Salesforce outage halts services for 7.5 hours

Salesforce endured an outage lasting approximately seven and a half hours, which prevented users from accessing the platform and generated “severe delays” together with sporadic errors on Wednesday, even as the company’s premier Dreamforce conference proceeded. The disruption began at 3:50 a.m. EDT, affecting “multiple instances across all regions,” according to the firm’s incident report, and was declared resolved near 3 p.m. EDT after several hours of verification that the applied remedies were effective.

External dependency failure

Initial analysis linked the problem to an “external dependency failure” that involved the older login server. A central system element faced a surge in demand, which reduced its ability to handle incoming requests. Salesforce asserted that no third-party services were implicated. The team first attempted rolling restarts to bring the service back online, then deployed patches on a region-by-region basis.

By 7:20 a.m., a portion of customers reported that functionality was returning to normal, and the company announced work on a “code-level permanent fix.” Nevertheless, the deployment did not finish for several instances, and some automated remedies failed to fully correct the problem. Users noted that scheduled jobs continued to malfunction even after the service appeared restored, prompting Salesforce to manually restart the affected instances.

Later, Salesforce indicated that the scope of impact was “narrower than initially understood,” observing recovery signs around 11 a.m. EDT, with the majority of clients back online. A subset of Hyperforce instances remained the last to regain service. By 11:39 a.m. EDT, mitigations were in place for all instances, and the company kept monitoring the situation until it marked the incident closed at 2:59 p.m. EDT.

For organizations that treat Salesforce as their system of record, several hours of authentication and availability loss can generate a “temporal data problem.” Events that should have occurred at distinct times may be delayed, fail entirely, or arrive out of order. Consequently, a customer interaction might be captured through an alternate channel while Salesforce is down, yet the integration, workflow, or scheduled process that normally logs the event cannot execute.

This situation can cause delays in transactions and service processes, lead APIs and middleware to accumulate retries, timeouts, and queue backlogs, and produce temporary inconsistencies in records. Missed scheduled jobs and workflows may occur, and staff might lose visibility into case histories or customer details even though the underlying data remains intact, resulting in what has been described as “data divergence.”

Architectural lessons and risk assessment

Info-Tech’s Jaffery advises enterprises to move promptly into a reconciliation and data-integrity stage. This effort should verify not only interactive access but also APIs, integrations, scheduled jobs, queues, workflows, automation, authentication flows, and downstream systems.

A thorough post-incident report from Salesforce ought to outline a causal chain that includes the trigger, dependency failure, technical propagation, customer impact, detection, mitigation, recovery, and lasting corrective measures, according to Jaffery. The organization should be prepared to answer detailed questions about which transactions failed, were only partially completed, or were duplicated, as well as which scheduled or asynchronous processes did not run and whether integrations retried successfully or generated a backlog.

One architectural insight highlighted by Jaffery is that a legacy component does not need to be large to be critical. An older authentication service can persist within a modern stack and become a dependency for newer functionalities. The component’s age matters less than its position in the dependency graph, its potential blast radius, and the robustness of its isolation and failure-handling mechanisms.

Read Also: Apple begins Ternus era after Cook

Potential causes and context

David Shipley, chief executive of Beauceron Security, said there are currently no clear indications that the incident was a security breach. “Right now, this bears all the hallmarks of an update gone horribly wrong.”

Shipley referenced a December 2025 event in which Amazon’s internal AI coding agent, Kiro, triggered a 13-hour AWS outage in a mainland-China region, observing, “I’m not going to be shocked if we don’t see some kind of agent role in this kind of scale disaster.”

He added that the series of significant layoffs at Salesforce over recent years may have contributed to the severity of the outage and the subsequent recovery effort. “Having it happen during Dreamforce had to be all kinds of hell, though, for their sales and customer support teams,” he remarked. “Pour one out for them as they work on rebuilding relationships, face-to-face.”