# Salesforce Suffers Global Outage on Day Two of Dreamforce 2026

Page: https://hirantha.me/blog/salesforce-suffers-global-outage-on-day-two-of-dreamforce-2026
Published: 2026-09-26
Written by: hirantha.me Newsroom (AI-generated). An AI model wrote this from the sources listed below; how it is made: https://hirantha.me/about/newsroom
Topics: salesforce, cloud outage, saas reliability, dreamforce, agentforce

> An overloaded internal login service knocked Salesforce offline worldwide for about three hours during its own Dreamforce conference.

## Down during its own showcase

Salesforce, the customer relationship management (CRM) software company, suffered a global service outage on September 16, 2026, the second day of Dreamforce, its flagship annual conference in San Francisco. The timing was awkward: while Salesforce executives were on stage in front of customers and press pitching Agentforce, the company's artificial intelligence (AI) agent product, which it says is already generating over one billion dollars in annual recurring revenue and growing more than 240 percent, customers around the world could not log in to the basic software the AI agents are supposed to sit on top of.

## What actually broke

The outage was not the result of an attack. Salesforce said the trouble began when an internal login service started stalling on requests, and those stalled requests piled up and consumed the server resources a core system component needed to keep processing new ones. Once that component was saturated, logins, application programming interface (API) calls used by integrated third-party tools, and access to some individual instances began failing across the board. In Salesforce's own words, "requests are stalling while waiting for a response from an internal login service, which is subsequently consuming available server resources."

The problem started at roughly 7:50 a.m. Coordinated Universal Time (UTC), or 12:50 a.m. Pacific time, on September 16. Salesforce engineers validated a fix on a test instance a few hours later and began rolling it out across the fleet, with the company confirming that service had returned to normal by around 11:00 a.m. Coordinated Universal Time, putting the core disruption at a little over three hours.

## How far it reached

Salesforce runs on a multi-tenant architecture called Hyperforce, spread across many regional instances, and this outage did not stay confined to one of them. That is part of what made it notable: a single internal dependency was able to degrade service across a large share of Salesforce's global footprint at once, rather than the failure staying contained to one region or one customer's instance the way a more isolated infrastructure problem typically would. Hundreds of instances were affected, with reported impact stretching across the United States, the United Kingdom, Germany, France, India, Japan, Australia, Brazil, and Canada, among others. Beyond basic logins, the disruption also hit Marketing Cloud's preferences page and the ability to create new support cases through Salesforce's own customer portal, which meant some affected customers could not even file a ticket about the outage while it was happening. Salesforce said its separate GovCloud environment, used by government customers, was not affected, and that no customer data was read, altered, or lost during the incident.

## The market's reaction

Investors did not shrug this one off. Salesforce shares fell roughly 3 percent in the trading sessions that followed the outage, settling around $242 to $243, even though Wall Street analysts remained broadly bullish on the company's underlying business, with several price targets still sitting well above $270. A three-hour login outage is not, by itself, the kind of event that normally moves a large enterprise software stock, which suggests investors were reacting as much to the optics, a reliability failure during the week the company was asking customers to trust it with autonomous AI agents, as to the outage's direct financial cost.

## What it means in practice

For any business that runs its sales, service, or marketing operations on Salesforce, this is a reminder that even a three-hour login failure at a single vendor can freeze customer-facing work across an entire organization at once, since so many downstream tools and workflows depend on that one login layer staying up. Practical steps worth taking after an incident like this include confirming your team has an offline or alternate way to reach customers during a CRM outage, checking whether your support and escalation contacts with Salesforce work when the primary support portal itself is down, and treating vendor status pages and trust portals as something to actually monitor rather than assume are irrelevant until the day they are not. The deeper lesson for the industry is more specific to this moment: as vendors push customers toward AI agents that are meant to run increasingly unsupervised inside these platforms, the reliability of the plumbing underneath those agents, in this case a single login service, matters just as much as the intelligence of the agents themselves.

## Questions

**What caused the Salesforce outage?**
An internal login service began stalling on requests, and those stalled requests consumed the server resources a core system component needed, which cascaded into login, API, and instance-access failures. It was not caused by an attack.

**How long did the Salesforce outage last?**
The core disruption ran a little over three hours, starting around 7:50 a.m. UTC and resolved by roughly 11:00 a.m. UTC on September 16, 2026.

**Was customer data affected?**
Salesforce said no customer data was read, altered, or lost, and that its GovCloud environment for government customers was not affected.

**Why did the timing matter so much?**
The outage hit during Dreamforce, Salesforce's own conference, on the same days it was promoting Agentforce, its AI agent product, which undercut the message that its platform is reliable enough to run autonomous agents on.

**What should businesses that use Salesforce take away from this?**
Make sure your team has a fallback way to reach customers if your CRM login goes down, confirm your support escalation path works even when the main support portal is affected, and monitor vendor status pages proactively.

## Sources

- [Salesforce Hit by Global Service Disruption During Dreamforce 2026](https://www.salesforceben.com/salesforce-hit-by-global-service-disruption-during-dreamforce-2026/), Salesforce Ben
- [Salesforce's Global Outage Hit Right As It Pitched Agentforce At Dreamforce](https://startupfortune.com/salesforces-global-outage-hit-right-as-it-pitched-agentforce-at-dreamforce/), Startup Fortune
- [Salesforce staggers back to feet after global outage](https://www.theregister.com/saas/2026/09/16/salesforce-staggers-back-to-feet-after-global-outage/5296800), The Register