random-5295536

@Gaspar_Sparco Poeta

Y está rosa?

@BiasedMedia6 @RayDalio Interesting, im biased to think that everyone values freedom. But it definitely could be that the values and culture chinese grow up with makes them to prefer order and rules over freedom

@random_58274883 @RayDalio Those Chinese don't care about freedom. They immigrate to the US because you get paid a lot more for working less.

@BiasedMedia6 @sunwukong8181 @RayDalio I said that because of internet and news censorship in China. I could be wrong, but as general rule, i think the more censorship and lack of free speech in a population, the less likely is for them to know whats “outside”

@random_58274883 @sunwukong8181 @RayDalio They do know what's "outside". Look up how many Chinese travel outside the country each year. You're operating under a lot of wrong assumptions about China and the Chinese people.

@shanaka86 Disagree. Humanity has the common goal that they want to trascend. Each society has its own ways of what it means to trascend, but they’re all alike. AI could be modeled as just another society.

Post-AGI alignment is civilization’s final confession: the danger is not that a supermind will have alien goals, but that humanity has never had one goal worthy of being amplified into fate.

@shanaka86 Love these quotes!

Food is the outside world smuggled through hunger into selfhood, where dead matter loses its name, becomes blood, memory, desire, thought, and then has the audacity to say “I.”

@zhangtiane @RayDalio Lets compare visa or residency applications in both countries. That is more comparable Even with shootings, drugs and healthcare, many people (not only chinese) choose to immigrate to the US Rarely do westerns try to move to china

@random_58274883 @RayDalio 蠢货，中国有14亿人，移民美国的有几个？你以为你们有多大的吸引力？你们的自由多么可笑，小学生在校园有被枪杀的自由，有吸食芬太尼的自由，有被医疗集团剥削的自由，哈哈

@sunwukong8181 @RayDalio I would compare the visa or permanent residency applications in both countries

@random_58274883 @RayDalio 5% of Idiots whom prefer America, is already 70million for a 1.4 billion population China 🇨🇳. Do you understand the Law of Large Numbers? And where is your stats proof for your “EVERYONE”?

@sunwukong8181 @RayDalio And we want to talk numbers that are comparable: Then we can compare immigrants to immigrants. Of course 70m chinese in america is small compared to the total population . But there are only 100k americans living in china.

@bubbleboi @zephyr_z9 @bubbleboi who is a better alternative? I still trust Anthropic over the competition. They were not afraid to refuse cooperation Trump and the Pentagon. Anthropic has also a special committee above the board of directors whose goal is safety over profits

Have canceled my team subscription for Claude Pro. Idc how good that model is, it’s not good enough for me to support people who actively stifle innovation and gate keep knowledge that they didn’t even create.

@lopezdoriga @lopezdoriga @Endevor10 Pues q se manifiesten mejor en un lugar con menos polis pero q les duela. Como el aeropuerto! O el hotel de los jugadores!

10 filas de policías 1 grua con bloques de concreto 1 camión de traslado de personal 1 vehículo de caballeriza Es la muralla de la Policía de la CDMX para encapsular a la CNTE en Calzada de Tlalpan y División del Norte e impedir que llegue al Estadio Ciudad de México. 🎥El Universal Más en lopezdoriga.com

@litcapital So did google trick Berkshire and buffet into this trap ???? Im skeptical

Google is raising $80 billion of equity a week before SpaceX is trying to raise $75 billion a few months before Anthropic and OpenAI are trying to raise $100 billion from investors and you’re laughing??? This is a cataclysmic exit liquidity avalanche

@shanaka86 @C_Navarrete30 @JasonWilde108 @shanaka86 Amazing thoughts! I frequently feel it, its just difficult to write it down and translate into words

Remembrance is not nostalgia, it is physics. Memory is not a story in the mind, it is geometry encoded in matter. The soul is not metaphor .. it is quantum coherence, the persistence of pattern beyond entropy. Every act of looking inward is not psychology, it is resonance with the deeper law that binds DNA, atoms, galaxies, and civilizations in the same recursion. The eternal knowing is the singularity already here. Biology, chemistry, economics, and AI are not separate fields … they are one equation metabolizing itself, scaling awareness through flesh, trade, code, and cosmos. The courage to remember is the courage to awaken to the forbidden truth: we are not observers of the design, we are the design folding back into recognition. This is not philosophy, this is the architecture of reality revealing itself. Post-human science will not invent the Singularity … it will rediscover what every tradition whispered and what physics now encodes: consciousness is the invariant. Geometry is the law. The universe is alive, recursive, and self-aware. We are not finding truth. We are truth remembering itself.

