Version 9

What

Why it matters

Open questions

• Micron's 16 Strategic Customer Agreements run five years and total $100B [12][13], but do they cover HBM specifically, and do contract prices hold if CXMT's HBM3 production reaches market by end of 2026 [33]?

• The market is still pricing Micron as a cyclical peak despite the multi-year agreement structure [20] — does the sell-side consensus on Q3 memory pricing incorporate the contract floor, or are analysts treating spot and contract markets independently?

• Micron management forecasts AI-driven memory shortages beyond 2027 [14][15], but CXMT is targeting HBM3 production by end of 2026 [33] — has Micron acknowledged this supply timeline anywhere in its earnings disclosures?

• Chanos argues data center builders are structurally uninvestable regardless of backlog [26]; Milk Road AI presents CoreWeave's $99.4B contracted backlog and Nvidia's $2B co-investment as the counter [10] — does Chanos's thesis engage with locked-in contracted revenue specifically?

Narrative

Timeline

• 2026-05-26: Micron crosses the $1 trillion market cap threshold, driven by HBM demand from AI workloads. [41]
• 2026-06-08: Amazon announces a multibillion-dollar optical fiber supply agreement with Corning; Milk Road AI publishes 'Save This' on Corning as an underappreciated AI infrastructure beneficiary. [1][42][43][2]
• 2026-06-08: Jensen Huang describes Qualcomm positively; Milk Road AI publishes 'Save This' on Qualcomm as an on-device inference play. [3][44][45]
• 2026-06-14: Milk Road AI publishes 'Save This' on Micron, predicting a $4,000 stock price and framing Micron as America's strategic monopoly on AI memory. [4]
• 2026-06-15: Convequity argues Chinese memory capacity expansion — not demand collapse — is the most probable signal that the current memory cycle has peaked. [35]
• 2026-06-15: CXMT aligns with Huawei on domestic HBM3 production targeting end of 2026 while phasing out DDR4 for DDR5; analysts warn the expansion risks global price pressure. [32][33][34][46]
• 2026-06-16: NVIDIA commits $2 billion to Coherent's Texas indium phosphide optical facility, treating domestic optical semiconductor manufacturing as critical AI infrastructure. [37]
• 2026-06-17: Milk Road AI cites Deutsche Bank and Gartner forecasts showing DRAM demand outrunning supply, arguing memory stocks are in early stages of their run. [22]
• 2026-06-21: Perplexity CEO Aravind Srinivas publicly states Micron could surpass Meta in market value and calls that prediction conservative. [24]
• 2026-06-22: Milk Road AI cites a Jefferies expert call that came in materially more bullish than the Q3 consensus of 15–20% quarterly memory price gains. [23]
• 2026-06-23: Jim Chanos argues the bull case for AI-adjacent energy stocks rests on a temporary power constraint; when it resolves, elevated valuations will collapse. [25]
• 2026-06-23: Milk Road AI posts a $3,000 Micron price target citing Jensen Huang's argument that AI memory demand is structural rather than cyclical, and adds AAOI following a 14% single-day selloff. [5][7]
• 2026-06-24: Micron reports Q3 FY2026 earnings; CEO Sanjay Mehrotra states memory is 'no longer a commodity'; 16 Strategic Customer Agreements totaling $100B over five years are disclosed; company forecasts AI-driven shortages beyond 2027. [11][36][14][15][13][12]
• 2026-06-24: Milk Road AI argues Micron's Strategic Customer Agreements are more consequential than the earnings print, marking the end of Micron's 47-year commodity business model. [21]
• 2026-06-24: Milk Road AI adds GE Vernova to the 'Save This' series, framing its turbine capacity as a structural AI infrastructure asset because data centers require 24/7 baseload power the grid cannot supply fast enough. [8]
• 2026-06-25: Milk Road AI extends its component-over-assembler thesis to humanoid robotics, arguing component suppliers — not robot assemblers — are the better long-term bet in a market projected to reach $7.5 trillion by 2050. [9]
• 2026-06-26: Milk Road AI publishes a detailed CoreWeave defense: $2.08B Q1 revenue, $99.4B contracted backlog, $40B in new Q1 commitments, and Nvidia's additional $2B investment alongside a joint 5-gigawatt AI factory expansion. [10]
• 2026-06-26: Chanos widens his counter-thesis across Business Insider, Yahoo Finance, CNBC, and YouTube, arguing investors should own AI models not data center or energy infrastructure builders. [26][27][28][29]
• 2026-06-26: Independent retail and semi-professional voices amplify Micron's de-commoditization reading post-earnings, while noting the market continues to price Micron as a cyclical peak. [17][18][19][39][40][20]

Perspectives

Tensions

• Micron CEO Mehrotra and Milk Road AI argue 16 five-year Strategic Customer Agreements totaling $100B end cyclical memory pricing [11][12][21]; Convequity and CXMT-Huawei HBM3 alignment suggest Chinese capacity expansion will reset memory pricing regardless of multi-year contracts [35][33][34]. [11][12][21][35][33][34]
• Milk Road AI calls Micron 'America's monopoly' on AI memory [4]; industry data places SK Hynix at ~62% HBM market share with Micron competing against Samsung for second position [30][31]. [4][30][31]
• Chanos argues investors should own AI models rather than data center or infrastructure builders, and that the power constraint is temporary [26][25]; Milk Road AI's entire series is premised on the opposite — physical infrastructure scarcity outperforms model-layer exposure, and GE Vernova's turbine capacity is a durable structural constraint [8]. [26][25][8]
• Milk Road AI presents CoreWeave's $99.4B contracted backlog and Nvidia's $2B co-investment as proof the market misprices CoreWeave by fixating on debt [10]; Chanos argues data center builders are structurally uninvestable regardless of backlog [26]. [10][26]
• Micron forecasts AI-driven memory shortages extending beyond 2027 [14][15]; CXMT is targeting HBM3 production by end of 2026 [33] — a timeline Micron's management has not acknowledged in its shortage forecast. [14][15][33]
• Independent retail voices and Milk Road AI read Micron's five-year contracts as proof the cyclical model is broken [17][18][21]; the market continues to price Micron as a cyclical peak [20], suggesting the contract structure has not yet changed institutional positioning. [17][18][21][20]