When datacenters lose power, generators should kick in. That didn’t happen for Google in London

Google attributed the October 20 outage at its europe-west-2-b London availability zone to an external power-cable failure combined with feeder B being open for repairs and a voltage-sensing design flaw and PLC misconfiguration that prevented generators from starting. UPS units provided about five minutes of backup, then ran out, disrupting services used by Lloyds, HSBC, Royal Mail and Revolut. Google says it reprogrammed the PLC, tested electrical-system changes and plans faster automated customer notifications and broader infrastructure-event testing.

Analysis

The more important signal is operational control, not a single-site equipment failure: a maintenance state and faulty control logic combined to defeat layers intended to be independent. That raises diligence questions about change management and testing across Google Cloud, but the disclosed incident alone does not establish a platform-wide weakness. Near term, customer switching is likely constrained by migration costs; competitors such as Amazon Web Services and Microsoft Azure may gain leverage in RFPs and resilience reviews before they gain meaningful workload share. Over 1–3 months, watch for customer remediation costs, SLA credits, renewal language, and any evidence that this was not isolated to the affected zone. Over 6–18 months, broader testing or additional redundancy could raise infrastructure costs; conversely, better multi-zone design may increase customer usage of cloud services rather than prompt exits. The contrarian point is that the outage could strengthen demand for resilient architectures while leaving the underlying provider relationship intact. The stock signal is modest absent repeat incidents or evidence of churn: one zone-specific event is not enough to infer a material change in Alphabet’s consolidated economics. Thesis weakens further if Google demonstrates verified control fixes and no deterioration in Cloud growth or customer retention; it strengthens materially with another infrastructure-level failure or adverse customer/contract disclosures.

Key Decisions for Investors

• No standalone GOOG trade on this disclosure: the likely immediate equity impact is limited, and the article provides no quantified customer, SLA, or financial exposure.
• Track Google Cloud growth commentary, renewal/retention indicators, and any disclosed service credits over the next 1–3 months; treat a measurable deterioration, rather than outage headlines alone, as the trigger to reassess GOOG.
• For cloud-exposed portfolios, review concentration in single-region deployments and customer dependency on Google Cloud. A shift toward multi-region or multi-cloud designs could benefit AWS and Microsoft Azure at the margin, though migration and operating costs may slow substitution.
• Escalate the risk view if Google reports another control-plane or power-related disruption, broadens remediation beyond the affected facility, or customers disclose material contract changes; otherwise keep the issue on watch rather than expressing a directional position.