Micron (MU) on Solana

Japan's Kioxia Holdings is reportedly exploring a US listing via American Depositary Receipts that could raise at least $10 billion, with Bank of America, Goldman Sachs, and JPMorgan engaged on the offering. The move follows SK Hynix's landmark $26.5 billion Nasdaq listing in July — the largest US share sale by a foreign company — and signals that global memory chipmakers are racing to tap US capital markets on the back of AI-driven memory demand. Kioxia's Tokyo shares have surged roughly 400% this year, pushing its market cap to approximately $183 billion, giving it substantial balance-sheet firepower heading into any potential listing.

For Micron, a US-listed Kioxia would introduce a better-capitalized, more liquid NAND competitor into the same investor pool at a time when MU shares have traded in a narrow range for two months. Analysts note that while AI-related memory fundamentals remain supportive — tight supply and sustained hyperscaler demand — investors are reassessing infrastructure spending trajectories, weighing on both Micron and SanDisk. Retail sentiment on StockTwits is currently bearish for MU. If Kioxia's listing proceeds and captures momentum comparable to SK Hynix's ADR debut, it could draw capital and analyst attention away from Micron even as the two compete across overlapping NAND and SSD product lines.

Micron is set to report fiscal Q4 results on September 30, with analysts expecting a continuation of the memory market upcycle that propelled fiscal Q3 DRAM revenue to $31.3 billion — up from $7.1 billion year-over-year — and NAND revenue to $9.9 billion from $2.2 billion. Average selling prices for DRAM rose roughly 60% sequentially last quarter, and NAND prices climbed around 80%, as AI-driven demand for high-bandwidth memory and data center SSDs kept supply tight. Memory prices hit record highs in August 2026 and are expected to continue rising through at least Q3 2027.

The structural constraint behind the shortage is compounding: HBM production requires approximately three times the wafer capacity of standard DRAM, limiting how quickly manufacturers can shift supply. Competitor SK Hynix has indicated the DRAM market imbalance will worsen through 2027, with equilibrium not expected until 2030 at the earliest. Micron's shares trade at around 6x forward earnings on fiscal 2027 estimates ahead of the print.

DeepSeek's V4.1-Flash model has demonstrated a 75% reduction in HBM memory demand and an 87.5% reduction in SSD storage requirements for KV-cache operations — the key-value attention states that accumulate during inference on large language models. The efficiency gains stem from a sparse-activation architecture in which a 552-billion-parameter model activates only 8–16 billion parameters per operation, dramatically shrinking the memory footprint that has historically driven high-bandwidth memory procurement.

For Micron, which reported cloud memory revenue of $13.77 billion in its most recent quarter at 83% gross margins and shipped more than $1 billion of HBM4, the development is a demand-side variable worth monitoring. The analysis notes that "DeepSeek has not disproved the memory cycle" — efficiency gains have historically coexisted with rising token volumes that sustain total memory demand — but it flags that if DeepSeek-style architectures achieve broader adoption, Micron's HBM pricing power could face pressure at the margin. Sandisk is seen as more directly exposed given that a larger share of its recent revenue gains came from pricing rather than volume.

Unions representing roughly 10,000 of Micron's 15,000 Taiwan employees — based in DRAM and HBM manufacturing hubs in Taoyuan and Taichung — are demanding a one-time bonus equivalent to 83 months of salary for fiscal 2026 and, starting fiscal 2027, a standing arrangement that allocates 15% of operating profit to quarterly worker bonuses. Over 80% of surveyed union members backed strike action, a notable escalation given Taiwan's status as a critical node in Micron's memory production network. Micron countered with an offer of up to 68 months of salary — which it described as the highest payout in company history — and said it remains committed to dialogue through existing channels.

The standoff comes as Micron reports extraordinary financial results driven by AI infrastructure demand: Q3 fiscal 2026 revenue reached $41.46 billion (up from $9.30 billion a year prior), GAAP net income hit $28.24 billion against $1.89 billion previously, and gross margin expanded to 84.6%, with Q4 revenue guidance of approximately $50 billion. The company has committed $43.9 billion to Taiwan manufacturing capacity. Labor disputes at facilities this central to HBM supply carry real operational risk, and the outcome of negotiations could set a precedent for other memory-sector manufacturers facing similar workforce pressure — Samsung narrowly averted a 48,000-worker strike in South Korea through last-minute talks in May.

Micron Technology (MU) has pulled back roughly 19% from its June 2026 peak, trading around $1,017 as of September 8, even after a 304% year-to-date gain through Q2 and a 239% advance in 2025. One analysis argues the drawdown creates a re-entry window for investors with a multi-year horizon, applying a 14x forward earnings multiple to projected 2030 EPS of $90–$120 to arrive at a price target range of $1,300–$1,700 — roughly 70% upside from current levels. The company currently trades at an unusually compressed forward P/E of approximately 6.5x, reflecting the market's tendency to discount memory names on cycle skepticism despite data center revenue reaching $25 billion in fiscal Q3 — 76% of total DRAM revenue.

