正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

$ASML CEO warns that without major gains in power efficiency, training frontier AI models could eventually consume the world’s energy supply. If scaling laws hold and no breakthroughs emerge, models trained in 2027 may require $100B+ compute clusters just to run.

Shay Boloor

418,363 subscribers

290,369 次观看 • 4 个月前 •via X (Twitter)

新闻政治科学技术财经

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

OpenAI CEO, Sam Altman, questions: how far could classical pre-trained models go if given unlimited resources but no new algorithmic breakthroughs? could they potentially reach something like a "GPT-5.5"? oAI employee: based purely on current ML and algorithmic understanding, no clear limit has yet been found for scaling classical pre-trained AI models.

OpenAI CEO, Sam Altman, questions: how far could classical pre-trained models go if given unlimited resources but no new algorithmic breakthroughs? could they potentially reach something like a "GPT-5.5"? oAI employee: based purely on current ML and algorithmic understanding, no clear limit has yet been found for scaling classical pre-trained AI models.

Haider.

102,958 次观看 • 1 年前

Sam Altman "Internally, we've reached GPT-4.5, and getting to GPT-5.5 won't require 100x more compute" - progress in reasoning models and RL techniques have greatly improved compute efficiency - it allows smaller models like internal GPT-4.5 to achieve GPT-6 level performance without requiring 100x more computing power

Sam Altman "Internally, we've reached GPT-4.5, and getting to GPT-5.5 won't require 100x more compute" - progress in reasoning models and RL techniques have greatly improved compute efficiency - it allows smaller models like internal GPT-4.5 to achieve GPT-6 level performance without requiring 100x more computing power

Haider.

344,581 次观看 • 1 年前

"It all comes down to compute." Dmitri Alperovitch from Silverado Policy Accelerator explains: "The single most important input to winning is compute the processing power used to train and run AI models. Let me say that plainly, because it is defining that everything that congress and this committee do on AI policy. The binding constraint in this competition is not talent, it is not data, and it is not cash. Both the United States and China have talented researchers and financial resources, and we have learned that frontier models trained on generic data routinely outperform smaller models, trained on more specialized inputs. It all comes down to compute."

"It all comes down to compute." Dmitri Alperovitch from Silverado Policy Accelerator explains: "The single most important input to winning is compute the processing power used to train and run AI models. Let me say that plainly, because it is defining that everything that congress and this committee do on AI policy. The binding constraint in this competition is not talent, it is not data, and it is not cash. Both the United States and China have talented researchers and financial resources, and we have learned that frontier models trained on generic data routinely outperform smaller models, trained on more specialized inputs. It all comes down to compute."

Select Committee on China

41,507 次观看 • 2 个月前

Frontier models that use reasoning to "think" during inference are generating 5X more AI tokens per year. ✨ "Inference is now a thinking process. And in order to teach AI how to think, reinforcement learning and very significant computation was introduced into post-training," said NVIDIA CEO Jensen Huang in a recent keynote at #CES2026. Reinforcement learning is increasing computation demands across all AI scaling laws: pre-training, post-training, and test-time scaling. Learn more about reinforcement learning and AI scaling ➡️

Frontier models that use reasoning to "think" during inference are generating 5X more AI tokens per year. ✨ "Inference is now a thinking process. And in order to teach AI how to think, reinforcement learning and very significant computation was introduced into post-training," said NVIDIA CEO Jensen Huang in a recent keynote at #CES2026. Reinforcement learning is increasing computation demands across all AI scaling laws: pre-training, post-training, and test-time scaling. Learn more about reinforcement learning and AI scaling ➡️

NVIDIA AI Infrastructure

13,329 次观看 • 5 个月前

“I don’t like calling them scaling laws. They're not actual laws.” Illia (root.near) (🇺🇦, ⋈), co-author of Attention Is All You Need and co-founder of NEAR Protocol, joined our CEO Omer Goldberg to discuss why future AI breakthroughs may come from improved training and formal reasoning rather than larger models.

