China is killing the US on energy. Does that mean they’ll win AGI? — Casey Handmer
Related videos
Chip design from the bottom up – Reiner Pope
Dwarkesh Patel
25.6k views
US Blockade on Iran (& China) Continues: As Does March to War with China in the Asia-Pacific
The New Atlas
15.2k views
US War on Iran & the Wider Dirty War on China: US/Ukrainian Mercenaries In Myanmar
The New Atlas
44.5k views
The Department of War is making a huge mistake.
Dwarkesh Patel
35.5k views
The El Mencho Killing is Crazy! Dozens Dead, US Tourists Stranded, & Cartel Wars in Mexico Expected
Philip DeFranco
339.9k views
What are we scaling?
Dwarkesh Patel
22.3k views
US Beta-Testing China Blockade with Attacks on Russian Energy Exports
The New Atlas
20.8k views
Ilya Sutskever – We're moving from the age of scaling to the age of research
Dwarkesh Patel
55.6k views
Satya Nadella – How Microsoft is preparing for AGI
Dwarkesh Patel
55.1k views
Arm CEO Rene Haas on AI: Nvidia Lessons, Intel’s Decline and the US-China Chip War
All-In Podcast
109.1k views
Solving the AI Energy Singularity Through Solar Dominance
Learn why solar power is positioned to dominate future industrial energy supply for AI, even amidst current natural gas dependency, and how extreme scale resolves logistical hurdles like land access and environmental regulation.
Short Summary
- Hyperscalers currently favor flexible natural gas for rapid data center deployment (near-term expediency).
- Solar power exhibits an astonishing 43% learning rate (cost reduction per doubling of production), leading to inevitable long-term cost dominance.
- Overcoming grid constraints requires massive, localized, off-grid solar power solutions, effectively turning data centers into self-sufficient generation hubs.
- Environmental regulations unnecessarily block fast solar deployment, creating higher friction than physical constraints like land use.
This discussion with Casey Handmer explores the massive energy needs of future AI growth (hundreds of GWs) and argues why solar, despite current reliance on existing infrastructure, will provide the necessary scale. We analyze why current hyperscaler choices prioritize speed (natural gas) over ultimate cost, and map out the extreme industrial mobilization required for a fully solar-powered computational future, including strategies for circumventing current regulatory inertia.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
Chip design from the bottom up – Reiner Pope
Dwarkesh Patel
25.6k views
US Blockade on Iran (& China) Continues: As Does March to War with China in the Asia-Pacific
The New Atlas
15.2k views
US War on Iran & the Wider Dirty War on China: US/Ukrainian Mercenaries In Myanmar
The New Atlas
44.5k views
The Department of War is making a huge mistake.
Dwarkesh Patel
35.5k views
The El Mencho Killing is Crazy! Dozens Dead, US Tourists Stranded, & Cartel Wars in Mexico Expected
Philip DeFranco
339.9k views
What are we scaling?
Dwarkesh Patel
22.3k views
US Beta-Testing China Blockade with Attacks on Russian Energy Exports
The New Atlas
20.8k views
Ilya Sutskever – We're moving from the age of scaling to the age of research
Dwarkesh Patel
55.6k views
Satya Nadella – How Microsoft is preparing for AGI
Dwarkesh Patel
55.1k views
Arm CEO Rene Haas on AI: Nvidia Lessons, Intel’s Decline and the US-China Chip War
All-In Podcast
109.1k views
Top Comments (10)
China has spent a few hundred billion on its high speed rail and the US spent $6 trillion failing in Iraq and Afghanistan. Which was the better capital allocation, Casey?
There is NO way this video isn't sped up 1.5x
This talk has made me realize I am *much* more interested in hearing from experienced people than from smart people.
