Apple AI server chips remain the key debate, Morgan Stanley maintains Overweight and $360 target price

Overview

This report focuses on Apple's progress and risks in AI server chips. Morgan Stanley notes that market sources say Apple's M2 Ultra AI server chip is "struggling" in performance, which could affect the latency and performance of more complex Siri AI queries in Private Cloud Compute, or force Apple to migrate more Siri AI workloads to Google Cloud using NVDA chips, thereby increasing costs. Meanwhile, supply chain checks show Apple has not canceled TSMC packaging capacity, Baltra ASIC is still advancing according to the C1H27 small-volume mass production plan, and the second-generation AI ASIC may enter mass production in 2028.

Core views

The core view is that insufficient AI server chip capability in the short term creates uncertainty for product experience and costs, but Apple's investments in iPhone, Services, and the long-term AI ecosystem still support an Overweight rating. The report forecasts FY26 EPS of $8.89 and FY27 EPS of $10.30, with a base target price of $360; there is upside potential if the iPhone 17 cycle, Apple Intelligence adoption, and services growth exceed expectations, while weak consumer spending, rising memory costs, limited AI feature progress, or greater regulatory pressure would pose downside risks.

Analysis framework

The report uses a combination of supply chain checks, risk-reward scenario analysis, peer regression valuation, and breakdown of key earnings drivers. The valuation framework centers on the FY27 EV/Sales multiple, cross-referenced with FY27 P/E; business analysis focuses on iPhone revenue growth, Services revenue growth, gross margin, EPS growth, AI server chip mass-production timing, and capital expenditure.

Methodology notes

• Valuation methodsEV/Sales and P/E cross-valuation

  The target price is based on 9.1x FY27 EV/Sales, corresponding to about 35x FY27 EPS.

  Morgan Stanley derives the FY27 EV/Sales multiple using regression against technology and consumer platform peers, then combines it with FY27 EPS of $10.30 to test the reasonableness of the target price.

• scenario_analysisBull/base/bear scenario analysis

  Risk-reward under different business and valuation assumptions is measured through Bull/Base/Bear target prices.

  The bull-case scenario is $453, the base case is $360, and the bear case is $194, reflecting changes in the iPhone cycle, AI adoption, services growth, costs, and valuation multiples, respectively.

• supply_chain_checkAI chip supply chain tracking

  TSMC SolC, CoWoS, ASIC tape-out, and Apple capex are used as signals of AI server chip progress.

  These indicators are used to assess whether Apple has sufficient high-performance, low-latency, and cost-controlled Private Cloud Compute inference capability.

Asset mapping & comparison

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

• Apple, Inc. (AAPL.US)

  Core covered name

  Strengths

  The iPhone replacement cycle may accelerate, Services is maintaining double-digit growth, and vertical integration capability plus long-term investment in AI, payments, cloud, health, and home support long-term value.

  Weaknesses

  Limited experience with high-performance AI server chips, possible insufficient performance of M2 Ultra AI servers, and PCC's complex inference capability still needs verification.

  Comparison

  Compared with the existing commercial AI GPU ecosystem, Apple's self-developed high-power AI server ASIC is still at an early stage; if the second-generation ASIC approaches NVIDIA GPU power envelopes, it may require advanced packaging such as HBM and CoWoS.

  Risks

  Rising GCP/NVIDIA usage costs, poor Siri AI performance or latency, higher memory costs, and regulatory and geopolitical risks.

• NVIDIA GPUs

  Potential substitute or external compute dependency

  Strengths

  Can handle higher-performance Siri AI workloads and represent mature commercial AI GPU capability.

  Weaknesses

  May be more costly for Apple and weaken the economics of its self-developed PCC.

  Comparison

  If Apple's second-generation AI ASIC enters mass production in 2028 and its power envelope approaches that of commercial AI GPUs, it may narrow the capability gap with NVIDIA GPUs.

  Risks

  Greater reliance on external chips and cloud infrastructure could affect costs, control, and margins.

• TSMC

  Key manufacturing and packaging supply chain observation target

  Strengths

  SolC and CoWoS capacity reservations can serve as leading indicators of demand for Apple's M-series and AI ASICs.

  Weaknesses

  Apple may use different technology naming within the foundry ecosystem, creating interpretation uncertainty for external observers.

  Comparison

  SolC corresponds more to demand for Apple's M-series and ASICs, while growth in CoWoS could indicate Apple is moving toward higher-power, HBM-related AI chip solutions.

  Risks

  Packaging capacity, tape-out progress, or mass-production ramp may fall short of expectations.

Key data

• Target price$360.00Based on 9.1x FY27 EV/Sales, corresponding to about 35x FY27 EPS of $10.30.
• Current share price$333.74Shown in the report table as the closing price on 2026-07-17.
• RatingOverweightThe most positive rating in Morgan Stanley's relative rating system.
• FY27 EPS$10.30Base-case earnings forecast in the report.
• FY26 EPS$8.89The report believes the iPhone replacement cycle and services growth will drive earnings higher.
• FY27 revenue forecast$557,504百万Fiscal Year 2027E revenue disclosed in the table.
• FY27 net profit forecast$149,643百万Fiscal Year 2027E net profit disclosed in the table.
• Bull/base/bear target prices$453 / $360 / $194Three-scenario target prices in the risk-reward chart.
• FY27 revenue growth14.9%Total Revenue Growth (Y/Y) in the key earnings inputs table.
• FY27 Services revenue growth12.1%Services Revenue Growth (Y/Y) in the key earnings inputs table.

Impact & implications

The investment implication is that progress in Apple's AI server chips could become a key variable affecting future Siri AI experience, cloud computing costs, and the credibility of its AI narrative. If Baltra and the second-generation ASIC advance on schedule, together with the iPhone replacement cycle and Services growth, valuation could be supported; if insufficient PCC chip performance causes more workloads to rely on GCP/NVIDIA, margins may be pressured and the delivery pace of AI features may weaken.

Risks

• Insufficient performance of the M2 Ultra AI server chip could drag on the performance and latency of complex Siri AI queries.
• Apple may need to place more Siri AI workloads on Google Cloud and use NVDA chips, resulting in higher costs.
• Initial mass production scale of Baltra ASIC is small, and as Apple's first high-power chip it carries experimental and ramp-up risk.
• Rising memory costs may compress gross margin.
• If the iPhone 17 cycle is weaker than expected, it may undermine FY26/FY27 growth assumptions.
• Weak consumer spending may limit iPhone upgrade rates.
• Limited progress in AI features or Apple Intelligence adoption below expectations.
• Geopolitical tensions and tighter App Store regulation.

What to watch

• Whether TSMC SolC capacity reservations continue to support demand for Apple's M-series and AI ASICs.
• Whether TSMC CoWoS capacity reservations indicate Apple is moving toward HBM and higher-power AI chips.
• The tape-out progress of the second-generation AI ASIC in 2027 and whether it can enter mass production in 2028.
• Whether Apple's capital expenditure accelerates further.
• Performance, latency, and user feedback for Siri AI after its autumn launch.
• Whether iPhone 17 upgrade intent, iPhone revenue growth, and the replacement cycle accelerate.
• Whether Services revenue maintains or re-accelerates to double-digit growth.
• The impact of memory costs and product mix changes on gross margin.