“I don’t like calling them scaling laws. They're not actual laws.” Illia (root.near) (🇺🇦, ⋈), co-author of Attention Is All You Need and co-founder of NEAR Protocol, joined our CEO Omer Goldberg to discuss why future AI breakthroughs may come from improved training and formal reasoning rather than larger models.

Chaos Labs

67,013 次观看 • 11 个月前

Microsoft CEO Satya Nadella says the AI scaling laws multiply intelligence in both compute and algorithmic dimensions: for every 10x increase in compute, capability increases 100x

Microsoft CEO Satya Nadella says the AI scaling laws multiply intelligence in both compute and algorithmic dimensions: for every 10x increase in compute, capability increases 100x

Tsarathustra

80,676 次观看 • 1 年前

“We want to orchestrate the world’s compute, in the same way Bitcoin did, but to train models that can rival ChatGPT” crux on the This Week in Startups podcast, sharing the motivations behind IOTA ・ SN9: to train frontier models efficiently and at scale, in an age where CapEx for centralized training is reaching unsustainable highs.

“We want to orchestrate the world’s compute, in the same way Bitcoin did, but to train models that can rival ChatGPT” crux on the This Week in Startups podcast, sharing the motivations behind IOTA ・ SN9: to train frontier models efficiently and at scale, in an age where CapEx for centralized training is reaching unsustainable highs.

Macrocosmos

14,467 次观看 • 2 个月前

Elon Musk: In the future, every country will have its own AI clusters. “It's currently very difficult to actually build an AI cluster and have it run. It's not like just pulling a computer out of a box. They are currently very difficult to run. And are you going to be training a frontier model? Because if you're training a frontier model, then you need a massive amount of compute and a level of technical skill that only a few companies possess. But, over time, I think every country will have AI compute clusters. It's just going to be a normal thing.” Future Investment Initiative, October 29, 2024

Elon Musk: In the future, every country will have its own AI clusters. “It's currently very difficult to actually build an AI cluster and have it run. It's not like just pulling a computer out of a box. They are currently very difficult to run. And are you going to be training a frontier model? Because if you're training a frontier model, then you need a massive amount of compute and a level of technical skill that only a few companies possess. But, over time, I think every country will have AI compute clusters. It's just going to be a normal thing.” Future Investment Initiative, October 29, 2024

ELON CLIPS

95,313 次观看 • 7 个月前

So what Warren Pies is saying: Anthropic’s Mythos rollout triggered a sharp collapse in $NVDA GPU availability, with supply effectively tightening to near zero as demand for frontier models accelerated. Q1 2026 revenue from Anthropic reinforces that this isn’t theoretical demand, it’s real dollars chasing compute at scale across the AI ecosystem, from $GOOGL GCP to $AMZN AWS. The takeaway is simple: better models require more compute, and that feedback loop continues to drive the entire AI infrastructure trade.

So what Warren Pies is saying: Anthropic’s Mythos rollout triggered a sharp collapse in $NVDA GPU availability, with supply effectively tightening to near zero as demand for frontier models accelerated. Q1 2026 revenue from Anthropic reinforces that this isn’t theoretical demand, it’s real dollars chasing compute at scale across the AI ecosystem, from $GOOGL GCP to $AMZN AWS. The takeaway is simple: better models require more compute, and that feedback loop continues to drive the entire AI infrastructure trade.

Sam Badawi

21,184 次观看 • 1 个月前

When your AI models dream of big compute, they dream of this. 🚀 Introducing Amazon EC2 P6e-GB200 UltraServers, purpose-built for training & deploying frontier #AI models. Big dreams demand even bigger computing power. #AWS 👉

When your AI models dream of big compute, they dream of this. 🚀 Introducing Amazon EC2 P6e-GB200 UltraServers, purpose-built for training & deploying frontier #AI models. Big dreams demand even bigger computing power. #AWS 👉

Amazon Web Services

35,909 次观看 • 11 个月前

There is no limit: Dario Amodei says that even without better algorithms, scaling will continue to lead to amazing improvements in AI models

There is no limit: Dario Amodei says that even without better algorithms, scaling will continue to lead to amazing improvements in AI models

