Morgan Stanley raises Alphabet target price to $415, betting on 2028 cloud and TPU compute monetization

Overview

This report is Morgan Stanley's deep research on Alphabet Inc., focusing on Google Cloud's 2028 compute expansion, external TPU sales, and the resulting re-rating of revenue, EBIT, EPS, and capital expenditures. The report believes GOOGL has recently been pressured by capital rotating toward hardware and semiconductors, but visibility into its 2027/2028 cloud business and AI compute capacity is increasing, and the current valuation does not fully reflect this change.

Core views

The report maintains an Overweight rating on GOOGL and raises the target price from $375 to $415. The main reasons are: first, GOOGL may add about 9GW of compute capacity in 2028, of which about 7GW will be TPU; second, about 4GW of 1P TPU may be sold externally, generating about $79bn in revenue; third, Google Cloud revenue is expected to reach about $308bn in 2028, with EBIT reaching about $132bn, accounting for about 46% of the company's total EBIT; fourth, 2028 EPS is raised to about $19, and the current share price implies about 18X 2028 EPS, below the long-term average level.

Analysis framework

The report uses a bottom-up compute capacity framework, combining TPU shipment assumptions from AVGO and MediaTek, revenue monetization assumptions per watt, and 1P TPU rack ASP and gross margin assumptions to derive Google Cloud revenue, EBIT, and capital expenditures. For valuation, the report rolls the target price forward to 1H28, using average 2027/2028 EPS and applying a P/E multiple of about 23X.

Methodology notes

• Compute capacity estimationGW compute increment model

  Estimate GOOGL's future deployable compute capacity based on rack power and chip supplier shipments.

  Based on TPU supply assumptions from AVGO and MediaTek, the report expects GOOGL to add about 7GW of TPU compute capacity in 2028, plus about 2GW of NVIDIA GPU capacity, for a total of about 9GW of additional on-premise compute capacity.

• Revenue monetizationRevenue per watt model

  Convert incremental compute capacity into core Google Cloud revenue on a $/W basis.

  In the base case, the report assumes the core Google Cloud business will monetize at about $15/$18 per watt in 2027/2028, leading to estimated core Google Cloud revenue of about $152bn/$229bn.

• External TPU sales1P TPU system sales model

  Estimate external sales revenue based on TPU rack ASP, power density, and gross margin.

  The report assumes a gross margin of about 20% for 1P TPU system sales, corresponding to about $20bn/GW of revenue, and forecasts 1P TPU sales revenue of about $62bn/$79bn in 2027/2028, respectively.

• Valuation methodsP/E and PEG relative valuation

  Derive the target price using average 2027/2028 EPS and a target P/E.

  The $415 target price is based on about 23X average 2027/2028 EPS, implying about 1.4X PEG, a premium of about 25% versus the peer median.

Asset mapping & comparison

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

• Alphabet Inc. (GOOGL.US)

  Core covered asset

  Strengths

  It has multiple AI monetization paths across Search, YouTube, Google Cloud, and TPU, and the report believes fundamental visibility for 2027/2028 is improving.

  Weaknesses

  Capital expenditure requirements are rising significantly, and the market still has differing views on AI search competition, cloud business monetization, and TPU sales profitability.

  Comparison

  The current valuation is about 18X 2028 EPS, which the report believes is below the long-term average NTM level of about 21X; the target price implies a higher PEG premium relative to peers.

  Risks

  New compute capacity may fail to come online on schedule, revenue-per-watt assumptions may be too high, gross margin on external TPU sales may come in below expectations, or AI innovation may fail to drive incremental growth in search and cloud businesses.

• Google Cloud

  Primary growth and profit driver

  Strengths

  The report expects 2028 revenue to exceed $300bn and EBIT to reach about $132bn, making it an important component of the company's AI EBIT engine.

  Weaknesses

  Growth depends on large-scale compute capacity buildout, customer demand, and backlog conversion, with relatively high capital intensity.

  Comparison

  Compared with the traditional advertising business, Google Cloud contributes faster revenue growth and greater incremental profit leverage in the report's model.

  Risks

  If cloud customer demand, pricing, compute utilization, or GenAI product deployment fall short of expectations, the logic for upward revisions to revenue and EBIT could weaken.

• Broadcom (AVGO)

  Important reference for TPU supply and compute shipment assumptions

  Strengths

  Based on management commentary and revenue estimates for AVGO shipments, the report infers total 2028 compute shipments of about 14GW, of which about 60% may be TPU.

  Weaknesses

  There is uncertainty over the final split between retained and externally sold TPU capacity at GOOGL.

  Comparison

  AVGO shipment assumptions are a key input into GOOGL's 2028 TPU capacity model.

  Risks

  Deviation in assumptions around AVGO shipments, revenue per watt, or TPU mix would affect GOOGL's compute capacity and TPU sales model.

Key data

• Target price revisionraised from $375.00 to $415.00The report maintains an Overweight rating.
• Current share price$353.65Closing price as of June 29, 2026.
• 2028 incremental compute capacityabout 9GWOf this, about 7GW is TPU and about 2GW is NVIDIA GPU.
• 2028 external 1P TPU salesabout 4GWThe report estimates this could contribute about $79bn in revenue.
• 2028 Google Cloud revenueabout $308bnIncludes core cloud revenue and 1P TPU sales revenue.
• 2028 Google Cloud EBITabout $132bnAbout 46% of the company's total EBIT.
• 2027/2028 Google Cloud revenue growth106%/44%Reflecting acceleration in the cloud business driven by AI compute capacity.
• 2028 EPSabout $19The report raises 2027/2028 EPS by about 4%/1%, respectively.
• 2027/2028 capital expenditures$350bn/$375bn2027 capital expenditure expectations were raised from $300bn to $350bn.
• Target price valuation multipleabout 23X average 2027/2028 EPSEquivalent to about 22X 2028 EPS.

Impact & implications

If the report's assumptions materialize, GOOGL's investment narrative will further expand from a traditional search and advertising platform into a compounded growth story around AI compute capacity, cloud infrastructure, and the TPU ecosystem. Greater revenue and profit contribution from Google Cloud could support a valuation re-rating; however, the scale of capital expenditures, gross margin on TPU sales, and efficiency of cloud compute monetization remain key factors determining whether the upside can be realized.

Risks

• The assumption of 20% gross margin and about $20bn/GW revenue for external TPU sales still requires more channel validation.
• Capital expenditure in 2027/2028 is enormous, and changes in data center leasing, power supply, racks, and chip costs could affect free cash flow.
• If Google Cloud's revenue-per-watt monetization assumptions are too high, the room for upward revisions to 2028 revenue and EBIT will be reduced.
• If AI search products fail to improve engagement and monetization, or cannibalize the core Search business, valuation re-rating could be constrained.
• If supply timelines from AVGO, MediaTek, and NVIDIA are delayed, the rollout of new compute capacity will be affected.

What to watch

• Whether incremental TPU and GPU compute capacity in 2027/2028 approaches the path of about 7GW/9GW.
• Disclosure on the scale, customer mix, gross margin, and revenue/GW of external 1P TPU sales.
• Whether Google Cloud backlog, partner projects, and revenue growth support a revenue path above $300bn.
• Whether the $350bn in 2027 and $375bn in 2028 capital expenditures can be partially mitigated through leasing or partnership structures.
• Whether GenAI features in Search, YouTube, and other core products can drive higher engagement and commercialization.
• Whether capital rotates back from hardware and semiconductors into AI platform companies such as GOOGL.