Gemini pause: critical error impacts users worldwide this Monda

An error message flashed across thousands of users’ screens on the night of May 12, 2025. The Gemini artificial intelligence service, developed by Google, experienced a significant outage at 11:33 PM Brasília time, leaving users unable to access its functionalities. The failure, which displayed the message “Wait a minute and try again,” caused frustration among individuals and businesses relying on the tool for tasks like text generation, data analysis, and real-time support.

The outage caught many by surprise, especially as it occurred during peak hours in various parts of the world. Reports of issues began surfacing on social platforms, with users sharing screenshots of the error. Below are some of the immediate impacts of the failure:

• Disrupted workflows: Companies using Gemini for process automation reported delays.
• User frustration: Students and freelancers couldn’t complete urgent tasks.
• Decline in trust: The unexpected pause raised questions about the service’s reliability.

The incident, though temporary, exposed the growing dependence on artificial intelligence tools across diverse sectors. Google has not yet provided a detailed official explanation, but the failure has reignited discussions about the robustness of large-scale AI systems.

Initial reports of the outage
Around 11:30 PM, users across different continents began reporting problems accessing Gemini. The error message, asking users to wait and try again, appeared on both the web interface and mobile apps. Within 30 minutes, the volume of complaints surged, with Gemini-related hashtags trending on social platforms.

The outage was not uniform. Some users accessed basic functions, like simple queries, while others faced a complete shutdown. Initial reports suggested the failure primarily affected high-user-density regions, such as North America, Europe, and parts of Asia.

Companies integrating Gemini into their systems, such as customer service platforms and data analysis tools, reported challenges in maintaining normal operations. A São Paulo-based software developer noted that their team lost two hours of productivity due to the outage.

History of Gemini instabilities
Gemini, launched as one of Google’s flagship AI offerings, has faced technical issues before. In March 2025, a flawed update compromised the model’s safety filters, allowing inappropriate responses in some cases. The May 12 outage, however, appears to be on a larger scale, directly impacting service availability.

Similar issues have affected other AI platforms. In 2024, for instance, ChatGPT experienced a multi-hour outage, sparking debates about the need for more resilient systems. In Gemini’s case, the May 2025 pause underscores the importance of investing in infrastructure to handle usage spikes.

Below are some notable incidents involving Gemini:

• February 2024: Pause in image generation due to accuracy issues.
• March 2025: Update that compromised safety controls.
• April 2025: Temporary halt in Gemini Advance 2.5 Pro due to app failures.
• May 2025: Global outage at 11:33 PM, the focus of this report.

The frequency of these events has led experts to question Google’s ability to maintain Gemini’s stability in a highly competitive market.

User reactions
The Gemini outage triggered a wave of reactions on social media. Small business owners, who use the tool for content creation or task management, expressed frustration over Google’s lack of immediate communication. A London-based user posted about missing a critical delivery due to inability to access Gemini for report revisions.

Cobertura completa: Internet

Students were also affected. Online forums saw reports from university students highlighting difficulties using Gemini for academic research on the night of May 12. The outage coincided with assignment deadlines at some institutions, amplifying discontent.

On a lighter note, some users took a humorous approach, sharing memes about the error message. The phrase “Wait a minute” became a target for jokes, with edits comparing Gemini to an overworked employee needing a break.

Recovery efforts
By 12:15 AM on May 13, Google began restoring Gemini access in some regions. The company issued a brief statement on its service status page, noting that technical teams were working to resolve the issue. However, the lack of details about the cause drew criticism.

The restoration was gradual. North American users reported partial recovery by 12:30 AM, while Asia saw prolonged instability. Google pledged ongoing updates but provided no clear timeline, leaving many users dissatisfied.

Businesses relying on Gemini for critical operations, such as real-time data analysis, sought temporary alternatives. Tools like Claude, from Anthropic, and Copilot, from Microsoft, saw increased traffic during the outage.

Speculated technical causes
While Google has not confirmed the outage’s cause, experts have suggested several possibilities. Server overload is one hypothesis, given the recent surge in AI tool usage. Gemini, which processes millions of daily requests, may have faced an unexpected demand spike.

