How a Google Cloud Bug Crashed Major Internet Services Worldwide

On June 12, 2025, the internet experienced one of its most significant disruptions in recent history. Platforms such as Snapchat, Spotify, and Discord began malfunctioning almost simultaneously. At the heart of the issue was a breakdown in Cloudflare’s Workers KV service, which encountered near-total error rates for over two hours. The source of the chaos was eventually traced back to a single code change made within Google Cloud Platform’s infrastructure.

The incident revealed just how vulnerable global internet infrastructure can be when a key provider like Google Cloud encounters issues. Though Google would later issue a public apology, the damage had already been done, and the tech world had been thrown into disarray.

Not only were third-party services affected, but Google’s own platforms were impacted as well. Gmail, Google Calendar, Drive, and Meet all experienced outages. Incidents of this scale are not just disruptive—they are financially devastating. Businesses dependent on Google Cloud’s systems lost millions of dollars in a matter of hours.

Typically, cloud providers guarantee uptime through service level agreements (SLAs), promising availability above 99.99 percent. When service drops below those thresholds, customers receive compensation via SLA credits. While these payouts might cost Google millions, the greater blow was to its credibility. Already behind industry leaders AWS and Microsoft Azure, Google Cloud’s standing suffered further as a result of this incident.

The bug itself originated from a code update made on May 29, 2025. Engineers had implemented a new quota policy check within a key API management service. This component is responsible for routing incoming API calls, confirming authorization, and enforcing usage limits. While the update was technically deployed weeks earlier, it remained dormant because its execution required a policy change—which never occurred during the testing phase.

On June 12, a global policy update triggered the flawed code across all 42 of Google’s data center regions. This activated an untested execution path that led to null pointer exceptions, crashing the binary and initiating an endless restart cycle. The failure effectively brought down the API gateway and created a ripple effect throughout the infrastructure.

The core issue was a lack of proper error handling. Null values weren’t accounted for, leading the new logic to crash when faced with unexpected data. Ironically, this failure came shortly after discussions within the tech community about the dangers of null pointers—particularly in languages like C where unchecked values can cause serious damage.

Google’s engineers began rolling back the change 40 minutes after the incident began. However, complete system recovery took several hours. Although services were eventually restored, the event served as a stark reminder of how even the most advanced systems are not immune to human error.

This widespread outage underscored a key lesson: while cloud computing brings efficiency and scalability, it also comes with substantial risks. One overlooked code path can impact millions of users and businesses around the world. As artificial intelligence increasingly contributes to writing production-level code, the necessity of thorough human oversight and rigorous testing only grows.

Software developers, quality assurance teams, and operations engineers must remain vigilant. It’s not enough to test likely scenarios—every possible execution route, even those that appear unreachable, must be evaluated. The financial and reputational stakes of an outage like this are immense and long-lasting.

Regardless of whether you’re building on Google Cloud, AWS, or Azure, this incident serves as a crucial reminder to implement robust safeguards and quick rollback mechanisms. Preventing the next global disruption starts with disciplined engineering and attention to the smallest details.