Frontier Scaling Laws and Superintelligence

Oct 27, 2026

The Builders Dario & Daniela Amodei: The Rise of Superintelligent AI, the End of Human Control & Humanity's Fate

• Dario Amodei Dario saw GPT-2’s scaling trajectory as consequential for nations, economies and humanity. He recalls extrapolating that another factor of a million in scale would make systems “better than humans at everything.” His alarm centered …

Oct 25, 2026

十字路口Crossing 10 Key Questions on the PhD Path in the AI Era | A Conversation with 李磊: HKU PhD, Xiaomi MiMo Researcher

• 李磊 李磊 expects rapid progress in RSI on well-defined tasks, but deciding what research to pursue remains the key gap. Better parameters, algorithms and synthetic data can all be found by scaling compute. But researchers still largely decide “which problems …

Oct 21, 2026

Matthew Berman Sam Altman: RSI, AI Consciousness, OpenAI's New Hardware, Compute, and more!

• Sam Altman Sam Altman expects better, cheaper intelligence to keep expanding compute demand, even if shortages ease in the short run. He anticipates supply constraints and gluts, while betting that custom chips, robotic construction and supply-chain work could improve cost …

Oct 20, 2026

The Builders Elon Musk: The End of Work, the Death of Money, and a World Run by AI

• Elon Musk Musk’s largest economic forecast depends on robots and expansion beyond Earth. His “rough guess” is that digital superintelligence increases goods-and-services output by 20%–30%, roughly $30 trillion annually. Robots, …

Oct 17, 2026

All-In CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags, Bill Gates Doomer

• Ben Horowitz The Gates segment: Ben and Chamath connected Bill Gates’s AI doomerism to his past pandemic warnings and speculated about his motives. Ben cited four preconditions for recursive takeoff, attributed to Databricks’ Ali Ghodsi — fewer GPUs per model, less training time, better …

Oct 16, 2026

Altimeter Capital CNBC Halftime Report with Brad Gerstner - October 15th, 2026

• Brad Gerstner His central thesis is that revenue follows model capability, and recursive AI development should accelerate the next wave of products. Opus 4.5 enabled Claude Code, which he credits with unlocking this year’s revenue; machines are now writing “80 or 90% of the code” for …

Oct 15, 2026

Interesting Times Tyler Cowen and Dwarkesh Patel Predict the Future | Interesting Times

• Dwarkesh Patel Patel separates today’s impressive LLMs from AGI, yet still puts human-level cognitive work uncomfortably close. He defines AGI as systems capable of any end-to-end cognitive job, while current models repeatedly arrive like interns on their “first day …
• Tyler Cowen Continual learning is the load-bearing technical assumption behind the forecast. Millions of AI sessions already touch company workflows, but their experience does not persist into the model. Cowen imagines the equivalent …

Oct 14, 2026

The a16z Show Can Open Models Keep Up With OpenAI and Anthropic? | Nathan Lambert

• Nathan Lambert Lambert is skeptical of fast takeoff on current evidence: “all the scaling laws are an exponential increase in compute for a linear increase in performance.” He cites Claude system cards as saying, roughly, that it would take “10× more progress to make this a fast takeoff.” His stated worry is the …

SemiAnalysis Ep. 038 - Is GLM 5.3 a Mythos-Class Cyber Model From China? (InferenceX)

• SemiAnalysis The next gains might come from scaling hybrid attention or post-training for specialized workflows. Kimbo calls GLM-5.3 Flash “almost like a Frankenstein model,” combining DSA with Kimi Delta Attention, and suspects scaling it to GLM-5.3’s …

Oct 12, 2026

Cheeky Pint Muse Wonder: Zuckerberg and Wang's $100B AI

• Cheeky Pint Muse Wonder moves Meta back to the frontier with a larger model optimized for personal and professional agents, not generic benchmark leadership. Meta claims state-of-the-art results across its agent evaluations, DeepSWE and SWE-Marathon, with strong Terminal-Bench 4 performance; its …

The a16z Show Recursive Self-Improvement and the Future of AI

• The a16z Show The founders of an RSI startup—Dhaval Adjodah, who did a PhD in reinforcement learning at MIT Media Lab, and Owen He, formerly a DeepMind research scientist—say they are seeing empirical glimmers of recursive self-improvement: performance improving while the resource denominator stays fixed or falls. Asked whether this is happening, Owen says, “I don’t think it’s just possible. I think it’s happening”; they say results have grown over the …
• Their definitional test distinguishes RSI from “AI creating AI”: ask what the optimizer and the optimizee are. Humans using better AI tools is “autocatalysis”—the ideas and improvement still come from human brains. RSI, they argue, would put the …
• Martin Casado Casado frames capability improvement in terms of performance over cost, but a closed loop alone doesn’t establish takeoff. A fixed numerator leads to a plateau; if it grows with capability, improvement can be constant, and if it grows as a higher power, the model …
• The denominator matters when evaluating RSI claims. Casado notes that model release times are falling, but says the trend looks different relative to capital raised. The founders track …
• Owen He They say many published RSI ideas don’t scale in their tests, and practical bounds remain unclear. Their framework includes human researchers as agents and is intended to generalize across problem domains, though physical processes and the …

Oct 9, 2026

No Priors Beam: The Great American Open Model with ReflectionAI Co-Founder and CEO Misha Laskin

• Misha Laskin The capital required to catch the frontier is rising from hundreds of millions toward tens of billions of dollars, even as model-assisted research improves compute efficiency. Laskin estimates the catch-up cost moved from roughly $100 million a year to 18 months ago to single-digit billions six months ago and …

Oct 8, 2026

The Cognitive Revolution AI:AM: A Level We Shouldn't Pass? Notes from The Curve + Tokens vs. Salaries & Is SaaS Cooked?

