From June 23 to June 26, 2026, the global semiconductor market underwent a dramatic narrative collapse and subsequent reconstruction. In just four days, the KOSPI index triggered circuit breakers twice, the Philadelphia Semiconductor Index plunged nearly 8% in a single day, and then Micron’s earnings report—with a staggering 84.9% gross margin—sent its stock soaring almost 16% after hours.
This wasn’t just routine volatility—it was a stress test for the sustainability of AI chip demand. When Broadcom failed to raise its full-year AI chip guidance in its early June earnings, the market began to question the narrative of "perpetually scarce computing power." As the KOSPI dropped from above 9,000 to circuit breaker levels in six trading days, investors began to reprice the supply-demand balance for HBM and AI memory. Then Micron, with $41.46 billion in revenue and an 84.9% gross margin, answered all doubts. The market responded with a 17% rally, signaling that the capital expenditure cycle for AI chips is far from over.
This article breaks down the logic behind the market’s emotional shifts over this 96-hour period, using a timeline of events. We analyze SMH fund flows, individual stock performance, and industry fundamentals to assess the pricing framework for AI chip stocks in the second half of 2026.
Collapse: The Chain Reaction of Expectation Gaps
Broadcom’s "No Upside" Sparks the First Crack
On June 3, 2026, after the market closed, Broadcom released its Q2 earnings: total revenue of $22.19 billion, up 48% year-over-year, and AI semiconductor revenue of $10.8 billion, up 143%. On paper, these are stellar results. But the market cares more about the future than the past. Broadcom’s Q3 AI semiconductor revenue guidance was $16 billion, below the consensus estimate of $17.2 billion. More importantly, management did not raise its full-year AI chip sales forecast of $56 billion, while analysts’ mean estimate hovered around $57.6 billion.
This is a textbook "expectation gap"—the numbers are good, but not better than the "already priced-in best case." After the report, Broadcom shares plunged 13% after hours. This drop wasn’t a rejection of Broadcom’s fundamentals, but rather the first correction to the implicit assumption of "unlimited AI chip demand."
KOSPI Circuit Breaker: The Multiplicative Effect of Leverage and Panic
On June 23, the Korean KOSPI index plummeted 9.99% to 8,203.84, triggering a circuit breaker. Samsung Electronics and SK Hynix both dropped over 12%. The immediate trigger was rumors that SK Hynix might slow HBM4 expansion, but the deeper cause was leverage amplification—Korean financial regulators had previously warned that leveraged funds tied to Samsung and SK Hynix had greatly increased market volatility since launch.
Year-to-date, the KOSPI index had still gained about 90%, making it one of the world’s best-performing stock markets. This meant Korean chip stocks had accumulated substantial unrealized gains before the crash. When expectations cracked even slightly, leveraged capital created a domino effect, turning sell-offs into circuit breakers.
US Chip Stocks Follow: Cross-Market Contagion of Sentiment
On June 24, the sell-off crossed the Pacific. The Philadelphia Semiconductor Index plunged 7.87% in one day, closing at 13,482. SanDisk and Micron fell over 13%, ARM dropped more than 10%, Qualcomm and Western Digital fell 8%, TSMC and Intel lost over 6%, AMD dropped more than 5%, and Nvidia slid 4.15%. The Nasdaq fell 2.21%.
This wave of selling wasn’t triggered by deteriorating fundamentals at any single company, but by multiple pressures: rising expectations for Fed rate hikes, investor skepticism about cloud providers’ AI capex returns, and the contagion of panic from the Korean market. As signals emerged that "compute rental prices are falling from their peak and tech giants are collectively tightening AI budgets," the simple narrative of "scarce compute justifies capex" faced its first systemic challenge.
Rebuilding: One Earnings Report Resets the Narrative
Micron’s "Explosive" Moment
After the close on June 24, Micron released its fiscal 2026 Q3 earnings. The numbers: revenue of $41.46 billion, up 346% year-over-year, far exceeding the $35.84 billion consensus; adjusted EPS of $25.11, beating the $20.78 estimate; and gross margin of 84.9%, up from 74.9% last quarter and 39% a year ago.
An 84.9% gross margin leads all major US tech companies, surpassing Meta’s latest 81.9% and Nvidia’s 75%. This isn’t just "selling more"—it’s the ultimate display of pricing power. With HBM supply still critically short, Nvidia, AMD, and Google all must accept Micron’s prices.
Even more bullish was the guidance: Q4 revenue expected between $49 and $51 billion, EPS between $30 and $32, and gross margin potentially reaching 86%. Micron has signed 16 long-term agreements with data center operators, automakers, and other customers, locking in sales for the next 3 to 5 years. CEO Sanjay Mehrotra stated, "Memory supply tightness will persist beyond 2027."
After the report, Micron surged nearly 16% after hours. On June 25, MU closed at $1,167.88, up 11.5%. Multiple investment banks raised their price targets: Needham from $1,550 to $1,650, B of A from $1,500 to $1,550, Susquehanna from $1,750 to $2,000.
Chain Reaction: From Memory to the Entire Industry
Micron’s earnings not only rescued its own stock, but also reset the narrative for the entire AI chip sector. After the report, US chip stocks rallied across the board after hours: Western Digital jumped over 11%, SanDisk more than 10%, Qualcomm over 10%, Seagate up more than 8%, ARM up more than 5%.
