正在加载视频...

视频加载失败

返回首页

Demis Hassabis just explained why the real AI bottleneck has nothing to do with training runs. Most people picture the AI arms race as who can build the biggest model. GPT-4 or Gemini Ultra style training runs, a few hundred million in compute, fired once or twice a year....

aakashgupta's profile picture

Aakash Gupta

302,060 subscribers

32,150 次观看 • 5 个月前 •via X (Twitter)

新闻政治教育科学技术
LIVE

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

Jonathan Ross just revealed why AI companies aren’t growing faster. Not demand. Not competition. Physics. Ross: “The demand for compute is insatiable.” There isn’t enough compute in the world. Not a temporary shortage. A fundamental gap between what the market wants and what the infrastructure can deliver. Ross: “Right now, one of the biggest complaints of Anthropic is the rate limits. People can’t get enough tokens.” Rate limits aren’t product decisions. They’re rationing. Companies forced to regulate access because infrastructure cannot meet demand. Slower services. Token caps. The only things standing between these companies and a revenue surge they can’t access. Every token cap is a revenue cap. Every slowdown is a sale that didn’t happen. Ross: “If Anthropic was given twice the inference compute, within one month their revenue would almost double.” Read that again. Double the compute. Double the revenue. Within thirty days. That’s not a growth projection. That’s a measurement of how deep the backlog already is. The demand exists right now. It’s sitting in a queue. The only thing between these companies and that revenue is physical hardware they don’t have. This breaks every assumption about how tech companies scale. Usually you scale by finding customers. AI companies have infinite customers. They scale by finding hardware. The constraint isn’t market fit. It isn’t distribution. It isn’t competition. It’s processing power. This is why Jensen Huang is the most important person in the world right now. NVIDIA doesn’t just make chips. It makes the thing every government, every AI lab, and every company racing for this future needs more of and can’t get enough of. The compute bottleneck isn’t a tech industry problem. It’s a civilizational one. The winner of this era isn’t determined by who builds the smartest model. Every major lab has a frontier model. The winner is whoever secures the most compute fastest while everyone else rations what’s left. The race isn’t for intelligence. It’s for infrastructure. And right now there isn’t enough to go around.
0:53

Jonathan Ross just revealed why AI companies aren’t growing faster. Not demand. Not competition. Physics. Ross: “The demand for compute is insatiable.” There isn’t enough compute in the world. Not a temporary shortage. A fundamental gap between what the market wants and what the infrastructure can deliver. Ross: “Right now, one of the biggest complaints of Anthropic is the rate limits. People can’t get enough tokens.” Rate limits aren’t product decisions. They’re rationing. Companies forced to regulate access because infrastructure cannot meet demand. Slower services. Token caps. The only things standing between these companies and a revenue surge they can’t access. Every token cap is a revenue cap. Every slowdown is a sale that didn’t happen. Ross: “If Anthropic was given twice the inference compute, within one month their revenue would almost double.” Read that again. Double the compute. Double the revenue. Within thirty days. That’s not a growth projection. That’s a measurement of how deep the backlog already is. The demand exists right now. It’s sitting in a queue. The only thing between these companies and that revenue is physical hardware they don’t have. This breaks every assumption about how tech companies scale. Usually you scale by finding customers. AI companies have infinite customers. They scale by finding hardware. The constraint isn’t market fit. It isn’t distribution. It isn’t competition. It’s processing power. This is why Jensen Huang is the most important person in the world right now. NVIDIA doesn’t just make chips. It makes the thing every government, every AI lab, and every company racing for this future needs more of and can’t get enough of. The compute bottleneck isn’t a tech industry problem. It’s a civilizational one. The winner of this era isn’t determined by who builds the smartest model. Every major lab has a frontier model. The winner is whoever secures the most compute fastest while everyone else rations what’s left. The race isn’t for intelligence. It’s for infrastructure. And right now there isn’t enough to go around.

