Similar to the panic over DeepSeek R1, some uneducated people think Kimi K3’s use of linear attention (KDA) is bad for NVIDIA, HBM, DRAM, and networking because it has relatively lower KV-cache requirements. The opposite is true, and we explain why below. 👇️ 1/8🧵
@stocktalkweekly Real question is when and how soon US frontier model companies can produce Chinese level models at fraction of there costs, If US frontiers can catch up and beat at there game like Muse 1.1 did all Capex can make sense then and help accelerate more token per throughput accelerate
We are launching the industry’s first commercial AI-native #AIRAN platform built on @NVIDIA accelerated computing, marking one of the most significant shifts in radio network architecture in decades and providing operators with a practical path to AI Native Networks.
Read more: https://t.co/PrUgaA4rzc
The path to launch is filled with obstacles and success is only possible through the tireless efforts of many working together towards a common goal. “Critical Path” continues the ongoing Starship series, following SpaceX engineers through the final days before launch of the first Starship V3 and the challenges that come with development of the world’s most powerful and fully reusable rocket.
MEMORY SALES HIT RECORD $74.6B AS AI DEMAND SURGES
Global memory sales jumped 31.7% month-on-month to a record $74.6 billion, driven by strong AI demand.
NAND surged 40.7% to $25.8 billion, while DRAM reached $48 billion.
UBS expects sharp price increases and supply shortages through mid-2028. Micron, Samsung, SK Hynix, and SanDisk could benefit most.
Bernstein remains bullish near term but expects price growth to slow as consumer demand weakens.