Micron Technology has spent the past three weeks giving its shareholders whiplash. Having touched an all-time high above $1,200 in late June, the stock has since shed close to a quarter of its value, dragging the broader memory complex — SanDisk, SK Hynix, and the newly listed SKHY among them — down with it. For a market that only weeks ago was toasting Micron as the cleanest way to play the artificial intelligence build-out, the reversal has been abrupt enough to unsettle even seasoned holders.
Fenestra is not one of them. We entered this drawdown as Micron's highest-conviction position and we intend to remain so. Our 24-month price target of $1,550 is unchanged, and the recent weakness — while sharp — looks to us like a classic sector rotation rather than a crack in the underlying thesis.
What Actually Changed
Three forces converged to produce the sell-off. First, South Korea's SK Hynix completed a roughly $28 billion Nasdaq listing on 10 July, and capital that had been concentrated in Micron rotated toward the newly tradeable competitor. Second, China's CXMT confirmed plans for an $8.5 billion IPO of its own, reviving long-standing fears that Beijing-subsidised DRAM capacity will eventually erode pricing power across the industry. Third, and more idiosyncratically, a well-publicised short position taken against Micron added fuel to a technical unwind that was already under way, with the stock briefly entering bear-market territory relative to its June peak.
None of these three developments, in our assessment, alters the structural case for high-bandwidth memory. SK Hynix's listing is a financing event, not a demand event — it does not add a single wafer of capacity that was not already planned. CXMT's ambitions are real but remain years away from qualifying at the leading edge required for AI accelerators. And short positions, however well timed, are wagers on sentiment, not statements about the durability of sold-out order books.
The Order Book Still Speaks Loudly
What has been somewhat lost in the noise is that Micron's commercial position has, if anything, strengthened. The company has disclosed sixteen non-cancellable long-term agreements worth in excess of $22 billion of committed high-bandwidth memory revenue, effectively selling out capacity through the remainder of this year. It has also deepened its relationship with Anthropic, both as a memory and storage supplier for next-generation AI systems and, notably, as a participant in Anthropic’ s own financing — an arrangement that gives Fenestra's holding an indirect line of sight into the frontier-model buildout, rather than only the chips beneath it.
Analyst Conviction and Long-Term Investment
Sell-side reaction to the pullback has been telling. Rather than trimming targets into weakness, several houses have moved the other way: KeyBanc lifted its price target during the drawdown, and Citi reiterated its Buy rating while flagging a ninety-day catalyst window tied to AI-driven DRAM pricing. Micron has also committed more than $250 billion toward U.S. fabrication capacity through 2035, a scale of reinvestment that only a company confident in multi-year demand visibility would contemplate.
Why the Pullback Supports the GARP Case
Our GARP discipline requires more than a good story — it requires a reasonable price for growth, assessed against PEG, return on invested capital, and free cash flow yield. Micron's pullback has, if anything, made those metrics more attractive, not less; the earnings power behind the HBM super cycle thesis has not been revised away, only the multiple the market is presently willing to pay for it. That is precisely the kind of dislocation our concentrated, ten-position mandate is built to look through rather than flee.
Volatility Is Not a Verdict on the AI Memory Cycle
We would caution readers against reading near-term volatility as a verdict on the AI memory cycle itself. Memory has always been the most cyclical corner of semiconductors, prone to sharp corrections even amid genuine structural tailwinds. What matters over our investment horizon is whether the world needs materially more high-bandwidth memory over the next several years than it can currently produce. On that question, nothing in the past fortnight's news flow has changed our answer. Micron's next scheduled earnings release, on 29 September 2026, will be an important checkpoint, and we will update readers accordingly.
Volatility of this kind is the toll paid for owning the picks-and-shovels of the AI era, not a signal to abandon the position. If the past decade of memory cycles has taught disciplined investors anything, it is that the sharpest corrections in this sector have tended to precede its longest advances.
Request a Portfolio Review
If you are not happy with your portfolio performance or would like a second opinion, please do not hesitate to contact Fenestra for a free review of your portfolio.
William Meyer
079 624 4031
Frequently Asked Questions
Why has Micron Technology’s share price fallen?
The article attributes the decline to three converging forces: capital rotating toward SK Hynix following its Nasdaq listing, concerns surrounding CXMT’s proposed IPO and future DRAM capacity, and a publicised short position that intensified an existing technical unwind.
Has Fenestra changed its investment thesis on Micron?
No. Fenestra states that Micron remains its highest-conviction position and that its 24-month price target of $1,550 is unchanged.
Why does high-bandwidth memory matter to Micron’s investment case?
High-bandwidth memory is required by advanced artificial intelligence accelerators. The article argues that demand for this specialised memory remains structurally strong and that recent market developments have not changed the underlying demand outlook.
What supports Micron’s commercial outlook?
The article highlights sixteen non-cancellable long-term agreements representing more than $22 billion in committed high-bandwidth memory revenue, together with Micron’s relationship with Anthropic and its planned investment in United States fabrication capacity.
Why could the pullback improve Micron’s valuation case?
Fenestra’s GARP approach considers growth alongside valuation measures such as PEG, return on invested capital and free cash flow yield. The article argues that the lower share price has made these metrics more attractive while the underlying earnings thesis remains intact.
What is the next important checkpoint for investors?
The article identifies Micron’s scheduled earnings release on 29 September 2026 as the next important checkpoint.
Important information: Past performance is not indicative of future performance. Investment values are not guaranteed, and investors are subject to normal market risks.
