UBS Maintains Neutral on Alphabet; Ups Cloud Revenue, Pressured Margins

Overview

This report dissects the structure of Google Cloud's backlog, modeling core, Vertex AI, and AI Labs (Anthropic, OpenAI, Meta) contracts separately. Revenue forecasts are lifted due to earlier recognition of TPU chip sales, while earnings are tempered by lower-margin TPU sales and incremental costs. The neutral rating and $410 target are upheld.

Core views

The core adjustment is a reworked backlog conversion model: 1) Splitting Anthropic orders into cloud services and TPU hardware (fully recognized revenue); 2) Accelerating TPU recognition—25% in 4Q26, remainder in 1Q27-3Q27 (vs. original 2028 timeline). This drives: 2026 cloud revenue up 24% to $126.1B (7% above consensus), with Anthropic/OpenAI/Meta contributing 21%/7%/1%; 2027 up 34% to $232.0B (16% above consensus). However, lower TPU margins and incremental costs pressure 2027 cloud margin to 27.3% (vs. 32.9% in 1Q26). Earnings impact: 2026 GAAP diluted EPS down 1% to $14.21; 2027 down 7% to $14.64 (margin有望回升至38.7% in 2028 as Vertex AI share grows). 2028 cloud revenue down 4% to $181.1B (32% below consensus) due to prior-period recognition shifts.

Analysis framework

The firm uses a layered backlog model: identifying incremental contract nodes tied to major clients (e.g., inferring Meta contract size via disclosure obligations), then applying tailored conversion timelines—TPU sales follow a 25%/75% split across 4Q26 and 1Q27-3Q27, while other backlog converts 50% within 24 months. This highlights non-linear growth: 2Q26 cloud revenue is revised down to $24.1B (from $26.7B), reducing y/y from 96% to 76.8%, yet full-year growth is forecast at 114.9% vs. consensus of 63%. Cost analysis (TPUs sourced from Broadcom then resold) confirms margin pressure sources.

Methodology notes

• Industry/Segment Analysis FrameworkVolume-price decomposition

  Cloud backlog modeled by client and technology segment

  Total backlog is disaggregated into traditional cloud, AI platform (Vertex), and enterprise hardware (TPU) tranches, each assigned conversion rates and margin assumptions to better forecast revenue mix and margin shifts, mitigating linear conversion biases.

Asset mapping & comparison

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

• ALPHABET INC (GOOGL.US)

  Direct beneficiary of faster cloud revenue but pressured by lower-margin TPU sales

  Strengths

  AI Labs contracts add revenue; Vertex platform有望 improve margins longer-term

  Weaknesses

  TPU sales have lower margins; AI search competition intensifies

  Comparison

  Cloud revenue growth notably exceeds consensus

  Risks

  Investigation-related costs, cloud scale-margin challenges, and advertising spend volatility

Key data

• 2026 Cloud Revenue126.1BUp 24%, 7% above consensus
• 2027 Cloud Revenue232.0BUp 34%, 16% above consensus
• 2027 Cloud Margin27.3%Down 5.6 ppt vs. 1Q26

Impact & implications

The model highlights a near-term trade-off: TPU sales boost top-line but weigh on margins, while Vertex AI is expected to improve margins over the medium term. The 26x PE implies market pricing has largely absorbed cloud growth, but monetization uncertainty around generative AI search (e.g., Al Overviews), competitive risks from rivals like Meta, and sustained high investment remain key headwinds.

Risks

• Generative AI search rivals (Perplexity/OpenAI/Meta) may erode market share
• Monetization efficiency of generative AI features remains uncertain
• Regulatory probes and antitrust litigation increase legal costs

What to watch

• Timing and volatility of backlog conversion to revenue
• Acceleration of Vertex AI's share of cloud revenue
• Actual margin performance of TPU chip sales