"with no good mountain ranges to separate them", well the Himalayas :)
"I've played Factorio" spoken like a true engineer
Example of solar in my country. So it's year 2022 and we had legacy hydro generating approx 3 TWh of power and gas plant generating 2,5 TWh a year. We also had solar which generated approx 50 GWh of power a year. War in Ukraine starts, we want to ditch Russian gas and produce electricity using small nuclear modular reactor, none of which exist in actual real life in Western world, which would produce like 3,5 TWh additional power a year. I suggest solar, I say it will be cheaper, faster, more flexible, banks will fund it more eagerly, will not need any price guarantees by government or its involvement. They laugh at me and point at 50 GWh produced a year by solar... production has barely moved in last 10 years. It will not scale, it's expensive, it does not work. But the door is left open for any new energy, government basically deregulates solar, if you have land, you can set up a solar park. What happened? 2023 - 150 GWh produced by solar 2024 - 450 GWh produced by solar 2025 - 1 TWh produced by solar 2026 - will produce 2 TWh using solar 2027 - will climb to 3,5 TWh in solar Meanwhile the nuclear plant plan is said to maybe come online after 2035, no real work on that has been started. Not sure who will finance it seeing the boom in solar, wind and early signs of battery storage in region... I came back to nuclear guys and show all the solar projects in construction phase, but they come back ''you need storage but there is 0 storage and it is expensive, solar can only be solution with storage. 2025... First 50 GWh from storage appear on the grid... It's not that gas or nuclear aren't possible solutions to issues, the thing is that when these projects and big power plants are financed and built as per 2025 specs and plans, we will have the equivalent of Star Trek in how solar can ramp up if paired with storage, and maybe also actual need of energy, how that will skyrocket or not. As the guy said, banks are looking at those projects and they do not know how much electricity will cost by 2035 and what will the world look like. With solar it will be up and preducing in a year or two...
No good mountain ranges? You mean the HIMALAYAS?
Happy Friday everyone !
the biggest hole in all these arguments is when people start referring to "9's" and obviously they're referring to reliability, but what they All (somehow) fail to bring up is time frames. like, once we start talking human life (or simulation) the 9's need to go absurdly deep. ever more problematic is that the "9's" have the be refreshable.
There are brush-corn/sorghum farms in Kansas in the 10,000's of acres. Each.
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
China has spent a few hundred billion on its high speed rail and the US spent $6 trillion failing in Iraq and Afghanistan. Which was the better capital allocation, Casey?
There is NO way this video isn't sped up 1.5x
This talk has made me realize I am *much* more interested in hearing from experienced people than from smart people.
"with no good mountain ranges to separate them", well the Himalayas :)
"I've played Factorio" spoken like a true engineer
Example of solar in my country. So it's year 2022 and we had legacy hydro generating approx 3 TWh of power and gas plant generating 2,5 TWh a year. We also had solar which generated approx 50 GWh of power a year. War in Ukraine starts, we want to ditch Russian gas and produce electricity using small nuclear modular reactor, none of which exist in actual real life in Western world, which would produce like 3,5 TWh additional power a year. I suggest solar, I say it will be cheaper, faster, more flexible, banks will fund it more eagerly, will not need any price guarantees by government or its involvement. They laugh at me and point at 50 GWh produced a year by solar... production has barely moved in last 10 years. It will not scale, it's expensive, it does not work. But the door is left open for any new energy, government basically deregulates solar, if you have land, you can set up a solar park. What happened? 2023 - 150 GWh produced by solar 2024 - 450 GWh produced by solar 2025 - 1 TWh produced by solar 2026 - will produce 2 TWh using solar 2027 - will climb to 3,5 TWh in solar Meanwhile the nuclear plant plan is said to maybe come online after 2035, no real work on that has been started. Not sure who will finance it seeing the boom in solar, wind and early signs of battery storage in region... I came back to nuclear guys and show all the solar projects in construction phase, but they come back ''you need storage but there is 0 storage and it is expensive, solar can only be solution with storage. 2025... First 50 GWh from storage appear on the grid... It's not that gas or nuclear aren't possible solutions to issues, the thing is that when these projects and big power plants are financed and built as per 2025 specs and plans, we will have the equivalent of Star Trek in how solar can ramp up if paired with storage, and maybe also actual need of energy, how that will skyrocket or not. As the guy said, banks are looking at those projects and they do not know how much electricity will cost by 2035 and what will the world look like. With solar it will be up and preducing in a year or two...
No good mountain ranges? You mean the HIMALAYAS?
Happy Friday everyone !
the biggest hole in all these arguments is when people start referring to "9's" and obviously they're referring to reliability, but what they All (somehow) fail to bring up is time frames. like, once we start talking human life (or simulation) the 9's need to go absurdly deep. ever more problematic is that the "9's" have the be refreshable.
There are brush-corn/sorghum farms in Kansas in the 10,000's of acres. Each.