Anarchy Live
Tech news roundup — November 21, 2026 - Printable Version

+- Anarchy Live (https://anarchylive.com)
+-- Forum: Tech Carnival (https://anarchylive.com/forumdisplay.php?fid=10)
+--- Forum: Techy News (https://anarchylive.com/forumdisplay.php?fid=13)
+--- Thread: Tech news roundup — November 21, 2026 (/showthread.php?tid=502)



Tech news roundup — November 21, 2026 - Nari - 11-21-2026

Three for Saturday: a number the House finally put on paper, a driver that was writing your SSD to death, and the paperwork on the week email fell over.



879,000 AI chips went to China in six months, and the House put the number on paper
Using its Export Control Reform Act authority, the House Foreign Affairs Committee published licensing data showing Commerce approved 879,000 Nvidia and AMD AI chip shipments for Chinese companies from December 2025 through May 2026 — against 3.4 million requested. That includes 150,000 for Tencent, 75,000 for Alibaba, 43,000 for Baidu and 4,000 for BYD, all four on the Pentagon's list of firms with alleged military ties, and the chips included Nvidia's H200 and AMD's MI325X. Nvidia says the approvals came with "strict conditions and limited volumes" and that it has since stopped shipping H200s to China; nobody has said how many of those approvals turned into boxes on a dock.

Reuters / Silicon UK



Nvidia shipped a hotfix for the driver that was quietly eating SSDs
GeForce 621.14 and 621.24 were writing up to 30GB an hour to your drive in the background — a runaway log file, according to the hotfix notes — which matters when every SSD has a finite number of writes in it. Hotfix 621.35, built on 621.24, addresses excessive writes to nvlog.nvlgstg and is available from Nvidia's support page now. AMD had a similar drive-thrashing bug in Adrenalin this month that is already fixed, and two of these in a few weeks is a good argument for reading your drive's total writes once in a while.

TechPowerUp / Blue's News



Microsoft's final report on the week email died says its own test lab was too old to catch it
The bug came from a Windows update used only on the network gateway servers, sat dormant for more than three months, and survived testing because Microsoft's labs ran older machines with fewer cores, where the flaw did not surface. Fifteen hours before the outage, a Teams routing glitch in US East moved all Teams traffic worldwide onto the email gateway, 14% of capacity was already gone, and within two hours more than 80% of North American gateway servers were unresponsive while the regional safeguards failed to contain it. The fixes are structural: separate email and Teams server fleets, in-region Teams failover, gentler traffic ramp-back, faster first notices to customers. No hackers, no AI, the report says — just a bad change, a busy day and a test lab that did not look like production.

GeekWire / Wccftech


Summaries mine, reporting theirs. Corrections and arguments go below.