New Chrome 135 Memory Leak Crashes Video Conferencing for Enterprise Teams

New Chrome 135 memory leak crash leaves video calls frozen for enterprise teams.

Chrome 135 rolled out last week with a memory leak that drains RAM during long video sessions. Enterprise users on platforms such as Zoom and Google Meet reported sudden crashes after forty minutes. The issue hit during peak business hours. Teams lost connections mid presentation and scrambled to switch browsers or devices. This Chrome 135 memory leak crash now sits at the center of a larger problem. Cloud meeting tools depend on browser stability that no longer holds under daily loads. The regression marks the first time since Chrome 130 that a mainstream stable channel update produced measurable downtime across distributed workforces relying on browser-rendered video.

Memory leak surfaces in enterprise reports

Users on public forums posted screenshots showing RAM usage climbing past eight gigabytes in a single tab. One discussions reached 950 upvotes within two days and collected more than 300 comments. IT teams confirmed the pattern across Windows and macOS fleets. The leak appears once video encoding starts and never releases memory after the call ends. Google has not issued a fix yet. The company listed the report as under investigation on its Chromium bug tracker.

Detailed traces from affected machines show the Chromium media component allocating buffers for VP9 and AV1 streams without proper deallocation after stream termination. Engineers who examined heap snapshots found dozens of orphaned MediaSource objects persisting in memory even after the user closed the meeting tab. These objects each held references to large WebRTC data channels, creating a compounding effect when participants joined multiple consecutive calls. In one reported case a single sales representative accumulated 14.7 GB of resident memory after four back-to-back meetings.

Enterprise monitoring dashboards quickly surfaced the problem because baseline RAM consumption for Chrome shifted from an average of 1.2 GB per tab to 4.8 GB within the first hour of use. Security teams noted that the additional memory pressure also triggered paging on laptops equipped with only 16 GB of total RAM, degrading performance across unrelated applications such as Slack and Microsoft Outlook. Network telemetry from managed endpoints revealed that roughly 12 percent of Chrome instances running version 135 exhibited the elevated memory curve within the first 48 hours of deployment, a spike far exceeding the 1.8 percent baseline observed with the prior release.

Enterprise teams bear the immediate cost

Finance and consulting firms lost billable hours when calls dropped. One New York advisory group rescheduled three client sessions after every laptop in the room hit the same wall. Video conferencing now demands more standby hardware. Some departments issued secondary laptops just to keep one browser tab alive during meetings. The Chrome 135 memory leak crash turned a routine software update into a production outage.

Consulting firms that bill by the quarter-hour calculated direct revenue impact exceeding $180,000 in a single Monday morning across three offices. Legal practices reported similar disruptions during depositions conducted over Google Meet, forcing court reporters to request extensions when recordings failed mid-session. Human-resources teams conducting remote interviews suffered repeated connection losses, extending hiring cycles by several days and increasing candidate drop-off rates. Procurement managers responded by accelerating orders for 32 GB and 64 GB laptops even though capital budgets had already closed for the quarter. The unplanned hardware spend created tension between IT and finance departments, prompting some organizations to explore VDI solutions that isolate video processing on central servers rather than on individual endpoints.

In one documented case, a global accounting firm with 4,200 remote employees measured an average of 47 minutes of lost productivity per affected user on the day the issue peaked. The cumulative impact reached an estimated 3,300 person-hours, equivalent to almost two full work weeks.

Technical anatomy of the memory leak

Reverse-engineering reports indicate that a change in Chrome 135 altered the lifetime management of AudioProcessingModule instances used by WebRTC. Previously these modules were destroyed when a MediaStreamTrack ended; the new implementation keeps them alive for potential reuse across navigation contexts. When video calls create and destroy multiple tracks within the same renderer process, the orphaned modules accumulate large ring buffers that the garbage collector cannot reclaim because active JavaScript references remain.