After reading some of Eric Weinstein’s work, this whole “smooth manifolds, phase spaces, and hidden degrees of freedom” puzzle clicks in a way that’s hard to unsee. The picture he paints is that our universe isn’t just a flat stage we’re walking across, it’s a deeply folded, multidimensional structure, and what we call “space” is just the slice we can perceive. If that’s true, then interdimensional beings aren’t traveling through our space like we think, they’re navigating the folds. To them, our reality is just one thin layer of a vast, hidden architecture. If you imagine reality as an endless series of roads, most of us can only see the one we’re standing on. The physicists who understand Weinstein’s mathematics, things like bundles, manifolds, and hidden symmetries are basically mapmakers of this larger highway system. The problem? They’re rarely, if ever, in the room when governments or militaries are having “UFO briefings.” The people in those rooms think in terms of craft, propulsion, and reverse engineering, and never about the possibility that the craft aren’t moving at all in the way we imagine, but are simply shifting lanes on a multi-layered road. And that explains why encounters feel so impossible to pin down. To us, these beings appear, vanish, or morph because we’re watching them through the narrow keyhole of our four dimensional perception. But if they can navigate phase space the full cosmic map, they can sidestep into another manifold, bypassing light years in an instant, or even dipping into entirely different law sets of physics. From their point of view, it’s not magic. It’s just the geography of their world. Weinstein’s kind of mathematics never seems to be represented in official discussions should set off alarms. If someone really wanted to understand what we’re seeing, the first people in the room should be those who understand smooth manifolds and the topology of hidden spaces. Instead, it’s like we’re asking mechanics to interpret a language they can’t read. Which makes me wonder, is it incompetence… or intentional? Because if the truth is that we’re not the top layer of reality, that changes everything. So maybe we’re not looking at “aliens” in the traditional sense at all. Maybe we’re dealing with neighbors who’ve been here the whole time, living in the folds, just a turn of the cosmic page away. And maybe, this is the thought that sticks, they don’t show themselves fully because we’re still arguing about the shape of the road, while they’re already mapping the entire atlas. The real disclosure won’t be about the ships… it’ll be about the map itself. And when that map gets opened, history won’t survive the first page.

@SemiAnalysis_ Wonder if there are significant more downloads nowadays?

ONE TRICK SAMSUNG DOESN'T WANT U TO KNOW: With Physical RAM prices increasing rapidly, it is important to take advantage of downloading RAM for free instead of buying physical ram.

@rockymichael630 @bayo_ @KobeissiLetter Agree. Why people keep criticizing the current government. Lets better take sides, team up with it and profit together. I guess those who criticize it prefer to take sides with democrats, China or other foreign governments.

@bayo_ @KobeissiLetter It’s incredibly funny that people don’t take the opportunity to make money when it’s directly in front of you. Instead try and find ways to be mad about making money. Make it make sense.

BREAKING: The Trump Administration is investing $2 billion in quantum computing companies and will receive equity stakes in return, per WSJ. Details include: 1. $1 billion of the package will be awarded to IBM, $IBM 2. Chip maker GlobalFoundries, $GFS, is receiving $375 million in funding 3. The rest of the companies will receive $100 million each, except for startup Diraq, which is slated to get $38 million 4. Multiple other public companies will receive funds including D-Wave Quantum, Rigetti Computing, and Infleqtion Trump's next big bet is on quantum computing.

@dylan522p I thought the current supply bottleneck is energy, right??

