Microsoft Confirms August Patches Break NDI Streaming—Here’s the Immediate Fix

Microsoft’s August 2025 Patch Tuesday updates are causing severe stuttering, lag, and choppy audio/video for professional video streamers and broadcasters who rely on the NDI protocol. The company acknowledged the regression on its Release Health dashboard after widespread reports that installing KB5063878 on Windows 11 24H2 or KB5063709 on Windows 10 21H2/22H2 devastates performance in tools like OBS Studio and NDI Tools, particularly when Display Capture is active.

NDI, or Network Device Interface, is the de facto standard for moving high-quality, low-latency audio and video between PCs on a local network. It is essential for multi-camera productions, live streaming, and corporate presentation systems. After the August 12 updates, however, NDI streams began suffering from rhythmic frame drops, audio sync drift, and repeated micro-stalls—not just the occasional glitch, but a constant degradation that makes live switching and remote contribution unusable. The problem is most acute on receiving machines, where NDI’s default Reliable UDP (RUDP) transport appears to be the flashpoint.

What Broke and Why It Matters

RUDP is an application-layer protocol designed to combine UDP’s low latency with selective retransmissions and sequencing. It relies on precise timer granularity, socket buffer semantics, and scheduling behavior from the Windows networking stack. When a cumulative update changes kernel timer resolution, I/O completion handling, or buffer management—even subtly—RUDP’s retransmission heuristics can mistakenly interpret normal jitter as packet loss. This triggers a storm of unnecessary retransmissions and queuing pressure, which manifests as the stuttering observed by users.

Field reproductions from multiple vendors and independent press confirm that switching to Single TCP or UDP (Legacy) instantly stabilizes the stream. This pattern strongly suggests the regression resides at the transport level, not in NDI’s media pipeline. Microsoft’s advisory explicitly states: “Severe stuttering, lag, and choppy audio/video might occur when using NDI … The issue affects streaming apps, including OBS and NDI Tools, especially when ‘Display Capture’ is enabled on the source PC.”

Practical Mitigation: How to Keep Shows Running

The fastest path to restoring NDI functionality without rolling back security fixes is to change the receive mode on all affected receivers. Follow these steps:

1. Identify impacted machines: Check for KB5063878 (Windows 11 24H2) or KB5063709 (Windows 10) in Settings → Windows Update → Update history. Alternatively, run winver and look for OS build 26100.4946 or the corresponding Windows 10 build.
2. Install NDI Tools (if not already present) from the official NDI website. The free pack includes NDI Access Manager.
3. Launch NDI Access Manager, open the Advanced tab, and change Receive Mode from RUDP to Single TCP or UDP (Legacy).
4. Click OK and restart all NDI-receiving applications (OBS, vMix, Studio Monitor, etc.). Many clients read the Access Manager setting only at startup.

Choosing Between Single TCP and UDP (Legacy)

• Single TCP provides the most stable fallback but adds a few milliseconds of latency and can suffer head-of-line blocking under heavy packet loss.
• UDP (Legacy) preserves lower latency but lacks RUDP’s retransmission resilience, making it unsuitable for lossy networks.

For most production LAN environments, Single TCP is the safer choice. Always run a full end-to-end rehearsal after making the change, checking lip-sync, frame pacing, and overall latency.

Operational Guidance for IT and AV Teams

Deploying the fix across dozens of endpoints requires planning:
- Pilot rings are critical. Keep a representative set of AV hardware in your early rollout group to catch regressions before they hit production.
- Schedule maintenance windows. Changing receive mode and restarting applications may disrupt ongoing workflows; coordinate with event calendars.
- Do not uninstall the security updates wholesale. Removing KB5063878 or KB5063709 reintroduces over 100 patched vulnerabilities, including a zero-day in Kerberos. The transport mitigation preserves the security baseline.
- For remote guests who cannot reconfigure their machines, consider deploying a LAN relay that accepts a safe transport and forwards the stream internally using the organization’s chosen mode.

Vendor and Community Confirmation

Multiple sources have independently verified the issue and the workaround:
- Microsoft’s Release Health dashboard now includes an entry acknowledging the regression and pointing to vendor guidance.
- NDI’s knowledge base article titled “Traffic Drops After Windows Update” instructs users to change Receive Mode away from RUDP.
- vMix, a leading live production software, published identical guidance with step-by-step screenshots.
- BleepingComputer, Tom’s Guide, The Register, and Neowin all reproduced the symptoms in lab environments and confirmed that the mitigation works.

Broader August Update Context

The NDI regression is not the only hiccup in the August releases. Microsoft also addressed a separate recovery/reset issue with out-of-band fixes in mid-August. Additionally, some outlets reported that the updates caused certain SSDs (especially those with Phison controllers) to disappear or corrupt data under heavy write loads. Microsoft has not yet uniformly acknowledged that storage claim, and it remains partially verified. Administrators should monitor vendor firmware advisories and not conflate it with the NDI problem.

Technical Analysis: Why RUDP Is So Fragile

RUDP sits at the intersection of real-time performance and OS-level networking internals. A change in Windows’ socket buffering or timer coalescing can distort the protocol’s retransmission timeout calculations. Because RUDP interprets any delay in acknowledgement as potential loss, a new timing behavior in the kernel can create a feedback loop of excessive retransmissions. The fact that UDP and TCP—both of which rely on different kernel paths—are unaffected strengthens the hypothesis that the regression is a low-level platform interaction rather than a bug in the NDI library itself.

Risks and Tradeoffs

For the vast majority of cases, the transport workaround is the correct path. Wholesale rollback of security updates should be reserved only for scenarios where the mitigation fails and strong compensating controls are in place.

What to Watch Next

• Microsoft is actively investigating the root cause. Expect a permanent fix in a future cumulative update, likely documented in the release notes.
• SSD-related reports require separate verification. Before taking action, check for firmware updates from your drive manufacturer and monitor Microsoft’s known-issues list.
• Community telemetry will continue to refine understanding. Post your results in forums and on vendor KBs to help narrow down whether certain hardware configurations are more susceptible.

The Bottom Line

The August 2025 patches delivered crucial security protections but also introduced a serious performance regression for NDI-based workflows. By moving NDI Receive Mode away from RUDP, production teams can restore stability immediately while keeping their systems secure. The incident underscores a recurring challenge in modern computing: when a platform update tweaks low-level networking behavior, specialized protocols can break in unexpected ways. A disciplined, tested mitigation—rather than a reflexive rollback—keeps both security and continuity intact. Monitor Microsoft’s Release Health channel and your equipment vendors’ advisories for the final fix. In the meantime, the show must go on, and with a simple configuration change, it can.