GOOGL Down 7% in the Past 3 Months: Buy, Sell or Hold the Stock?

Alphabet raised 2026 AI-related capital-expenditure guidance to $195-$205 billion after Q2 capex doubled to $44.9 billion, contributing to negative free cash flow of $5.855 billion and a 7% share-price decline over three months. Offsetting these investment and profitability risks, Google Cloud revenue rose 82% year over year to $24.8 billion and backlog reached $514 billion, with more than half expected to convert within 24 months. Analysts lifted 2026 EPS estimates 41.4% over 60 days to $20.51, but regulatory liabilities, cloud competition and elevated infrastructure costs support a wait-and-see stance.

Analysis

The key underwriting question is not demand but capital efficiency: Alphabet is bringing forward a multi-year depreciation burden while a meaningful portion of enterprise monetization is deferred. This creates a likely 2-4 quarter mismatch in which reported revenue can remain strong while incremental operating margin and FCF disappoint. The market will increasingly value Google Cloud on backlog conversion quality—duration, cancellation terms, and gross-margin profile—rather than headline contracted backlog, particularly where capacity is sourced from third parties.

Competitive dynamics favor hyperscalers with the most flexible capacity and enterprise distribution. MSFT is best positioned to monetize AI through existing software-seat economics, while AMZN can use its scale and merchant cash flows to tolerate infrastructure returns that would be dilutive for a standalone cloud provider. Conversely, Alphabet's TPU commercialization is strategically important because it could reduce NVDA dependence for customers and create a credible alternative accelerator stack; however, it is not yet a near-term earnings offset without evidence of external unit economics and customer adoption beyond internal workloads.

Consensus may be too focused on AI search cannibalization and too little on remedy-driven changes to distribution economics. Even modest deterioration in default-placement economics or data-sharing obligations can lower the durability of high-margin search cash flows precisely as fixed infrastructure costs rise. The near-term setup is therefore asymmetric: strong top-line prints may not rerate the stock unless management demonstrates stabilization in capex intensity, cloud margin, and FCF conversion; a miss on any one of these metrics could compress the multiple despite revenue growth.

The contrarian bull case is that capacity scarcity, rather than weak ROI, is suppressing near-term cloud margins and that utilization rises faster than depreciation over the next 12-18 months. That thesis is falsified if backlog conversion decelerates, third-party capacity spending expands faster than cloud revenue, or 2027 capex growth remains materially above revenue growth after new capacity comes online.

Key Decisions for Investors

• Maintain a neutral-to-underweight GOOG stance into the next 1-2 earnings reports; add only if Cloud operating margin expands sequentially despite higher depreciation and management provides a credible path to capex growth below revenue growth in 2027. Downside trigger: another FCF deterioration alongside unchanged capex plans.
• Express relative AI monetization through long MSFT / short GOOG over 3-6 months. MSFT has more immediate software-based monetization and less dependence on preserving legacy search distribution economics; reassess if Google Cloud margin expansion exceeds Azure-related growth expectations or regulatory remedies prove economically immaterial.
• For a lower-beta cloud-capacity expression, favor AMZN over GOOG for 6-12 months: AWS can absorb infrastructure investment through a broader cash-generative ecosystem and may benefit if customers diversify workloads across providers. Exit if AWS growth materially lags Google Cloud while AWS margin fails to hold.
• Do not add directional NVDA exposure solely on Alphabet's capex trajectory. Monitor external TPU sales disclosures and accelerator mix: verified TPU displacement at enterprise customers would be a medium-term negative read-through for NVDA, but current evidence supports an alert rather than a trade.
• Watch for regulatory remedy implementation details over the next 3-12 months; use any evidence of reduced default-search payments, materially lower traffic-acquisition economics, or mandated data access as a catalyst to increase the GOOG underweight.