During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few. From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.

Been running Opus 5.5 agents over the notebooks of Gauss, Riemann, Hilbert and Grothendieck. around 12,000 pages so far. Heres the best finding so far: Riemann's 1860 shock wave paper is famous for getting the energy part wrong however his unpublished sheets at Göttingen have it right the Rankine Hugoniot conditions a decade or two before Rankine and Hugoniot plus results usually dated 19061937, all in his own notation. A second sheet in another folder has the same physics. There are probably thousands of potential undiscovered stuff in the research papers of the smartest ever to live imo all of them should be made public and reviewed by ai years of work can probably be transcribed by agents in days

Why do GB300 GPUs rent for just $12-15/hour when serving DeepSeek can generate $34/hour? Morph's @skeptrune argues the AI infrastructure boom has an overlooked problem: there aren't enough customers to buy all the inference capacity being built. "All of OpenRouter is probably 16 racks of GB300 at most, which is not a lot of compute. There’s tens of thousands of racks available in the world." "If you were to build a 20 megawatt data center and you wanted to just sell the tokens for that, there's less than 1,000 companies in the world that would actually have the appetite to buy tokens from you. You can't sell a gigawatt of compute as inference. There’s not enough demand." "DeepSeek, for example, on a GB300, monetizes at 34 bucks an hour. If there was infinite demand at 34 bucks an hour, then GB300s would sell for that amount. But in practice, they sell for 12 to 15. Sometimes even less." "You can't just turn on a GPU and then find a customer on the other side that’ll let you monetize at 34 bucks an hour. We need more people to build software and nice applications that use these tokens." @morphllm

You have to feel a little bad for Ilya at this point. He, perhaps more than anyone, kicked off the current AI arms race by leaving DeepMind to join OpenAI back in 2015. He chose to stay during the first wave of departures for Anthropic, then launched a failed coup against Sam after realizing that he was not being "consistently candid". This eventually led to him getting kicked out of OpenAI, but instead of joining Anthropic like everyone else, he decided to start his own company that would not pursue product in the meanwhile. Now it's late 2026, and frontier models at both Anthropic and OpenAI are significantly accelerating internal research beyond what any single brilliant researcher can do. The age of heroes is ending. The machines Ilya helped bring into existence are rapidly obsoleting the skills that once gave him such leverage, and power is concentrating in the organizations that can bring industrial might to bear on the problem of how to make AI more powerful. And now Ilya is on the outside, just like the rest of us, watching the singularity happen without him.

Fearless Engineering: The CS-4 is our first engineering documentary. Through lab footage and conversations with the engineers behind each breakthrough, the film follows the creation of the CS-4, our latest wafer-scale AI system. 0:00 introducing the engineering team 1:09 Wafer-scale history 2:28 Powering the wafer 4:00 The cooling problem 4:18 Designing the cold plate 5:11 Holes in the wafer 6:33 the alignment problem 7:47 The CS-4 PRD 8:35 Designing for deployability and serviceability 11:01 Removing the motherboard 11:25 Removing internal cooling loops 11:41 inventing the power rack 12:35 A one-person lab deployment 14:03 Backpack: 50% fewer components 14:55 Connecting the compute backpack to cooling manifold 17:58 shifting Cerebras’s focus from training to inference 19:09 Building the Cerebras Cloud 20:01 CS-3 vs. CS-4 performance 20:49 3D Stacked DRAM

Jensen Huang reportedly instructed his team at Nvidia "never to think about" planning for a Taiwan contingency. Despite the fact that their business is reliant on TSMC. The Jensen-Megaspeed mess really puts these incredible — and barely-noticed — details from Stephen Witt's The Thinking Machine in a new light: Huang refuses to consider the impact on his global business if China were to blockade or invade Taiwan, despite the regular threats emanating from Beijing to do just that. “To my great surprise, Nvidia had done no contingency planning for this eventuality,” Witt writes. He quotes Deb Shoquist, Nvidia’s executive vice-president of operations: “If something happens to Taiwan and TSMC, the ramifications are so large it’s almost like asking me what I’d do if California fell into the ocean.” Huang had instructed her “never to think about this question.” —from my @nybooks essay earlier this year on Nvidia and American policy toward China: —more thoughts from me today:

No surprise that western governments are running short of China experts. People with genuine China expertise in western governments find themselves often regarded as too understanding of Chinese positions. Their expertise is exploited when expedient and more often disregarded as politically inconvenient. They usually will have a boss who is hostile to China, not by accident. Their promotion prospects will tend to depend on confidence that although expert, they are not interested in China and, preferably, hostile to China. Seeing all this happening, younger staff steer clear of making too much investment in acquiring Chinese experience and skills. Anyone with actual recent experience of living in China is likely to be excluded from ability to get a security clearance and so unemployable in government. The result is western governments flying ever blinder and more prejudiced on China.