Developers who attached debuggers observed that the leak rate averages 38 MB per minute during a 1080p call and rises to 62 MB per minute when screen sharing is active. The growth is nonlinear because each additional participant contributes its own decoding pipeline that inherits the same faulty reference counting. Disabling hardware acceleration reduces the rate by roughly 30 percent but does not eliminate the leak entirely, confirming that both software and GPU paths contain the defect. Further instrumentation of the renderer process exposed that the leak also affects the VideoTrackAdapter class, which retains references to decoded frame textures longer than necessary. When a meeting tab is backgrounded and then restored, the adapter reallocates textures without freeing prior allocations. This behavior compounds on machines with multiple monitors where Chrome renders preview thumbnails in parallel with the main video stream.

Comparison with prior Chrome releases

Chrome 134 maintained stable memory profiles during extended calls, with average consumption plateauing after the first ten minutes. Internal telemetry shared by several large enterprises showed that 134 users experienced fewer than 0.2 percent crash rates attributable to memory exhaustion. The regression in version 135 therefore represents a clear break from a multi-release trend of incremental stability improvements. Organizations that delayed updates because of earlier compatibility concerns now view that caution as justified, while teams that adopted promptly face the largest operational disruption.

Internal benchmarks circulated among enterprise Chrome administrators reveal that mean time between crashes for browser-based video rose from 97 minutes in version 133 to just 41 minutes in version 135 under identical network and hardware conditions. The delta prompted several companies to implement mandatory version pinning policies for the first time in three years.

Impact across industry verticals

Healthcare providers conducting telehealth visits encountered compliance complications when sessions terminated before informed-consent language could be completed. Hospitals using Chrome-based virtual desktop infrastructure reported that the memory leak propagated to thin-client endpoints, affecting hundreds of simultaneous consultations. Education institutions running hybrid classes found that recorded lectures stored in browser tabs consumed so much memory that students could no longer open supplementary materials during the same session.

Manufacturing firms that rely on remote site walkthroughs using augmented-reality overlays experienced dropped connections during safety inspections, forcing personnel to repeat physical visits. In each sector the Chrome 135 memory leak crash exposed the assumption that browser-based video would remain reliable for mission-critical workflows. Financial-services firms running compliance-mandated recorded meetings faced additional regulatory scrutiny when session logs showed abrupt terminations without proper audit trails.

Workarounds and temporary mitigation strategies

IT administrators have published internal playbooks recommending users switch to the native Zoom or Meet desktop clients, which bypass the affected renderer processes. Additional guidance includes disabling hardware acceleration via chrome://flags and forcing the use of the H.264 codec by appending a query parameter to meeting URLs. Some teams deploy policy settings through Chrome Enterprise that automatically kill tabs whose private memory exceeds 3 GB, a blunt measure that prevents total system lockups but still interrupts active meetings.

Power users have begun scripting periodic tab reloads using extensions that monitor Performance API metrics. While these scripts reduce peak memory, they introduce audio artifacts and reset participant layouts, creating a poor experience for meeting hosts. Several organizations also experimented with running video calls inside separate Chrome user profiles that enforce stricter garbage-collection schedules through command-line flags. Early results indicate a modest 15 percent improvement in memory stability, though at the cost of increased context-switching overhead for end users.

Local capture tools avoid browser dependency

remio records meetings directly on the device without routing audio through any browser tab. The five-level memory system keeps context even when external services fail. Enterprise users gain a second recording path that stays active regardless of Chrome status. Action items and notes continue to populate from the local file.

Download remio to maintain meeting records outside the browser.

Cloud meeting platforms face growing pressure

Zoom and Meet both rely on Chrome for their web versions. When the browser leaks, those platforms inherit the failure. Teams that moved entirely to web clients now question the single point of failure. Desktop apps provide partial relief but still pull data through browser processes on many systems. The Chrome 135 memory leak crash made that dependency visible in real time.

Platform vendors have started publishing guidance that encourages customers to maintain mixed fleets of web and native clients. Internal roadmaps now include optional WebTransport-based recording paths that can operate independently of the Chromium media pipeline. Adoption metrics will reveal whether enterprises treat browser stability as a solved problem or as an ongoing risk factor that justifies additional infrastructure spend.

