Morgan Stanley previews MU & CBRS earnings: Memory price surge exceeds expectations, bullish on AI-driven cycle extension

Overview

This report by Morgan Stanley provides an earnings preview for two key semiconductor industry players - Micron Technology (MU) and Cerebras Systems (CBRS). Despite macroeconomic uncertainties, the core thesis centers on persistent demand exceeding supply for memory chips, particularly DRAM, leading to substantial price increases that are expected to sustain for 2-3 years. The firm has upgraded Micron's profit forecasts and target price while maintaining its 'Overweight' rating. For Cerebras, the firm views it as a new entrant in the AI chip space where short-term focus is on execution capability related to capacity deployment, while long-term prospects hinge on validating its wafer-scale engine technology, also maintaining an 'Overweight' rating.

Core views

Demand-side dynamics and pricing trends: The research highlights that demand for memory chips, especially in the DRAM segment, continues to outpace supply. Micron's prior guidance for May quarter revenue growth of 40% implied a year-over-year increase in average selling prices (ASPs) for DRAM and NAND of 30-35%. However, through channel checks, Morgan Stanley found conditions to be even stronger, revising their model to project a 45% increase in DRAM ASP (up from 40%) and a 50% increase in NAND ASP (up from 35%). While third-party predictions suggest potentially higher growth rates of 60%-75%, the firm believes these are already priced into the market. For the August quarter, they expect another approximately 20% increase in both types of memory prices, with room for further upside. Financial projections and valuation for Micron (MU): Based on more favorable pricing assumptions, Morgan Stanley upgraded their May quarter revenue forecast for Micron to $36.454 billion (a 52.8% increase quarter-on-quarter and a 291.9% year-on-year increase), surpassing the market consensus of $35.555 billion. Their EPS projection for the same period stands at $21.31, above the consensus of $20.57. They anticipate a gross margin of 83.1% for this quarter. Looking ahead to the August quarter, they forecast revenues of $43.328 billion and EPS of $26.01. On valuation, the firm has increased their estimate for Micron's through-cycle earnings to $35 per share and applied a cyclically appropriate P/E ratio of 29.5, arriving at a target price of $1,050. This multiple reflects the premium currently attached to AI-related opportunities within the semiconductor sector, consistent with broader industry levels. Execution focus and long-term rationale for Cerebras (CBRS): With Cerebras preparing to release its first quarterly results post-IPO, Morgan Stanley does not anticipate any major surprises in performance but holds a constructive view towards its differentiated architecture. Given existing large take-or-pay contracts, the short-term narrative pivots around execution rather than demand fundamentals. The firm will closely monitor management's progress in bringing online the initial 250 MW of capacity, with current expectations pointing to completion around mid-2027. In the longer term, the critical test will be whether the company can demonstrate that its wafer-scale engine represents a durable competitive advantage by delivering tangible value to customers and achieving economic scale expansions. Supply chain and capital expenditure considerations: From a supply standpoint, although overall DRAM production growth has exceeded expectations, constraints due to construction timelines, HBM trade-off ratios, and less efficient node migrations limit responsiveness, resulting in single-digit percentage increases in bit shipments that lag significantly behind demand growth. Furthermore, following Micron's acquisition of Tonglu Factory from PSMC, equipment deliveries may occur faster than anticipated, necessitating upward revisions to capital expenditure plans. Morgan Stanley revised its forecast for Micron's August quarter capex from $8 billion to $10 billion and annual spending from $25 billion to $27 billion.

Analysis framework

The report employs a typical 'supply-demand framework' combined with 'volume-price decomposition' methodology to analyze the memory chip industry. Firstly, by comparing Micron's official guidance with channel survey data, it identifies an upside surprise in average selling prices (ASP), leading to updated revenue and profit forecasts. Secondly, it applies a 'cyclical valuation approach,' estimating an average earnings figure across a full business cycle (through-cycle earnings) and combining this with the prevailing market risk appetite towards the AI theme (justifying a higher P/E multiple) to derive a target price. For Cerebras, the analysis adopts an 'execution-driven logic,' distinguishing between the near-term catalyst of planned capacity rollouts and the longer-term validation of technological moat, emphasizing that efficient execution is crucial given pre-existing customer commitments.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply-demand framework

  The central tension in the memory chip industry lies in matching supply with demand, particularly how high-value products like HBM impact wafer capacity requirements.

  By analyzing the gap between bit shipment volumes (supply) and demands from data centers/AI applications (demand), one can infer price trends. When demand grows much faster than supply, prices tend to rise, boosting manufacturer profits.

• Industry/Industrial Analysis FrameworkVolume-price decomposition

  Decomposing revenue growth into separate components attributable to volume changes (bit shipments) and price adjustments (average selling prices).