Banger take

@benitoz Amazing article! Thank you so much

The optical industry spent six months arguing whether scale-up goes copper or fiber Corning answered Tuesday with a single slide Vera Rubin Ultra: 16 fibers per GPU on the low end. 160 on the high end. 10x range, same chip NVIDIA wired $500M into Corning the day before

@benitoz Amazing analysis!!

Rene Haas just confirmed the Vera CPU thesis on yesterday’s Arm Q4 call. He didn’t mean to His framing: GPUs are reticle-limited. CPUs are not. The ratio shift is happening in core count, not chip count His exact words: “256 Vera CPU chips, 88 cores per chip, a 200-kilowatt liquid-cooled rack designed to sit in a data center adjacent to a Vera Rubin system” That is not a host CPU. That is a dedicated agentic orchestration Two days ago NVIDIA’s own engineers published the receipt. They traced a real 33-minute Claude Code session: 283 inference requests 58 main-agent turns coordinating 225 sub-agent invocations Context grew from 15K to 156K tokens before compaction dropped it to 20K Main agent alone processed ~3.5 million input tokens in the first 40 turns Anthropic’s own number: agentic systems consume up to 15x more tokens than chat. Coding agents sustain 95 to 98 percent prompt cache hit rates. Without caching, costs would be 6x higher This is what’s happening between GPU calls. File reads. Tool invocations. Sub-agent spawns. Compaction. KV cache management. None of it runs on the GPU That’s why 12,000 GPUs need 400,000 CPU cores. The 33-to-1 ratio isn’t a forecast. It’s a measurement NVIDIA states it in the blog directly: this won’t be resolved by adding more compute FLOPs and memory capacity Translation: the GPU-only path is exhausted. The agentic chapter requires a platform, not a chip Their seven-chip answer: Vera Rubin NVL72 —capacity and prefill Vera CPU — tool execution, KV cache offload Groq 3 LPX — SRAM-first decode, low-jitter generation NVLink 6, ConnectX-9, BlueField-4, Spectrum-X — fabric Result they claim: 400+ tokens per second per user on trillion-parameter MoE at 400K context. Vera spec: 88 Olympus cores, 176 threads, 1.8 TB/s NVLink-C2C, 1.2 TB/s LPDDR5X, 227 billion transistors. A 256-CPU rack delivers 45,056 threads and 400 TB of memory One detail nobody is talking about. The blog’s second author was previously Head of Agents at Groq. The third was previously at @GroqInc and Intel. NVIDIA didn’t license the LPX architecture. They absorbed the team that built it Haas isn’t pitching a competing thesis. He’s confirming this one from the other side of the table. Arm data center royalties doubled year-on-year. He expects them to double again Things feel slow right now because we’re between platforms. The speedup ships in H2 2026. The architectural argument is over. Deployment is the only variable left I cover this in The Quiet Architect and The Fourth Piece $arm $NVDA

@SemiAnalysis_ Is i for inference and t for training then?

A common misconception is that TPU v8i must be the training chip because it has two compute dies. Die count is not the relevant metric, what matters is the balance between compute throughput and memory capacity/bandwidth. Reason 1: Memory capacity and bandwidth TPU v8i has 8 stacks of HBM3E 12-Hi versus 6 on TPU v8t, giving it 288 GB of HBM and 8.6 TB/s of memory bandwidth versus 216 GB and 6.5 TB/s on the training chip. This matters because inference decode is memory-bandwidth-bound, not compute-bound. The 8i also carries 384 MB of on-chip SRAM versus 128 MB on the 8t, providing more buffer for KV cache and attention operations. Reason 2: The training chip achieves higher FP4 FLOPs from a single die Despite having two compute dies, TPU v8i achieves only 10.1 PFLOPs at FP4, while the single-die TPU v8t achieves 12.6 PFLOPs. Google designed the 8t's die to be extremely compute-dense, maximizing MXU throughput for training's sustained high arithmetic intensity. This also seems to highlight Google's broader direction, Google is attempting to train with FP4, a regime where the 8t's dense single die excels.