Morgan Stanley sees Cerebras’ pullback as an attractive entry point and flags ON’s Analyst Day as a key catalyst

Overview

This semiconductor weekly update centers on Morgan Stanley’s renewed conviction in Cerebras following a sharp share-price decline and on the forthcoming ON Analyst Day. It also provides sector valuation, earnings, inventory and short-interest context.

Core views

Morgan Stanley argues that Cerebras’ 17% selloff over the last month creates an attractive entry point despite two good quarters since its IPO. The firm’s underlying thesis is that low-latency inference and the disaggregation of inference workloads are becoming important AI-computing growth drivers. While investors focus on competing offerings from NVIDIA, Groq and others, Morgan Stanley sees Cerebras as retaining a meaningful early lead in fast inference. The report addresses the principal bear argument that very large AI models cannot fit within a single wafer-scale system. In Morgan Stanley’s view, Cerebras has continued to demonstrate strong results with large-parameter-count models, while the introduction of CS4 shows an ability to scale beyond one wafer. The firm also expects increasing disaggregation of inference workloads, separating prefill from more memory-bound work to optimize around DRAM constraints. It identifies this architecture as central to AMD and Amazon partnerships and increasingly relevant to hyperscalers trying to work around memory constraints. Morgan Stanley retains its $279 Cerebras price target, based on 13x its estimated 2028 adjusted revenue of $6.3 billion. The multiple is described as in line with other small- and mid-cap AI companies with near-100% AI exposure that are growing faster than the broader market. The report further argues that, given the nearly $7 trillion combined market capitalization of the two largest AI processor companies, one percentage point of market share could be worth more than $50 billion. It says Cerebras’ existing single-customer contract could take the company toward that scale by calendar 2028, with material further upside potential. Its risk-reward table shows Cerebras at $191.9, with a $90 bear case, $279 base case and $400 bull case, implying 45% base-case upside. For ON, Morgan Stanley expects the September 16 Investor Day—its first Analyst Day since May 2023—to focus on an updated long-term growth model spanning AI data center, automotive/silicon carbide and industrial markets. At the prior event, management had outlined 10–12% revenue CAGR through 2027, 53% gross margin, about 40% operating margin and 25–30% free-cash-flow margins. The company has continued to reiterate longer-term margin targets, although the original 2027 timing has become less explicit. Morgan Stanley expects discussion of recent content and total-addressable-market expansion, utilization, remaining Fab Right benefits, the contribution of higher-margin new products, capital intensity and free cash flow, as well as how the pending Synaptics acquisition fits into ON’s longer-term strategy and financial model. The weekly sector context shows inventory remaining elevated in parts of the semiconductor chain. Semiconductor-company days of inventory were 118, up five days quarter over quarter versus a seasonal four-day increase, and 24 days above the historical median. Customer inventory was flat sequentially at 60 days versus a seasonal zero-day decline and remained above its historical median. Distributor inventory was 56 days, down five days quarter over quarter versus a seasonal one-day decline, in line with its historical median. The report also presents Morgan Stanley-versus-consensus 2027 estimates and a broad semiconductor risk-reward framework, while maintaining an Attractive North America industry view.

Analysis framework

Morgan Stanley combines company-specific technology and demand analysis with relative valuation, scenario-based risk-reward cases, and sector monitoring. For Cerebras, it links inference architecture, model scalability, DRAM constraints and customer deployment potential to a revenue-multiple valuation; for ON, it frames the Analyst Day around management’s long-term operating targets and strategic updates. The weekly sector view supplements this with estimates-versus-consensus, inventory days and short-interest data.

Methodology notes

• Valuation methodsPS valuation

  Revenue-multiple valuation

  Morgan Stanley values Cerebras at 13x its estimated 2028 adjusted revenue of $6.3 billion, benchmarking the multiple against small- and mid-cap AI peers.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Inventory tracking across semiconductor companies, customers and distributors

  The report uses days of inventory at different points in the semiconductor chain to assess inventory conditions and their sequential movement relative to seasonal patterns.

• Event-Driven and Behavioral FinanceEvent-driven analysis

  Analyst Day catalyst analysis

  The report identifies ON’s September 16 Analyst Day as an event that could clarify its long-term growth, margin, capital-intensity and acquisition strategy.

Asset mapping & comparison

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

• Cerebras Systems (CBRS.O)

  Primary highlighted stock; positioned as a potential beneficiary of fast inference and disaggregated AI workloads.

  Strengths

  Morgan Stanley sees a significant early lead in fast inference, strong large-model results and CS4-enabled scaling beyond a single wafer.

  Weaknesses

  The market’s concern is that larger models may not fit into a single wafer-scale structure.

  Comparison

  The firm compares its 13x 2028 revenue multiple with small- and mid-cap AI companies with near-100% AI exposure; it cites NVIDIA and Groq as competing products.

  Risks

  Capacity deployment and infrastructure risk, customer concentration risk, and competitive risk.

• ON Semiconductor Corp. (ON.O)

  Company facing a forthcoming Analyst Day that may update its long-term financial and strategic model.

  Strengths

  Potential growth opportunities span AI data center, automotive/silicon carbide and industrial markets, alongside higher-margin new products and Fab Right benefits.

  Weaknesses

  The timing of the original 2027 long-term model has become less explicit.

  Risks

  The report does not state a specific ON risk list.

Key data

• Cerebras share-price move-17%Decline over the last month cited as creating an attractive entry point.
• Cerebras price target$279Based on 13x Morgan Stanley’s 2028 adjusted revenue estimate.
• Cerebras 2028 adjusted revenue estimate$6.3 billionThe revenue base used in the firm’s 13x valuation multiple.
• Cerebras base-case upside45%Risk-reward table based on a $191.9 reference price and $279 base case.
• ON prior long-term revenue target10–12% CAGR through 2027Target presented at the May 2023 Analyst Day; the original 2027 timing has since become less explicit.
• ON prior long-term margin targets53% gross margin; ~40% operating margin; 25–30% free-cash-flow marginLong-term model metrics Morgan Stanley expects management to revisit.
• Semiconductor-company inventory118 daysUp 5 days quarter over quarter; 24 days above the historical median.
• Semiconductor-customer inventory60 daysFlat sequentially and above the historical median.
• Distributor inventory56 daysDown 5 days quarter over quarter and in line with the historical median.

Impact & implications

Morgan Stanley’s central implication is that rising demand for low-latency inference and disaggregated AI workloads could support Cerebras’ competitive positioning and create upside if deployments and customer demand expand. For ON, the Analyst Day may determine whether management can provide a refreshed path for growth, margins, free cash flow and the Synaptics acquisition within its long-term model.

Risks

• Cerebras faces capacity-deployment and infrastructure risk.
• Cerebras has customer-concentration risk.
• Cerebras faces competitive risk.

What to watch

• Cerebras’ contract extension, potential additional customers and capacity deployment.
• ON’s September 16 Analyst Day updates on growth in AI data center, automotive/SiC and industrial markets.
• ON’s updated long-term margin, capital-intensity and free-cash-flow framework, including the role of the pending Synaptics acquisition.
• Inventory trends at semiconductor companies, customers and distributors.