  The report independently forecasts quarterly bit shipment growth rates (e.g., 5.0% for DRAM and 4.0% for NAND) alongside average selling price increases (like 45% for DRAM and 50% for NAND), allowing for a more precise estimation of revenue and margin shifts without introducing errors from lump-sum growth predictions.

• Valuation MethodPE/PEG valuation

  Applying a price-to-earnings ratio based on through-cycle earnings.

  Given the inherently cyclical nature of the semiconductor industry, using peak or trough figures from individual years can be misleading. Instead, the report calculates an average earnings level over a complete business cycle (here set at $35 per share) and multiplies this by a relevant P/E ratio (29.5 here) reflecting the sector's activity driven by AI factors, thereby smoothing out distortions caused by cyclical fluctuations.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Micron Technology (MU.O)

  Beneficiary: Rising DRAM/NAND prices directly drive substantial revenue and profit growth, with AI-driven storage needs ensuring continued prosperity.

  Strengths

  Leadership in HBM performance with stable market share; strong bargaining position amidst tight supplies; strategic customer agreements providing revenue visibility.

  Weaknesses

  Significant increase in capital expenditures affecting short-term free cash flow; uncertainty surrounding adjustments to LPDDR content.

  Comparison

  Compared to peers, Micron's execution capabilities in HBM are undervalued, and its valuation appears attractive relative to its underlying cyclical earning potential.

  Risks

  Rapid slowdown in end-user demand leading to inventory build-up; intensified competition in HBM segments pressuring prices; macroeconomic uncertainties impacting customer capital spending decisions.

• Cerebras Systems (CBRS.O)

  Beneficiary: Surge in AI computing demand gives its wafer-scale engine architecture a distinctive competitive edge.

  Strengths

  Unique wafer-scale design; secured large take-or-pay contracts offering short-term revenue guarantees.

  Weaknesses

  Unproven scalability and economic viability at mass-market levels; capacity rollout pace acts as a bottleneck.

  Comparison

  Positioned as a newcomer among AI chips, compared to established GPU manufacturers, its architecture shows promise for specific large-model training scenarios despite having limited ecosystem maturity.

  Risks

  Delayed capacity launches; failure to achieve economic feasibility at scale; high client concentration exposing revenue to risks tied to loss of major clients.

Key data

• MU May Quarter Revenue Forecast$36.454 BillionRepresents a 52.8% sequential increase, exceeding consensus of $35.555 Billion
• MU May Quarter EPS Forecast$21.31Above consensus of $20.57
• DRAM ASP Sequential Growth Forecast45%Previously estimated at 40%, indicating tighter supply-demand conditions
• NAND ASP Sequential Growth Forecast50%Previously estimated at 35%
• MU FY27 Through-Cycle EPS Expectation$35.00Used as a long-term valuation anchor
• MU Target Price$1,050Based on a 29.5 times through-cycle P/E multiple
• CBRS Initial Capacity Deployment Target250 MWExpected to come online by mid-2027

Impact & implications

The report suggests that strong performances from Micron, along with ongoing price appreciation trends, will reinforce investor confidence in extended cycles for memory chips. Confirmation of additional strategic customer agreements during Micron's earnings call could further boost sentiment and support elevated valuation multiples. As for Cerebras, assuming steady progress on capacity rollout—even without immediate financial breakthroughs—their unique technology platform retains meaningful long-term upside potential. Overall, AI-driven demand for storage is reshaping the industry landscape, shifting traditional cyclical patterns such that supply-demand imbalances might persist longer than initially expected.

Risks

• Sudden reversal in memory chip prices: A sharp decline could ensue if terminal demand weakens concurrently with high inventories.
• Lower-than-anticipated HBM demand or heightened competition: Such scenarios could compress margins and reduce profitability.
• Macroeconomic volatility: Economic concerns might prompt customers to cut back on inventory stocking or investing in new technologies.
• Cerebras' capacity ramp risks: Delays stemming from infrastructural challenges or poor implementation could hinder timely activation of the 250 MW facility.
• Cerebras' client concentration exposure: Heavy reliance on few big clients means departures could disproportionately affect operational outcomes.

What to watch

• Any fresh disclosures or developments regarding Micron’s Strategic Customer Agreements (SCAs).
• Comments from Micron about the adjustments made to its Rubin platform's LPDDR content and corresponding responses from clients.
• Concrete schedules and progress reports on Cerebras’ 250 MW capacity installation timeline.
• Actual price trajectories observed for DRAM and NAND during the August quarter together with negotiation outcomes.
• Whether Micron reaches its adjusted annual capital expenditure target of $27 billion and the status of WFE delivery schedules.