If I had to stereotype my X experiences with markets: China 🇨🇳: set on cloning me with AI, can only think of trades in short term timeframes from A-shares PTSD. America 🇺🇸: bullish on anything futuristic like $SPCX, don’t care about valuations Europe 🇪🇺: from $SIVE to $SOI, cares more about water usage than the AI buildout. Somehow can only look at past 12 months. (Belgium is cool so far), looking at you France Sweden Korea 🇰🇷: leveraged degens. I’ve never seen a market so volatile. Equivalent of 50x hyperliquid traders but with stock markets. Japan 🇯🇵: somehow supportive of everything, haven’t seen any Japanese person aggressively bear post and short stocks before. Not enough data on other places yet like Latin America, but will have some soon enough ig.

避孕方式有效率：

BREAKING: OpenAI has confidentially filled for an IPO.

“We expects the prices to increase to $53/GB in 2027, when Vera Rubin will be shipping in volume.” Bernstein is very bullish on 2027 HBM4 pricing. LFG!!!

$AMD says unified memory architectures (UMA) are becoming more important for AI PCs, workstations, and future high-performance platforms The idea is that CPU, GPU, and memory share one large memory pool instead of separating system RAM from GPU VRAM. This helps AI workloads because large models need a lot of accessible memory David McAfee, VP & GM of Ryzen CPU and Radeon Graphics, said: "With Strix Halo, with $NVDA entering the space as well, there's going to be a lot of focus on UMA systems, and on identifying the right architecture for what these UMA systems can do." "This is a totally new workload, a totally new computing space that we're solving for here, and I think it will shape lots of things around our product choices, roadmap products, and deployment options." "What NVIDIA did with their announcements is an endorsement of that architecture.""The emergence of agentic compute and running supersized models because of the unified memory pool of these systems is an incredibly unique value." Unified memory architectures are rapidly becoming a foundational pillar of next-generation computing With agentic AI driving demand for massive shared memory pools, both AMD and NVIDIA are now validating the UMA approach AMD’s confidence in this direction, highlighted by Strix Halo and the expected Ryzen AI MAX 400 series, suggests we are still at the beginning of this trend As unified systems blur the lines between CPU, GPU, and memory, they could unlock new levels of performance, efficiency, and capability, not just for AI workloads, but potentially for gaming and high-end desktops as well

Jensen - “All of my friends at Korean companies: LG, SK hynix, Samsung, Hyundai, and Naver, are all booming.” One special highlight was how important memory has become in NVIDIA products.

Anthropic's Opus 4.7 and 4.8 models are experiencing degraded performance, which is causing a higher rate of failures for users selecting these models in Notion AI. To mitigate impact, all Anthropic models have been disabled in the model picker and requests have been rerouted to alternative providers. Most users should now be able to continue using Notion AI with minimal disruption, though Anthropic-specific features remain unavailable. Please refer to notion-status.com/ for the details.

Google has published a paper that might end the transformer era. For the last 7 years, every major AI, ChatGPT, Claude, Gemini, has been built on the exact same architecture: The Transformer. But Transformers have a fatal flaw. To remember context, they have to process every single word against every other word. It’s called quadratic complexity. As your prompt gets longer, the compute cost explodes. The alternative is the old-school RNN (Recurrent Neural Network). RNNs are incredibly cheap and fast, but they have a fixed memory size. If you give them a long document, they get amnesia. Until today. Google researchers published Memory Caching: RNNs with Growing Memory. And it fixes the biggest bottleneck in AI. Instead of an RNN having a fixed, rigid memory that constantly overwrites itself, Google gave it a "save" button. The technique allows the RNN to cache checkpoints of its hidden states as it reads. The memory capacity of the RNN can now dynamically grow as the sequence gets longer. They built four different variants, including sparse selective mechanisms where the AI actively chooses exactly which checkpoints matter most. The results rewrite the rules of efficiency. On long-context understanding and recall-intensive tasks, these new Memory-Cached RNNs closed the gap with Transformers. They achieved competitive accuracy without the explosive, quadratic compute cost. It perfectly bridges the gap between the cheap efficiency of an RNN and the massive capability of a Transformer. We have spent billions scaling Transformers because we thought they were the only way an AI could remember a long conversation. But Google just proved we don't need to process the whole history every single time. We just needed a smarter cache.