Following a significant de-risking event in US chip stocks on June 5, the market witnessed a rapid sector rotation by June 8 and June 9. Micron rebounded nearly 10% on June 8, while SK Hynix and Samsung Electronics posted noticeable gains in the Korean market on June 9.

This shift marks a transition in investor focus from speculative AI narratives to assets with immediate earnings per share (EPS) upgrade potential, driven by DRAM/NAND price hikes and robust export data. The core assets involved include Micron, SK Hynix, Samsung Electronics, Western Digital, SanDisk, NVIDIA, Broadcom, Marvell, Coherent, Credo, and semiconductor ETFs SOXX and SMH.

The initial plunge on June 5 erased over $1 trillion in market capitalization for US-listed chip stocks, with the Philadelphia Semiconductor Index dropping nearly 8.5% intraday. Individual equities suffered sharp declines: Micron fell approximately 13.25%, NVIDIA by about 6.2%, AMD by roughly 10.86%, and Broadcom by around 7.92%.

However, the subsequent recovery was not uniform. Funds did not exit the AI semiconductor sector entirely but rather engaged in internal rotation. As valuations faced scrutiny, market attention pivoted from 'who has the AI story' to 'who can most quickly turn AI demand into profit.' Data compiled by Woofun AI shows that while parts of the AI hardware chain trade on future product cycles and expected capital expenditure, storage demand growth is already reflected in orders, prices, and financial reports.

A primary trigger for this derisking cycle was a gap in expectations following Broadcom's earnings report. In absolute terms, Broadcom's fundamentals remain strong, with FY2026 Q2 revenue reaching $22.2 billion, a 48% year-on-year increase. The company projects FY2026 Q3 total revenue around $29.4 billion, with AI semiconductor revenue expected to hit $16 billion, representing over 200% year-on-year growth. Despite this, the market sold off, indicating that high expectations built over the past year have raised the pricing threshold. Simply being part of the AI supply chain is no longer sufficient; growth trajectories, profit realization, and next-quarter guidance must align with valuations. Woofun AI notes that the June 5 drop was not a demand collapse stress test but a pressure test for high-expectation trades where the narrative alone could no longer sustain premiums.

The market's valuation logic has shifted from proximity to AI capital expenditure to the speed of earnings translation. Previously, GPUs, ASICs, high-speed optical modules, and copper interconnects commanded premium valuations based on their role in AI cluster expansion. Now, the focus is on whether orders translate into EPS. Storage holds a distinct advantage due to a shorter EPS propagation chain. AI server demand alters the supply-demand relationship for high-value-added products like HBM, server DRAM, and enterprise SSDs. As manufacturers shift capacity to these high-end products, traditional DRAM and NAND supply tightens, driving contract prices up. This chain translates quickly into revenue, gross margin, and EPS without relying on long-term imagination.

Financial reports from key players validate this logic. Micron reported record revenues, gross margins, EPS, and free cash flow for FY2026 Q2, with significant year-on-year growth in data center revenue and a forecast for another high in FY2026 Q3. SK Hynix's 1Q26 results were even more pronounced, with revenue of 52.5763 trillion Korean won and an operating profit of 37.6103 trillion Korean won, yielding an operating profit margin of 72%. The company attributed this growth to HBM, high-capacity server DRAM, and eSSDs. Industry pricing data from TrendForce supports this trajectory, expecting a 58% to 63% quarter-on-quarter increase in 2Q26 conventional DRAM contract prices and a 70% to 75% increase in NAND Flash contract prices.

Furthermore, South Korean exports hit a record high in May 2026, with semiconductor exports surging 169.4% year-on-year to approximately $37.16 billion, with chips accounting for over 40% of total exports for the first time.

While NVIDIA remains the ultimate gatekeeper of AI demand, its valuation is already concentrated on top AI assets, making it susceptible to export controls and expectation gaps. Similarly, the ASIC direction, driven by cloud providers' self-developed chips, faces visibility challenges regarding project onboarding and volume production windows. Optical modules and copper interconnects also depend on future architecture roadmaps and capital expenditure cycles. In contrast, storage's pricing basis is straightforward: HBM demand drives high-end products, capacity shifts compress traditional supply, and price increases improve revenue and margins. Woofun AI analysis suggests that while risks remain, storage offers a more easily modeled path to EPS validation compared to future architecture generations.

The rebound from June 8 to 9 does not prove a permanent switch from P/E expansion to EPS validation. Micron's nearly 13% drop and subsequent 10% rebound may include technical repairs and short covering, while SK Hynix's rise was catalyzed by news of data center cooperation with NVIDIA. Storage remains a cyclical industry where rapid price increases could stimulate supply expansion or suppress end-customer purchasing willingness. HBM annual contracts, yield ramp-up, and market share allocation are still evolving. If DRAM/NAND price increases slow or HBM market share falls short, the logic for upward EPS revision will be challenged.

However, the current trend indicates that funds within the AI chain prefer sections with a shorter path to EPS realization, where orders, prices, capacity, and profit margins are simultaneously visible.