Tsarathustra

157,362 次观看 • 2 年前

Scale CEO Alexandr Wang says AI has "genuinely hit a wall" in terms of pre-training, but there is still progress in AI with evals climbing and models getting smarter due to post-training and test-time compute

Scale CEO Alexandr Wang says AI has "genuinely hit a wall" in terms of pre-training, but there is still progress in AI with evals climbing and models getting smarter due to post-training and test-time compute

Tsarathustra

193,767 次观看 • 1 年前

ASML CEO Christophe Fouquet describes the next evolution of AI semiconductors, telling Tom Mackenzie that chipmakers are looking to increase transistor density and energy efficiency

ASML CEO Christophe Fouquet describes the next evolution of AI semiconductors, telling Tom Mackenzie that chipmakers are looking to increase transistor density and energy efficiency

Bloomberg

12,430 次观看 • 6 个月前

What could scaling unlock for biology? Introducing ProGen3- our next AI foundation models for protein generation. We develop compute-optimal scaling laws up to 46B parameters on 1.5T tokens with real evidence in the wet lab. +we solve a new set of challenges for drug discovery

What could scaling unlock for biology? Introducing ProGen3- our next AI foundation models for protein generation. We develop compute-optimal scaling laws up to 46B parameters on 1.5T tokens with real evidence in the wet lab. +we solve a new set of challenges for drug discovery

Ali Madani

80,638 次观看 • 1 年前

Anthropic co-founder Jared Kaplan on scaling and the road to human-level AI. At AI Startup School in San Francisco. 00:17 - From Physics to AI 01:41 - Initial Skepticism and Shift to AI 02:12 - AI Training Phases 02:32 - Pre-Training 03:16 - Reinforcement Learning 04:19 - Scaling Laws in Training 08:19 - Unlocking AI Capabilities 11:27 - Organizational Knowledge and Memory 12:19 - Oversight and Nuanced Tasks 13:38 - Preparing for the Future 15:48 - Claude 4 and Beyond 21:18 - Human-AI Collaboration 29:50 - Scaling Laws and Compute Efficiency 35:26 - Audience Q&A

Anthropic co-founder Jared Kaplan on scaling and the road to human-level AI. At AI Startup School in San Francisco. 00:17 - From Physics to AI 01:41 - Initial Skepticism and Shift to AI 02:12 - AI Training Phases 02:32 - Pre-Training 03:16 - Reinforcement Learning 04:19 - Scaling Laws in Training 08:19 - Unlocking AI Capabilities 11:27 - Organizational Knowledge and Memory 12:19 - Oversight and Nuanced Tasks 13:38 - Preparing for the Future 15:48 - Claude 4 and Beyond 21:18 - Human-AI Collaboration 29:50 - Scaling Laws and Compute Efficiency 35:26 - Audience Q&A

Y Combinator

116,444 次观看 • 11 个月前

OpenAI's Noam Brown says letting AI models think for 20 seconds got the same boost in performance as scaling up the model by 100,000x and training it for 100,000 times longer, and scaling model size and compute will end up being a footnote compared to scaling model thinking

OpenAI's Noam Brown says letting AI models think for 20 seconds got the same boost in performance as scaling up the model by 100,000x and training it for 100,000 times longer, and scaling model size and compute will end up being a footnote compared to scaling model thinking

Tsarathustra

273,500 次观看 • 1 年前

$A crypto project actually trained a 72B parameter AI model from scratch using decentralized GPU compute. Not fine-tuned, not a wrapper: trained from zero. The model benchmarks competitively against Meta's LLaMA 3 on reasoning tasks, and the entire training run cost a fraction of what centralized labs spend. If decentralized compute can produce frontier-class models, the moat around OpenAI and Anthropic is thinner than people think.$

A crypto project actually trained a 72B parameter AI model from scratch using decentralized GPU compute. Not fine-tuned, not a wrapper: trained from zero. The model benchmarks competitively against Meta's LLaMA 3 on reasoning tasks, and the entire training run cost a fraction of what centralized labs spend. If decentralized compute can produce frontier-class models, the moat around OpenAI and Anthropic is thinner than people think.

VirtualBacon

40,543 次观看 • 2 个月前