GoKawiil
Tech News
Vonder, founded by Viture executives and chief brand officer Guillaume Hamon, has introduced audio-only smart glasses priced $299-$399 that use bone conduction to distinguish the wearer's voice from other speakers in a room. The glasses require a monthly AI subscription, use onboard microphones to summarize conversations via an app, and do not store audio recordings or include cameras.
The Trump administration is pursuing policies aimed at boosting domestic semiconductor production, including memory chips where the US lags foreign producers. However, proposed tariffs on imported components and materials could raise costs for American chipmakers trying to expand production.
Micron has marked its 2GB (16Gb) GDDR7 memory chips, in both 28 Gbps and 32 Gbps variants, as End of Life on its website, according to an X post from UNIKO's Hardware. The chips had only recently entered production and shipped on a limited number of Nvidia cards, including retail versions of the GeForce RTX 5060, with Samsung and SK Hynix supplying the bulk of GDDR7 for the RTX 50 lineup.
A Munich Regional Court ruled on September 18 that Micron infringed two YMTC utility model patents covering 3D NAND technology, granting injunctions with 10-year protection terms. Three other related cases in the same filing were adjourned. Micron says it disputes the ruling and has filed an appeal to the Munich Higher Regional Court.
A gray 32GB Corsair Vengeance DDR5-6000 RAM kit (two modules, CL38-48-48-96 timings, 1.35V) is now available for $439 using promo code SSF73844. This price matches a competing 32GB V-Color Manta XSky kit and is currently the lowest price found for a standalone kit of this capacity and speed.
A developer has built a web-based emulator that recreates the IBM 1620 Model 2, a 1960s transistorized decimal computer originally aimed at scientific and engineering users. The project also includes IPL-V, the 1963 operating environment originally used on the machine, letting users run period software directly in a browser.
A 2024 technical essay argues that the history of computing has overemphasized Charles Babbage's mechanical calculator designs while neglecting the development of electronic working memory, or RAM. The author traces this history back to relay-based bistable circuits, explaining how latching relays could store bits and laid groundwork for modern memory systems.
ChangXin Memory Technologies (CXMT) announced at the 2026 World Manufacturing Convention that it has begun mass production of its fifth-generation DRAM platform, called G5. The company says the platform uses new self-aligned quadruple patterning with an 11.95nm active-area half-pitch, a 45:1 capacitor aspect ratio, and a high-k metal gate process, delivering roughly 50% more die per wafer than prior generations.
Analyst Kurnal Insights calculated that at current AI-driven memory prices, the revenue value of DRAM per square millimeter of silicon now surpasses the nominal wafer processing cost of TSMC's advanced N2 and N3 logic nodes. Using an assumed DRAM price of $1.50 per gigabit, the newest 1b DRAM generation works out to about $0.654 per mm², compared to roughly $0.424 per mm² for an N3 wafer.
Acer CEO Jason Chen told DigiTimes that RAM-driven PC price increases could start reversing by the end of 2027, a far shorter timeline than rival memory executives have forecast. He argues that supply of most RAM and SSD types has already caught up with demand, with real shortages now confined to high-end components like top-tier DDR5 modules and certain CPUs.
A project called BootLife implements Conway's Game of Life entirely within a 512-byte x86 boot sector, using VGA memory to display the simulation grid. It's shared on GitHub as a Show HN post, built with NASM and runnable via QEMU, with source code and a Makefile provided for building and testing.
A commentary argues that most enterprise AI deployments today only store and retrieve information rather than genuinely learn from experience over time. It draws a comparison to an employee who interacts with customers and attends meetings for a year but retains none of the skills or judgment gained, remaining as capable as on their first day. The piece calls for AI systems that actually improve and adapt with use, rather than simply accumulating data.