Salesforce Service Outages Strike on Day 2 of Dreamforce Conference

Talk about bad timing. Just as Salesforce was riding high on the momentum of its annual Dreamforce conference in San Francisco, the company found itself dealing with the exact opposite of what it wanted the world to be talking about: a global service outage.

The trouble kicked off in the early hours of Wednesday, September 16, with Salesforce's own status page flagging the first signs of disruption at around 7:50 a.m. UTC, or roughly 3:50 a.m. Eastern time. Within a couple of hours, it was clear this wasn't a small, isolated hiccup. Salesforce confirmed shortly before 2 a.m. Pacific that multiple instances across all of its operating regions were affected, and the company's Trust site began warning that customers could face severe delays, intermittent errors, or trouble accessing some services altogether.

For anyone who relies on Salesforce to run their day-to-day business — and that's a lot of people — this wasn't just an inconvenience. The outage even blocked some customers from creating new support cases through the portal, which meant the very people trying to report the problem couldn't easily do so. That's the kind of ironic twist that makes IT teams want to pull their hair out.

And the scale here wasn't small. Hundreds of Salesforce instances were affected across countries including the United States, Japan, India, the UK, France, and Germany. Given that Salesforce counts giants like Amazon, Walmart, Coca-Cola, Toyota, and IBM among its customers, the ripple effects of a disruption like this reach a lot further than most people realize.

The Worst Possible Week

If you had to pick one week for something like this to happen, this was probably it. The outage landed squarely on day two of Dreamforce, Salesforce's flagship annual event, where the company had gathered more than 40,000 people in person and over 200,000 registered attendees online. This year's conference is leaning hard into what Salesforce calls the "Agentic Enterprise," spread across more than 400 sessions — basically, the company's big pitch for how AI agents are going to reshape how businesses operate.

There's something almost too on-the-nose about a company promoting the reliability of its AI-driven future while its core infrastructure is buckling in real time. It's the kind of moment that critics don't need to exaggerate — the timing does the work for them.

What Actually Went Wrong

So what caused it? According to Salesforce's own investigation, the root of the problem traced back to requests stalling while waiting on a response from an internal login service, which ended up eating up available server resources. In plainer terms: something in the login system got backed up, and that backup snowballed into a resource crunch that dragged down performance across the board.

Salesforce's engineers initially weighed restarting systems as a fix, but they eventually zeroed in on increased load on a core system component as the real underlying issue, and decided restarts alone weren't going to solve it. That's a telling detail — sometimes the obvious "turn it off and on again" move just doesn't touch the real problem underneath.

The Long Road Back

Getting things working again wasn't instant. The timeline shows a steady, methodical recovery rather than a quick patch. By around 10:56 a.m. UTC, Salesforce said it had successfully validated a fix on a test instance and had started rolling it out across affected systems. From there, momentum built: by 3:56 a.m. Pacific, the company confirmed the fix had been validated and rollout was beginning across all impacted instances, with a full fleetwide rollout underway roughly twenty minutes later.

Even with the fix in motion, Salesforce was careful not to declare victory too soon. At one point, an update noted that customers were still experiencing severe delays, intermittent errors, and an inability to access some services and create support cases, with the company promising fresh updates every 30 minutes or sooner. It's the kind of cautious, drip-fed communication you'd expect from a company that's learned the hard way not to overpromise during an active incident.

A Rough Stretch for the Stock, Too

The outage didn't do Salesforce's stock price any favors either. Shares had already been under pressure heading into the week, closing Tuesday at $255.65, down about 1.5% from Monday's close, after a sharp rally the day before. Wednesday's premarket trading saw shares slip a bit further, adding another layer of bad news to a week the company clearly wanted to be defined by innovation, not incidents.

The Bigger Picture

It's worth remembering that Salesforce isn't new to this kind of turbulence — the company has weathered high-profile outages before. But there's a particular sting to having it happen live, during the one week of the year when the whole industry is watching. As Salesforce continues to lean into AI and automation as the future of its platform, incidents like this are a reminder that the flashiest new features still depend on the boring, unglamorous plumbing underneath — and when that plumbing springs a leak, even the biggest announcements can get overshadowed.

For now, Salesforce says its fix is working and services are being restored. But expect plenty of follow-up questions in the days ahead about what exactly buckled under the pressure — and whether the company can keep it from happening again the next time all eyes are on it.