How Major Chip Companies Performed

Asia (Friday, June 26):

• South Korea's KOSPI tumbled more than 8%, triggering a circuit breaker .
• Japan's Nikkei slumped nearly 5% .
• Samsung and SK Hynix were among the heaviest weights, as memory stocks that had surged 41% over the prior month gave back gains .

U.S. (Friday, June 26):

• Sandisk plunged over 10% .
• Micron Technology fell 7% .
• Intel shed 2.6%, Arm lost 3.3%, Marvell dropped 3.5% in premarket trading .
• The S&P 500 fell 0.05% to 7,354, the Nasdaq Composite declined 0.24% to 25,297 — marking the fifth consecutive drop for both indices .

The broader June 23 'chip-wreck' saw Micron plunge 13%, Nvidia shed over $300 billion in market cap at one point, and the Nasdaq drop 4%+ in a single session .

AI Infrastructure Costs and Chip Price Inflation Driving Anxiety

• 'Chipflation': Korean memory makers Samsung and SK Hynix have raised server DRAM prices by 60–70% in 2026 as AI demand outstrips supply . Contract prices for DRAM and NAND have surged dramatically, with AI server demand consuming production capacity that would otherwise go to PCs and smartphones .
• Hyperscaler capex is consuming nearly all operating cash flow. Estimates put Big Tech's combined 2026 AI infrastructure commitments at roughly $725 billion, with Alphabet alone raising $80 billion in debt . This has shifted the narrative from AI growth to capital sustainability.
• Alphabet's $80 billion raise challenged the prior assumption that hyperscalers can self-fund AI buildout indefinitely, introducing dilution and balance-sheet concerns .
• Memory stocks (SK Hynix, Samsung, Micron) have been the primary beneficiaries of 'chipflation,' but the selloff reflects fears that pricing power has peaked and that hyperscalers may push back or slow orders .

Jefferies' Counterargument and 'Capital Destruction' Warning

Jefferies' Global Head of Equity Strategy Chris Wood has published a nuanced view :

The bull case (AI trade still holds):

• Jefferies argues the AI rally remains fundamentally supported by strong earnings growth, not just speculation .
• Their quantitative strategy team found that returns on AI-related stocks are backed by real revenue and profit expansion, unlike the 2021–2022 unprofitable tech bubble .
• AI stocks have driven more than 80% of the S&P 500's gains in 2026, and Jefferies sees this concentration as justified by earnings delivery — for now .

The 'capital destruction' warning:

• Wood warns that the AI trade will eventually be broken not by a sudden drop in chip demand, but by a market-wide realization that hyperscalers and leading AI labs (OpenAI, Anthropic) cannot earn an adequate return on the vast capex they are undertaking .
• He sees a risk that investor patience will run out when it becomes clear that ~$725 billion in annual AI spending is not generating proportionate revenue or profit, potentially triggering a structural repricing of the entire semiconductor complex .
• The concern is that the hyperscaler capex cycle resembles telecom infrastructure build-outs — massive upfront investment with long, uncertain payback periods and the risk of stranded assets .
• Wood's view: the selloff isn't about broken AI demand; it's about the price of that demand being too high relative to the returns it ultimately generates .

Bottom-line tension: Jefferies believes the AI trade is supported in the near term by real earnings momentum, but warns that if hyperscalers fail to monetize their investments within a visible timeframe, the result will be 'capital destruction' across the AI supply chain — starting with the chip stocks that have benefited most from the spending spree.

How Major Chip Companies Performed

Asia (Friday, June 26):

• South Korea's KOSPI tumbled more than 8%, triggering a circuit breaker .
• Japan's Nikkei slumped nearly 5% .
• Samsung and SK Hynix were among the heaviest weights, as memory stocks that had surged 41% over the prior month gave back gains .

U.S. (Friday, June 26):

• Sandisk plunged over 10% .
• Micron Technology fell 7% .
• Intel shed 2.6%, Arm lost 3.3%, Marvell dropped 3.5% in premarket trading .
• The S&P 500 fell 0.05% to 7,354, the Nasdaq Composite declined 0.24% to 25,297 — marking the fifth consecutive drop for both indices .

The broader June 23 'chip-wreck' saw Micron plunge 13%, Nvidia shed over $300 billion in market cap at one point, and the Nasdaq drop 4%+ in a single session .

AI Infrastructure Costs and Chip Price Inflation Driving Anxiety

• 'Chipflation': Korean memory makers Samsung and SK Hynix have raised server DRAM prices by 60–70% in 2026 as AI demand outstrips supply . Contract prices for DRAM and NAND have surged dramatically, with AI server demand consuming production capacity that would otherwise go to PCs and smartphones .
• Hyperscaler capex is consuming nearly all operating cash flow. Estimates put Big Tech's combined 2026 AI infrastructure commitments at roughly $725 billion, with Alphabet alone raising $80 billion in debt . This has shifted the narrative from AI growth to capital sustainability.
• Alphabet's $80 billion raise challenged the prior assumption that hyperscalers can self-fund AI buildout indefinitely, introducing dilution and balance-sheet concerns .
• Memory stocks (SK Hynix, Samsung, Micron) have been the primary beneficiaries of 'chipflation,' but the selloff reflects fears that pricing power has peaked and that hyperscalers may push back or slow orders .

Jefferies' Counterargument and 'Capital Destruction' Warning

Jefferies' Global Head of Equity Strategy Chris Wood has published a nuanced view :

The bull case (AI trade still holds):

• Jefferies argues the AI rally remains fundamentally supported by strong earnings growth, not just speculation .
• Their quantitative strategy team found that returns on AI-related stocks are backed by real revenue and profit expansion, unlike the 2021–2022 unprofitable tech bubble .
• AI stocks have driven more than 80% of the S&P 500's gains in 2026, and Jefferies sees this concentration as justified by earnings delivery — for now .

The 'capital destruction' warning:

• Wood warns that the AI trade will eventually be broken not by a sudden drop in chip demand, but by a market-wide realization that hyperscalers and leading AI labs (OpenAI, Anthropic) cannot earn an adequate return on the vast capex they are undertaking .
• He sees a risk that investor patience will run out when it becomes clear that ~$725 billion in annual AI spending is not generating proportionate revenue or profit, potentially triggering a structural repricing of the entire semiconductor complex .
• The concern is that the hyperscaler capex cycle resembles telecom infrastructure build-outs — massive upfront investment with long, uncertain payback periods and the risk of stranded assets .
• Wood's view: the selloff isn't about broken AI demand; it's about the price of that demand being too high relative to the returns it ultimately generates .

Bottom-line tension: Jefferies believes the AI trade is supported in the near term by real earnings momentum, but warns that if hyperscalers fail to monetize their investments within a visible timeframe, the result will be 'capital destruction' across the AI supply chain — starting with the chip stocks that have benefited most from the spending spree.