Uploaded September 2026 | Updated September 2026, 1 hour ago
Jensen Huang says AGI has arrived. What can you actually do with it? I explain why I moved from $20 to $200 a month on AI—and how I’m using it to finish work I used to put off.
We look at GPT-6 Astra, Claude Fable and Grok, the next compute expansion, and what increasingly capable agents could mean for your business and daily work.
My approach to “personal AGI” starts with work you already value. What can you create better? What can you delegate? What can you finally get done?
CHAPTERS
00:00 Jensen says AGI is here—your advantage
00:59 GPT-6, Fable and the competition
03:24 What does a completed task cost?
04:33 AI helping researchers build better AI
05:12 B300, Rubin and the next compute expansion
06:43 More reliable agents and longer assignments
08:56 What is AI worth to you?
09:53 The AI productivity ladder
12:27 Your personal AGI: where to start
What’s one task you would delegate to AI this week? Tell me in the comments.
More research and analysis:
NextBigFuture: nextbigfuture.com
Substack: nextbigfuture.substack.com
Substack: AI datacenter + Tesla/SpaceX buildout — physical stack → dollars.
SOURCES
Artificial Analysis: artificialanalysis.ai/articles/benchmarking-gpt-6-astra
Epoch’s Colossus 2 tracker: epoch.ai/data/ai-data-centers/directory/colossus-2
OpenAI research acceleration: openai.com/index/research-acceleration-view-inside-openai
Ramp AI spending: ramp.com/data/ai-index-august-2026
METR task horizons: metr.org/time-horizons
CLARIFICATIONS
Compute buildouts and future task durations discussed here are scenarios. METR horizons measure human task difficulty, not uninterrupted agent runtime. Spending figures alone do not establish productivity returns.
The Navier–Stokes solution claim mentioned in the recording has not been verified and should not be treated as an established result. Navier -stokes solution took thousands of agents working for 88 hours.
#AGI #AIAgents #AIProductivity
Jensen Huang says AGI has arrived. What can you actually do with it? I explain why I moved from $20 to $200 a month on AI—and how I’m using it to finish work I used to put off.
We look at GPT-6 Astra, Claude Fable and Grok, the next compute expansion, and what increasingly capable agents could mean for your business and daily work.
My approach to “personal AGI” starts with work you already value. What can you create better? What can you delegate? What can you finally get done?
CHAPTERS
00:00 Jensen says AGI is here—your advantage
00:59 GPT-6, Fable and the competition
03:24 What does a completed task cost?
04:33 AI helping researchers build better AI
05:12 B300, Rubin and the next compute expansion
06:43 More reliable agents and longer assignments
08:56 What is AI worth to you?
09:53 The AI productivity ladder
12:27 Your personal AGI: where to start
What’s one task you would delegate to AI this week? Tell me in the comments.
More research and analysis:
NextBigFuture: nextbigfuture.com
Substack: nextbigfuture.substack.com
Substack: AI datacenter + Tesla/SpaceX buildout — physical stack → dollars.
SOURCES
Artificial Analysis: artificialanalysis.ai/articles/benchmarking-gpt-6-astra
Epoch’s Colossus 2 tracker: epoch.ai/data/ai-data-centers/directory/colossus-2
OpenAI research acceleration: openai.com/index/research-acceleration-view-inside-openai
Ramp AI spending: ramp.com/data/ai-index-august-2026
METR task horizons: metr.org/time-horizons
CLARIFICATIONS
Compute buildouts and future task durations discussed here are scenarios. METR horizons measure human task difficulty, not uninterrupted agent runtime. Spending figures alone do not establish productivity returns.
The Navier–Stokes solution claim mentioned in the recording has not been verified and should not be treated as an established result. Navier -stokes solution took thousands of agents working for 88 hours.
#AGI #AIAgents #AIProductivity
