thursday morning, illustrate the divide.

The company reported record quarterly revenue of KRW171.5 trillion and operating profit of KRW89.5 trillion for the three months ending 30 June. Its semiconductor division generated KRW127.5 trillion of revenue and KRW89.2 trillion of operating profit, supported by AI-server demand and rising memory prices.

Samsung nevertheless expects supply constraints to continue. It said demand for server DRAM, enterprise solid-state drives and HBM should accelerate during the second half of 2026, while weaker consumer demand and rising component costs could weigh on mobile devices and personal computers.

SK hynix also posted record quarterly results on Wednesday. Revenue reached KRW79.3 trillion and operating profit KRW60.5 trillion, with higher DRAM and NAND prices and sales of premium AI-memory products supporting profitability.

However, strong current earnings do not automatically justify any valuation. Investors are questioning whether enormous expenditure on AI data centres will produce sufficient long-term returns and whether today’s exceptional margins can be maintained as production expands.

China adds another source of uncertainty

Competition is also becoming a bigger concern.

A report that China had begun manufacturing domestic immersion deep-ultraviolet lithography equipment contributed to a fall of more than 7% in ASML shares earlier this week. The equipment reportedly remains less advanced than ASML’s extreme-ultraviolet systems, but its development could reduce China’s dependence on overseas semiconductor tools over time.

That does not mean established manufacturers are about to lose their technological advantage. It does, however, challenge assumptions that today’s market structure and profit margins will remain unchanged indefinitely.

How could chipflation affect the wider economy?

Higher memory prices can increase production costs for smartphones, computers, servers, vehicles and other connected devices.

Manufacturers then face a choice: raise prices, reduce memory specifications or accept lower margins. Samsung’s mobile division reported an operating loss in the second quarter despite higher year-on-year handset revenue, with elevated component costs weighing on earnings.

For global markets, the effects are uneven. Semiconductor producers may benefit from scarcity, while hardware companies and consumers absorb higher costs. Technology-heavy stock indices are also more exposed to a change in AI sentiment than markets dominated by financial, energy or defensive businesses.

A repricing, not necessarily the end of the AI cycle

The semiconductor sell-off should not automatically be interpreted as evidence that AI demand has collapsed.

Company results continue to show strong demand, higher memory prices and constrained supply. The market is instead debating how much investors should pay for those earnings and whether the AI investment cycle can deliver the returns implied by earlier valuations.

Chipflation and falling semiconductor shares can therefore coexist. One describes the economics of limited manufacturing capacity. The other reflects changing expectations, leverage and risk appetite.

The distinction will be important as investors assess whether the current decline is a temporary valuation reset—or the beginning of a more lasting reassessment of the AI trade.

Sources: Reuters; Samsung Electronics Q2 2026 results; SK hynix Q2 2026 results; Micron Technology fiscal Q3 2026 materials.

Disclaimer: This article is for educational purposes only and should not be considered financial advice. Always conduct your own research before making investment decisions.
MSc Finance graduate from the London School of Economics and Political Science (LSE)
Avatar for Ria Vaghela

Ria V Vaghela is an M&A Associate at RSM UK and an MSc Finance graduate from the London School of Economics and Political Science (LSE). She has worked at Jefferies, Dial Partners, GP Bullhound and 7i Capital prior to RSM UK gaining an extensive experience in finance. She has also worked as an Editor and Content Writer for The Representative Media. Apart from finance, she is interested in reading books on philosophy, self-help and economics, likes to paint and play lawn tennis.

Leave a Reply

Your email address will not be published. Required fields are marked *