Microsoft has resolved the crippling NDI streaming issues that plagued broadcasters and multi‑PC streamers after the August 2025 Windows 11 cumulative update. The fix arrived quietly on September 9, 2025 via KB5065426 for Windows 11 and KB5065429 for Windows 10, directly addressing a regression that broke the Reliable UDP (RUDP) transport used by thousands of live production setups.

NDI, or Network Device Interface, is the backbone of countless professional and amateur broadcasts. It shuttles high‑quality video and audio over Ethernet with extremely low latency, allowing capture rigs, remote cameras, and software like OBS and vMix to talk to each other without dedicated SDI hardware. When the August 12 Patch Tuesday update (KB5063878, build 26100.4946) landed, it silently altered socket behavior at the operating system level, causing NDI’s RUDP mode to fail catastrophically.

The result was immediate and painful: streams froze, audio drifted out of sync, and frames dropped even on perfectly healthy local networks. The issue was most visible during screen‑capture workflows, but it could strike any RUDP‑based NDI feed. Microsoft acknowledged the problem on its Release Health page, while community forums and vendors scrambled for a workaround.

What Exactly Broke?

KB5063878 was a routine security rollup that patched over a hundred vulnerabilities. It wasn’t supposed to touch networking behavior intimately enough to disrupt NDI, but the cumulative nature of modern Windows updates means even small changes to kernel timing, socket buffering, or packet queuing can have outsized effects on application‑layer protocols.

NDI’s Reliable UDP adds sequencing, acknowledgements, and retransmission on top of standard UDP to deliver video that’s both fast and resilient. This reliability layer leans heavily on quick, predictable OS‑level packet handling. When KB5063878 shifted that handling subtly, RUDP’s heuristics went haywire. Retransmission storms flooded internal buffers, head‑of‑line blocking appeared out of nowhere, and delivery became so bursty that every few seconds the picture would stutter or freeze.

Crucially, the bug spared legacy UDP and single‑TCP transports. Because those modes use simpler, well‑understood semantics, they didn’t trigger the pathological feedback loops that paralyzed RUDP. That difference became the foundation of the temporary fix.

The Stop‑Gap Workaround That Kept Shows Running

Within days of the August update, vMix, NDI’s own documentation, and outlets like BleepingComputer and Tom’s Guide published a consistent mitigation: change the NDI Receive Mode on every affected receiving machine from Reliable UDP to either Single TCP or UDP (Legacy). The steps were straightforward:

  1. Download and install the free NDI Tools pack from ndi.video/tools.
  2. Launch NDI Access Manager and switch to the Advanced tab.
  3. Under Receive Mode, choose Single TCP or UDP (Legacy) and click OK.
  4. Restart any application that receives NDI (OBS, Studio Monitor, vMix, etc.) so it re‑reads the new configuration.

The change is reversible and preserved the critical August security patches. It didn’t require uninstalling KB5063878—a dangerous rollback that would reopen the very vulnerabilities Microsoft had just closed. For most studios, the fix was the pragmatic path: accept a slight latency increase or loss of retransmission in exchange for keeping the show on the air.

But the workaround wasn’t free. Single TCP adds a few milliseconds of latency and can suffer head‑of‑line blocking during real packet loss, while Legacy UDP offers zero retransmission—lost packets stay lost, causing visual glitches. For esports, live concerts, and high‑end corporate events, that tradeoff was noticeable. Engineers who needed the lowest possible delay found themselves stuck between a broken transport and a compromised alternative.

The Fix Arrives: KB5065426 and KB5065429

On September 9, 2025, Microsoft released the long‑awaited remedy. The new cumulative updates—KB5065426 for Windows 11 (24H2 and compatible versions) and KB5065429 for Windows 10—directly address the underlying networking change that broke RUDP. NDI developer documentation and vMix’s knowledgebase were updated instantly to confirm that the issue is resolved once these updates are installed.

The fix requires no additional configuration. After applying the September update, NDI streams using Reliable UDP will again function as they did before the August disruption. Users who had switched to Single TCP or UDP can now safely revert the Receive Mode back to RUDP and restart their applications. For those who postponed the August update entirely, Microsoft recommends installing both the August security rollup and the September fix together, as the latter is cumulative and includes all previous patches.