The bull case rests on sustained demand for high bandwidth memory, server DRAM, and data center SSDs driven by agentic AI, robotics, and edge deployments through 2030, underpinned by 16 multiyear strategic customer agreements with a combined minimum value exceeding $100 billion. Memory supply shortages are expected to persist into 2027–2028, with new manufacturing capacity in Idaho, South Korea, and New York projected to normalize supply around 2030. Risks are significant: the memory industry is inherently cyclical, multiple expansion to software-like levels is unlikely, and the stock has shown it can reprice sharply when sentiment shifts — making volatility tolerance a prerequisite for any long-duration position.

Micron Technology stock is down about 21% from its June highs — a dip that looks tempting on paper, especially with the stock trading at a P/E of 21, below the S&P 500 and Nasdaq-100. But analyst Anthony Di Pizio argues the discount exists for a reason: Wall Street expects earnings to cycle back down, and the current forward P/E of just 6 based on FY2027 estimates reflects how stretched the growth narrative has become relative to its durability. Micron reported FY2026 Q3 revenue of $41.4 billion — a 346% year-over-year increase driven by AI-related HBM memory demand — but Di Pizio's concern is that this peak masks the underlying cyclicality of memory markets.

The bear case rests on three interlocking risks. First, AI infrastructure spending faces monetization pressure: large hyperscalers collectively deployed around $800 billion in capex in 2026, and roughly 60% of businesses have reportedly begun routing workloads to cheaper AI models to cut costs, with companies including Uber, Walmart, and AT&T implementing AI usage caps. Second, every major memory manufacturer is racing to add HBM and DRAM capacity, a dynamic that historically erodes pricing power once supply catches up. Third, state-level data center construction restrictions could further delay demand. Di Pizio offers no explicit pivot point for turning bullish — the implicit threshold is clearer evidence that supply-demand has re-balanced without a steep margin decline, which the current setup does not yet provide.

Micron's Taiwan unit is preparing what the company describes as the highest compensation package in its history, with detailed proposals due in October, in an effort to head off a strike by local workers. Unions representing roughly 10,000 of Micron's 15,000 Taiwan employees secured backing from more than 80% of surveyed members for strike action in August, driven by demands for a one-time bonus equivalent to nearly seven years' pay. The core dispute centers on Micron's Incentive Pay Plan, which unions argue falls short of industry norms given the company's surging profitability.

Taiwan is Micron's largest manufacturing hub, with approximately $43.9 billion invested there, making it central to DRAM and high-bandwidth memory output at a time of elevated AI chip demand. Taiwanese authorities are monitoring the situation and have flagged potential broader supply-chain disruption if a work stoppage occurs. MU shares fell roughly 2.6% on Tuesday following the initial strike reports and slipped another 1% in premarket trading Wednesday; Micron is scheduled to report fiscal Q4 earnings on September 30.

Micron Technology carries roughly a 6x forward price-to-earnings multiple — specifically 5.6x on fiscal 2027 consensus estimates of $155.56 per share — placing it among the three cheapest stocks in the S&P 500 by that measure, behind only a distressed cable operator and a cyclical automaker. The discount is striking given the company's underlying performance: fiscal Q3 revenue hit a record $41.4 billion (up 364% year-over-year), and earnings per share surged 1,368%. On a trailing basis, Micron's P/E of roughly 20x still sits well below both the S&P 500's 25x and the Nasdaq-100's 33x. The core explanation, analysts say, is that the market continues to price Micron as a boom-and-bust commodity supplier rather than a structural AI beneficiary — the same skeptical lens applied to PC and smartphone memory cycles, where peak earnings historically signaled an imminent downturn.

The bull case rests on whether AI infrastructure demand has altered that pattern. Micron has signed multi-year supply agreements with hyperscaler customers that extend generally through 2030, with plans for those contracts to cover roughly half or more of revenue — trading some upside in a peak-pricing environment for significantly greater earnings visibility. Demand from physical AI applications including humanoid robots and autonomous vehicles is also cited as a durable growth driver beyond the current data center build-out. The bear case counters that the largest cloud providers have now collectively spent over $1.1 trillion on AI infrastructure since 2023, with end-users already pushing back on deployment costs — a dynamic that could compress upgrade cycles and reduce memory pricing power. Critics also note that as Micron and rivals expand manufacturing capacity, new supply coming online could recreate the gluts that historically crushed memory margins, making the stock's current cheapness a forward warning rather than an opportunity.