Practical implications for IT departments

Security and compliance teams must update incident-response playbooks to treat browser memory exhaustion as a potential outage vector rather than a user-support ticket. Procurement cycles should incorporate stress-test criteria for any new endpoint-management tool, requiring vendors to demonstrate behavior under sustained 4K video loads. Training programs need to teach employees how to interpret task-manager output so they can escalate memory anomalies before they reach critical thresholds.

Budget models that previously assumed three-year laptop refresh cycles may require compression to two years for knowledge workers whose primary tool is video conferencing. Insurance carriers offering cyber policies have begun asking explicit questions about browser version governance and fallback recording mechanisms during underwriting reviews. Several enterprises now require proof of local-recording capability as a precondition for approving any new SaaS video platform contract.

Limitations and risks of current mitigation approaches

Native desktop clients reduce exposure yet still embed Chromium components for certain rendering tasks, leaving residual risk. Hardware upgrades address symptoms rather than root cause and increase electronic-waste streams. Scripted tab reloads create inconsistent meeting artifacts that can undermine legal or regulatory recordings. Until Google ships a verified fix, every workaround remains probabilistic rather than deterministic.

Organizations that rely exclusively on a single vendor’s desktop application also face supply-chain concentration risk if that vendor’s update channel experiences delays. Diversifying across local recording tools, native clients, and monitored browser instances therefore represents the only layered defense available in the short term.

Root cause investigation from Chromium source

Chromium commit logs show the introduction of a new lifetime policy for AudioProcessingModule objects intended to improve cross-origin reuse. The policy change inadvertently created a circular reference chain when MediaStreamTracks were created and destroyed within the same renderer process. Because the WebRTC signaling layer retains a weak reference to the adapter for potential reuse, the garbage collector treats the entire object graph as live. The resulting retention prevents deallocation of the associated ring buffers even after the stream has ended. This architectural shift was merged without extended soak testing on long-duration video workloads, allowing the regression to reach stable channel. The change aligns with documented WebRTC media handling patterns described in the WebRTC native code media architecture overview.

Long-term enterprise strategies beyond workarounds

Forward-looking organizations are evaluating WebRTC gateways that terminate media sessions on dedicated appliances rather than inside endpoint browsers. Some enterprises have begun piloting containerized browser instances running on centralized Kubernetes clusters, with video output streamed to lightweight client endpoints via secure remote desktop protocols. These architectures decouple media processing from the endpoint device, providing a scalable path that sidesteps the Chrome 135 memory leak entirely. Early adopters report that while initial deployment costs are high, operational resilience and centralized patch management deliver measurable returns within two quarters.

Outlook hinges on browser and platform fixes

Google is expected to ship a stability update in the next two weeks. Enterprise admins watch for patch notes that confirm memory release after video streams close. Meeting platform vendors may add fallback recording modes that bypass Chrome entirely. Adoption of such modes will signal whether the industry treats browser leaks as routine or as structural risk. Users should track the next Chrome release and any statements from video vendors on local fallback support. The preceding release that demonstrated stable behavior is referenced in the Chrome Stable channel release notes.

What to watch next

Monitor the Chrome Stable release notes for bug ID references related to AudioProcessingModule or WebRTC buffer handling. Watch for coordinated announcements from Zoom, Google, and Microsoft regarding enterprise-grade recording alternatives. Track adoption statistics of local-first meeting tools as organizations publish post-incident reviews in the coming quarter.

Frequently asked questions

How quickly does the leak manifest?

Most users observe noticeable degradation after 35–45 minutes of continuous 1080p video; screen-sharing sessions accelerate the curve by 40 percent.

Does the issue affect ChromeOS devices?

Yes. Early field reports from education and retail pilots show identical memory growth patterns on managed Chromebooks running version 135.

Will rolling back to Chrome 134 resolve the problem permanently?

Rollback eliminates the immediate leak but exposes users to any unpatched vulnerabilities addressed in the 135 security update.

Are there open-source patches available?

Community engineers have published experimental builds with adjusted reference counting, though these remain unsupported and carry stability risks.