Imagine if China's top AI executive who single-handedly made the AI revolution possible through tireless pursuit of parallel processors was one day inexplicably disappeared following a lurid sex scandal alleged by intelligence agencies a week after a change in legislative power.

Very important point on sovereign buyers I think we may see a return to the Biden-era AI diffusion rule Democrats will push for it There is bipartisan support as well Trump will be under pressure, and Jensen's lobbying won't be effective

If Anthropic’s valuation drops below $1 trillion in the event of an IPO — I’m going to say it probably shouldn’t go public. There’s been a meaningful amount of volatility in markets. That could last a while. The buy side has been pretty stung, deleveraging meaningfully, with the potential for more. That adds volatility to public equity markets. The most recent jobs data and growth metrics prop up the macro — giving way to another rate hike. Growth is supposed to slow against that backdrop, with liquidity tightening. What’s more, a blue wave could still portend a debt ceiling debate next year that could keep rates elevated in the long end. Conditions are expected to tighten further. Yet companies have been successful in raising capital in smaller increments and keeping spending plans modest. With equity and debt market still very open (the VIX has not hit alarming levels by any stretch), and $4.5 billion in private capital dry powder on the sidelines, they are better off waiting at a value under $1 trillion.

I’m watching Ramayana right now, and it’s remarkable how such a well-known epic has been brought to life on such a grand scale. The vision, effort and presentation deserve real appreciation. Christopher Nolan’s The Odyssey is another ambitious project, bringing Greek mythology to the big screen. But imagine Mahabharata being made with the same scale, vision and cinematic ambition. It could be bigger than Game of Thrones in almost every sense. The depth of its characters, the political intrigue, the philosophical debates, the wars and the sheer scale of its narrative are unmatched. If brought to life with the right vision, it could become one of the greatest cinematic projects in the history of world cinema. Very few epics could even come close.

The story keeps getting crazier. To recap everything that has been reported so far: - The US intelligence community discovered that Jensen Huang, the CEO of the largest company in the world, was sleeping with Alice Huang, a Chinese spy and an executive at Megaspeed. (FT) - Multiple independent investigations have concluded that Megaspeed is likely a Chinese cutout, which helps Chinese companies access AI chips or circumvent export controls. (NYT and Bloomberg) - The US intelligence community also concluded that Alice Huang’s goal was to convince Jensen to lobby the U.S. government to loosen export controls on advanced AI chips. (NYT) - The intelligence community relayed this information to “a small group of top US government officials” earlier this year. (FT) - Over the last year, Jensen Huang has repeatedly lobbied the U.S. government to loosen export controls on advanced chips. @BISgov has done so, not only by permitting the export of H200 chips to China, but also not enforcing export controls that would have required a license for exports of the most advanced Blackwell chips to Megaspeed and Southeast Asia. - Although @CommerceGov leadership was presumably briefed on this information by the U.S. intelligence community, @BISgov has taken no actions to tighten controls on Megaspeed or Southeast Asia. - Megaspeed remains a preferred cloud partner of Nvidia, and continues to receive Nvidia Blackwell chips despite employing a Chinese spy who reportedly is romantically engaged with the CEO of Nvidia. - Tens of billions of dollars worth of AI chips have been exported to Southeast Asia this year alone, and Malaysia and Singapore are now two of the four largest importers of AI chips in the world. What is going on here??? We need answers. The question of who gets the most advanced AI chips in the world is literally the defining strategic question of our time. And not only does it seem like the MSS may be directly influencing the debate on export controls in the United States, but also @CommerceGov is refusing to listen to the warnings of the IC on this issue. What is a POSSIBLE good explanation for why Megaspeed is still receiving chips? Why is there not a license requirement for AI chip shipments to all of Southeast Asia? This is beyond strange at this point. The fallout from this story will be massive. Obviously, the motivations underpinning Nvidia’s policy views opposing export controls must now all be questioned. But also, how did @BISgov sit on this information and has still taken no actions? We need answers.

once a model is produced it’s infinitely replicable and downloadable across the world, to be run on toasters. anyone who is comparing the difficulty of regulating AI to regulating nuclear weapons is selling you a crock of shit