• The Cognitive Revolution The Curve’s frontier-lab insiders increasingly treat powerful AI as an operational reality, not a bubble that might pass. One senior lab leader said pretraining still works, synthetic data may improve its economics, and current models may already contain “that …

Oct 7, 2026

Moonshots The Fight Over Claude's Consciousness, AI's 1942 Moment, & Why Altman Says "Accept Some Bad Things"

• Alexander Wissner-Gross Frontier labs are allocating intelligence toward making more intelligence, widening the gap between their internal systems and everything enterprises can license. OpenAI’s Boris Power reportedly said 80%-90% of research targets GPT-7 and GPT-8, while 5.1 and 5.2 are considered short-term bets; …
• Moonshots Washington is treating superintelligence as a geopolitical mobilization, but the panel splits over whether centralized institutions can steer it. Alex calls the new Super Intelligence Force a shift from a “1939” to a “1942 moment,” signaling a Manhattan Project-style race, while Salim …

Oct 5, 2026

Sourcery If an AI Model Can Cheat, It Will | Turing CEO on Reward Hacking

• John Collison The guest rejects both imminent job extinction and the claim that AI is merely hype, favoring a slow takeoff over one or two decades. Recursive self-improvement can optimize today’s training loops, but it cannot yet discover an entirely new outer-loop paradigm beyond …

Oct 3, 2026

Moonshots Recursive's $670M Bet on Self-Improving AI, Sonnet 5.5 Hits 70%, Elon Co-Leads Pentagon Push EP 299

• Richard Socher Recursive’s $670 million bet rests on the claim that weak recursive self-improvement already exists because “AI is code, and AI can code.” Richard Socher says the strong version requires AI to own ideation, implementation, and validation, with an open-ended outer loop that …
• Alexander Wissner-Gross Socher’s strongest definition of ASI is much further away than the programming and mathematics milestones commonly marketed as superintelligence. He expects AI to exceed humanity in specific domains within a few years, but says surpassing “all of humanity combined” across ten forms of …

Oct 2, 2026

All-In Trump’s Super Intelligence Summit, AI Safety Accord, GDP Beats, Midterm Predictions

• All-In The summit’s strategic premise was blunt: “Whoever wins superintelligence wins.” Trump assembled chipmakers, hyperscalers, data-center operators and frontier labs whose leaders often compete or litigate against one …

Oct 1, 2026

Sourcery We’re Reaching the Physical Limits of Chips

• Annie Lamont Ever-larger models may be approaching their practical endpoint. Stephen cites open-source Kimi K3 as having more parameters than the human brain, while the chip needed to operate it is about a million …

Sep 27, 2026

Sourcery 10 of The Wildest Hot Takes From the All-In Summit 2026

• Wen Sang Wen Sang of Genspark argues capability isn’t the bottleneck — access is — so “we won’t slow down.” Echoing Jensen’s point that “AGIs are out of here” but not everywhere, he says coding-agent power users are in the “single millions,” $20 …

The Cognitive Revolution AI:AM: What If It Works Too Well? Colluding Agents, $200M Safety Orgs, Virtual Cells Saturate at 2%

• Prakash Narayanan The closing macro frame: "there's going to be more compute installed over the next 12 months than exists currently in the world." Nathan sees nothing yet suggesting the single biggest teacher model "couldn't just learn it all" — "a bit of a scary beast" — while Prakash …

Sep 23, 2026

The a16z Show Replit CEO Amjad Masad on What Young People Should Learn in the AI Era

• Amjad Masad The most thesis-relevant disclosure: Masad says AI is now capable of doing machine learning itself — “RSI” / “cursive self-improvement,” “vibe research,” or “autoresearch” — and Replit is building a product on it. A late-life chess hobby led him to discover LLMs “actually suck at chess,” to train his first model, and to conclude ML is at the same …

Sep 22, 2026

屠龙之术 AI After Entering Production: Where Will Intelligence, Compute, and Chips Go?

• 屠龙之术 The next generation of intelligence cannot be built by scaling data and compute alone; it must also solve few-shot reasoning, continual learning, and catastrophic forgetting. Zheng Qinghua uses the human brain—approximately 86B neurons, 20-23 watts of power consumption, and roughly 4% of neurons activated …

Sep 21, 2026

Latent Space From InstructGPT to Jev: What Comes After ChatGPT — Diogo Almeida, TypeSafe Co-founder & CEO

• Latent Space He dismisses both the "pace the frontier" consensus and much of the neo-lab wave. Pacing arguments, in his view, assume everyone must do more RLVR with unrestricted action spaces; for Jev's shape, "zero is the optimal …

Sep 20, 2026

Jordi Visser The Market Didn’t Crash, Humanity Didn’t Die — But Crypto Just Broke Out

• Jordi Visser The "why now": Aschenbrenner's automated-AI-researcher inflection from his 2024 *Situational Awareness* paper is arriving early — models "six months ahead" of his timeline — evidenced by Navier-Stokes being solved by 10,000 agents working 88 hours and emerging consumer agents such as Instinct and Meta's Muse. On doom odds he applies expected value: a 10% extinction / 80%-90% extreme-prosperity distribution is "basically all upside" — "if there's …