Another theory points to issues with Google’s cloud infrastructure integration. Recent Google Cloud updates, which support Gemini, may have introduced instabilities. In technical forums, developers speculated that the outage could be linked to quota limits, such as the “429 Resource has been exhausted” error seen in prior cases.

Possible speculated causes include:

• Traffic spike: High demand during peak hours.
• Cloud issues: Problems in Google Cloud infrastructure.
• Failed updates: Errors in recently applied patches.
• Quota limits: Restrictions on request processing.

None of these hypotheses have been officially confirmed.

Market impacts on AI
The Gemini outage comes at a time of fierce competition in the AI sector. Tools like ChatGPT, Claude, and Grok, from xAI, vie for market share with Gemini, which has been investing in enhancements like the 2.5 model launched in March 2025. The May 12 failure could erode confidence among corporate users who demand high availability.

Tech companies using Gemini in their products, such as automation platforms and virtual assistants, faced operational challenges. Some shifted tasks to competing AI models, potentially benefiting rivals in the short term.

The pause also raises questions about AI service scalability. With growing adoption of these technologies, incidents like this highlight the need for investments in redundancy and resilience.

Announced preventive measures
Following the partial service restoration, Google stated it is reviewing its systems to prevent future outages. Engineering teams were tasked with identifying the root cause, with a focus on enhancing server capacity. The company also committed to greater transparency in future communications.

Announced measures include:

• Capacity expansion: Scaling up server infrastructure.
• Proactive monitoring: Tools to detect usage spikes.
• Gradual updates: Phased patch rollouts to avoid failures.
• Priority support: Faster assistance for corporate users.

The effectiveness of these measures remains to be seen, but they signal an effort to rebuild public trust.

Repercussions on other Google services
The Gemini outage did not directly impact other Google products, such as Workspace or Firebase. However, it fueled speculation about the overall stability of the company’s infrastructure. Users of services like Google Cloud, which powers Gemini, reported slowdowns in some regions.

The interconnectedness of Google’s services amplifies the impact of such incidents. Companies using Gemini alongside tools like BigQuery faced disruptions in integrated workflows. Google assured users it is investigating potential ripple effects.

Event timeline
The Gemini outage followed a clear sequence. At 11:33 PM, the first error reports emerged, with the “Wait a minute” message dominating interfaces. Within an hour, the failure spread globally, affecting millions of users.

Cobertura completa: Tecnologia

By 12:15 AM, Google acknowledged the issue, beginning recovery efforts. Partial restoration started in some regions by 12:30 AM, but full normalization only occurred after 2:00 AM in some areas. Below are the key moments:

• 11:33 PM: First error reports.
• 11:50 PM: Peak of social media complaints.
• 12:15 AM: Google’s initial statement.
• 12:30 AM: Partial restoration in some regions.
• 2:00 AM: Service stabilized in most areas.

The swift response was praised by some, but the lack of clarity on the cause sustained dissatisfaction.

Alternatives sought by users
During the outage, many users turned to other AI platforms. Claude, known for its stability, was a popular choice among developers. Grok, from xAI, also saw increased usage, particularly for its streamlined interface.

Companies with Google support contracts sought direct assistance, but delays in responses prompted some to adopt internal solutions. Small businesses, without access to priority support, were hit hardest, resorting to free tools or manual processes.

The temporary migration underscores user flexibility but also highlights the risks of relying on a single service.

Compensation plans
Google announced it is evaluating compensation options for affected users, particularly corporate clients with premium contracts. Details on potential refunds or credits remain undisclosed, but the company promised updates in the coming days.

Individual users, accessing Gemini for free or through basic subscriptions, are unlikely to receive compensation. The decision sparked criticism, as many rely on the service for daily tasks. Google emphasized that service improvements are its primary response.

Lessons for the industry
Incidents like the Gemini pause serve as a reminder of the complexity of operating global-scale AI platforms. The reliance on cloud infrastructure, coupled with exponential growth in AI tool usage, demands continuous investment in redundancy and monitoring.

Competitors like OpenAI and Anthropic may leverage the incident to promote their services’ stability. For Google, the outage is a chance to refine processes and bolster user confidence.

The May 12, 2025, incident, though resolved in hours, leaves a warning for the industry: despite its advancements, artificial intelligence still faces significant technical challenges.