How to Confirm You’re Protected

Any Windows user who relies on NDI should verify their system state immediately:

  • Open Settings → Windows Update → Update history.
  • Look for KB5065426 (Windows 11) or KB5065429 (Windows 10). If present, you’re covered.
  • If you still have KB5063878 without the September update, install the latest cumulative update right away.
  • Once the update is applied, open NDI Access Manager, return the Receive Mode to RUDP (the default), and restart your receiving software.

A thorough test is non‑negotiable. Run a full rehearsal under the same capture conditions you’ll use live—especially if your workflow involves Display Capture or GPU‑accelerated encoding—to confirm that frame rates and audio sync are rock‑solid.

Lessons for Live Production Windows Management

The NDI/RUDP regression was a stark reminder that even routine security patches can destabilize specialized workflows. Windows 11’s servicing model bundles hundreds of fixes into a single package, and while Microsoft tests broadly, the sheer diversity of real‑world streaming setups makes pre‑release validation of every protocol interaction impossible.

For broadcast engineers, the incident reinforces a handful of battle‑tested practices:

  • Pilot updates before wide deployment. A dedicated test ring of identical hardware and software should always vet new patches against typical live production workloads—not just office applications.
  • Have a documented rollback plan. The ability to quickly revert a problematic KB without jeopardizing security is essential. Team members should know how to use DISM or Windows Recovery to pull back a cumulative update if necessary.
  • Monitor vendor and community channels. The vMix knowledgebase, NDI’s official FAQ, and Microsoft’s Release Health page became the clearinghouses for workaround instructions within hours. Being plugged into those sources can cut downtime dramatically.
  • Treat transports as configurable variables. The ease with which the production world switched from RUDP to TCP or UDP highlights the value of having multiple transport options. When protocol‑specific bugs emerge, having a fallback that’s already tested within your environment can avert a show‑stopping crisis.

Microsoft’s Response and What Comes Next

Microsoft’s handling of the issue was commendable for its speed, if not its initial transparency. After acknowledging the problem in late August, engineering teams isolated the root cause and delivered a targeted fix within four weeks—a timeline that, while painful for affected streamers, is relatively quick for a component as fundamental as the networking stack.

The fix is not a Known Issue Rollback (KIR) but a full servicing update. This means it permanently corrects the behavior and will be included in all future cumulative updates. Organizations that manage patches through WSUS, Intune, or other deployment tools can now approve the September update without fear of breaking NDI.

Looking ahead, the NDI ecosystem is likely to see deeper collaboration between Microsoft and protocol developers. The fact that RUDP was disabled by an OS‑level change—rather than an application bug—suggests that networking abstractions in Windows may need to expose finer‑grained controls to latency‑sensitive software. As live production moves further onto commodity hardware and standard Ethernet, the boundary between “application” and “OS” will only become more critical.

For Users Still on the August Update

If you applied the Receive Mode workaround and haven’t yet installed the September patch, now is the time to move. The security protections in KB5063878 remain essential; the new update layers the fix on top of them. After installation, you can restore RUDP and enjoy both security and optimal latency.

For users who rolled back KB5063878 entirely to keep productions running, the September update is your safe path back to a fully patched system. Re‑install the August update (it will come as part of the September cumulative), confirm that NDI works, and then remove any temporary compensating controls like extra firewall rules or network segmentation.

A Painful Detour, but a Clear Road Ahead

The saga of KB5063878 and its RUDP ruin will be remembered as one of 2025’s more disruptive Windows patching incidents—up there with print‑nightmare and early Windows 11 performance regressions. But it also demonstrated the community’s resilience: within 48 hours, a reliable workaround was documented in multiple languages, and within a month, a permanent fix was delivered.

For broadcasters, the takeaway is clear: trust but verify every Patch Tuesday, maintain alternative transport configurations, and keep a close eye on the Release Health dashboard. The September fix puts NDI back on solid ground, and for the thousands of streamers who rely on it daily, that means the show can go on—this time, without the stutter.