On June 25, the Philadelphia Semiconductor Index soared 482.68 points (3.59%), closing at 13,940.87. Even though the Nasdaq fell 0.46% to 25,358.60, the semiconductor sector had already completed a V-shaped reversal.
SMH Fund Flows: Retail Capital "Voting With Their Feet"
Fund flows offer another perspective on narrative reconstruction. In mid-June, the VanEck Semiconductor ETF (SMH) saw a single-day net inflow of about $6.9 billion, with AUM rising to $78.9 billion. In the second half of June 2026, US semiconductor ETFs saw retail buying of about $12 billion, up 1,200% from April—a record. The Roundhill Memory ETF (DRAM) accumulated nearly $17 billion in AUM within five weeks of launch, the fastest asset growth in ETF history.
Large-scale retail inflows usually signal crowded trades. But from another angle, $12 billion in retail buying also shows that after Micron’s earnings validated sustained AI memory demand, individual investors’ confidence in the "long-term AI chip narrative" didn’t collapse—instead, the sell-off created a buying opportunity.
The Deeper Logic Behind Narrative Shifts
From "Scarce Compute" to "Memory Bottlenecks"
The core narrative shift centers on the market’s focus moving from GPU compute to memory supply. Broadcom’s "no upside" shook faith in unlimited GPU/ASIC demand, but Micron’s earnings proved a tougher constraint—shortages of HBM and DRAM are physical, not demand-driven.
Micron’s 84.9% gross margin demonstrates that, with HBM supply limited, memory manufacturers have stronger pricing power than chip designers. This isn’t "AI demand slowing"—it’s "AI industry value distribution being rearranged," tilting from GPU makers to memory suppliers.
Demand Visibility for 2026-2027
Micron’s 16 long-term agreements provide direct evidence for demand through 2026-2027. Customers are willing to lock in supply at high prices for the next 3-5 years, indicating downstream AI memory demand expectations are strengthening, not weakening. This creates an interesting hedge against Broadcom’s lack of guidance raise, which implied "demand may fall short"—different segments are at different points in the supply-demand cycle.
Conclusion: Collapse Is Over, Volatility Remains
In just four days, the AI chip narrative underwent a thorough stress test. Broadcom’s "no upside" exposed excessive optimism embedded in market pricing. KOSPI’s circuit breaker highlighted the fragility of leveraged capital. Micron’s 84.9% gross margin provided indisputable evidence for the reality and persistence of AI memory demand.
As of June 26, Bitcoin remains in a weak range between $58,000 and $60,000. The Fear & Greed Index stands at 13, deep in extreme fear territory. Risk appetite in both crypto and semiconductor markets has not fully recovered. But SMH’s single-day $6.9 billion inflow and DRAM ETF’s $17 billion asset accumulation in five weeks show that capital hasn’t abandoned the AI chip sector—it’s just waiting for more definitive pricing anchors.
Micron’s earnings have provided that anchor. In the second half of 2026, the pricing logic for AI chip stocks will shift from "imagining unlimited demand" to "validating real supply constraints"—the former depends on imagination, the latter on gross margin and long-term agreements. For investors, this means a more quantifiable valuation framework, and more pronounced divergence among individual stocks.
FAQ
Q1: Why did Broadcom’s earnings beat expectations but its stock still plunge?
Broadcom’s Q2 revenue of $22.19 billion and AI semiconductor revenue of $10.8 billion both beat expectations, but the market cares about incremental growth, not existing numbers. The Q3 AI semiconductor guidance of $16 billion fell short of the expected $17.2 billion, and the full-year sales target remained at $56 billion. With AI chip stock valuations already at elevated levels, "meeting expectations" effectively means "falling short of implied expectations," triggering a sell-off.
Q2: How did the KOSPI crash transmit to US chip stock sell-offs?
The KOSPI’s 10% plunge on June 23 was led by chip giants Samsung and SK Hynix, triggering a circuit breaker. As Korean chip stocks are central to the global semiconductor supply chain, their collapse quickly sparked concerns about global AI chip demand. On June 24, the Philadelphia Semiconductor Index followed with a 7.87% drop, creating a "Korea → global" panic contagion.
Q3: What does Micron’s 84.9% gross margin signify?
An 84.9% gross margin tops all major US tech companies, beating Meta (81.9%) and Nvidia (75%). This means that with HBM and DRAM persistently undersupplied, memory manufacturers have immense pricing power. Whether it’s Nvidia, AMD, or Google’s AI processors, they all rely on Micron’s high-bandwidth memory.
Q4: Does SMH’s single-day $6.9 billion inflow mean the semiconductor sector is overcrowded?
SMH’s single-day net inflow of about $6.9 billion pushed AUM to $78.9 billion. In late June, US semiconductor ETFs saw retail buying of about $12 billion, up 1,200% from April. From a behavioral finance perspective, such concentrated inflows do suggest crowded trades and increased short-term volatility risk. But crowding alone isn’t a sell signal—the key is whether fundamentals continue to justify high valuations.
Q5: How will the investment logic for AI chip stocks change in the second half of 2026?
In the first half, AI chip stock pricing logic was "compute is always scarce," relying on imagination about long-term demand. In the second half, the focus will shift to "validating real supply constraints"—the market will pay more attention to gross margin, long-term agreement coverage, and capacity expansion rates—quantifiable metrics. Different segments (GPU/ASIC vs. HBM/DRAM) will enter distinct supply-demand cycles, and performance among individual stocks will likely diverge further.