Dustin

28,395 次观看 • 7 个月前

DeepMind CEO Demis Hassabis says the AI bottleneck is already moving from model ideas to physical supply chains. "How much of the restriction at this point is physical? I am talking memory shortage, GPU shortage, and electricity." Demis: "That is constraining a lot of the deployment for sure. We see so much more demand for our models - Gemini and the other models that we have - than we can possibly serve." Then the less obvious part: "It does constrain a little bit the research, because you need a lot of chips to be able to experiment on new ideas at big enough scale that you can actually see if they are going to work." And on memory: "The whole supply chain is kind of strained... It still in the end actually comes down to a few suppliers of a few of the key components. Anywhere where there is some kind of constraint on the capacity, there is a sort of choke point." Important framing: The next AI race is not just who has the best model architecture. It is who can secure memory, chips, power, suppliers, and enough experimental capacity to find the next breakthrough before everyone else.
1:09

DeepMind CEO Demis Hassabis says the AI bottleneck is already moving from model ideas to physical supply chains. "How much of the restriction at this point is physical? I am talking memory shortage, GPU shortage, and electricity." Demis: "That is constraining a lot of the deployment for sure. We see so much more demand for our models - Gemini and the other models that we have - than we can possibly serve." Then the less obvious part: "It does constrain a little bit the research, because you need a lot of chips to be able to experiment on new ideas at big enough scale that you can actually see if they are going to work." And on memory: "The whole supply chain is kind of strained... It still in the end actually comes down to a few suppliers of a few of the key components. Anywhere where there is some kind of constraint on the capacity, there is a sort of choke point." Important framing: The next AI race is not just who has the best model architecture. It is who can secure memory, chips, power, suppliers, and enough experimental capacity to find the next breakthrough before everyone else.

Karl Mehta

25,716 次观看 • 3 个月前

Elon Musk announced a chip factory 10 times the size of Tesla's Gigafactory. The goal is to produce enough AI compute to equal twice the entire electricity consumption of the United States. He called it the Terafab. Here is the number that stopped me cold. The entire global AI chip industry right now is on track to hit around 100 gigawatts per year of compute. Every Nvidia GPU, every Google TPU, every chip from every company on earth combined. 100 gigawatts. Musk wants one factory to produce a terawatt per year. A terawatt is 1,000 gigawatts. Ten times the output of the entire global industry. From a single building. To put the scale in physical terms, the Terafab would need to be around 100 million square feet. You would need Starship point to point transport just to get from one end to the other. But the reason for the scale is not ambition for its own sake. To launch meaningful AI compute into space, you need a billion chips per year running at a kilowatt each. That is not a number the current industry can produce. The Terafab is the only way to get there. The timeline he put out: a gigawatt of space AI compute annualized by end of next year. Then 10x per year from there. 10 gigawatts by year two. 100 gigawatts by year three. A terawatt beyond that. Most people think orbital data centers are a decade away. Musk is building the factory to make them possible by next year.
3:26

Elon Musk announced a chip factory 10 times the size of Tesla's Gigafactory. The goal is to produce enough AI compute to equal twice the entire electricity consumption of the United States. He called it the Terafab. Here is the number that stopped me cold. The entire global AI chip industry right now is on track to hit around 100 gigawatts per year of compute. Every Nvidia GPU, every Google TPU, every chip from every company on earth combined. 100 gigawatts. Musk wants one factory to produce a terawatt per year. A terawatt is 1,000 gigawatts. Ten times the output of the entire global industry. From a single building. To put the scale in physical terms, the Terafab would need to be around 100 million square feet. You would need Starship point to point transport just to get from one end to the other. But the reason for the scale is not ambition for its own sake. To launch meaningful AI compute into space, you need a billion chips per year running at a kilowatt each. That is not a number the current industry can produce. The Terafab is the only way to get there. The timeline he put out: a gigawatt of space AI compute annualized by end of next year. Then 10x per year from there. 10 gigawatts by year two. 100 gigawatts by year three. A terawatt beyond that. Most people think orbital data centers are a decade away. Musk is building the factory to make them possible by next year.

Ihtesham Ali

318,691 次观看 • 3 个月前

Lisa Su just dropped a massive reality check on the state of the AI hardware scramble, and the sheer scale of what is coming is wild. If you asked her six months ago, demand was just "strong." But now? The entire landscape has shifted into absolute overdrive. She specifically noted that in the last 60 to 90 days, there has been a massive acceleration, especially when looking at the forecasts and demand patterns for 2026. The most insane takeaway is her blunt assessment that "there's not enough compute out there for everything that wants to be done." It is a complete bottleneck of global ambition versus available silicon. What is really flying under the radar here is that this isn't exclusively a GPU story anymore. The desire to ramp compute quickly for heavy AI enterprise workloads is actively pulling massive CPU content right along with it. The demand is practically insatiable across the entire board. The heavyweights are scrambling, the trenches are fighting for allocation, and the hardware supercycle is officially accelerating to a whole new level.
1:19

Lisa Su just dropped a massive reality check on the state of the AI hardware scramble, and the sheer scale of what is coming is wild. If you asked her six months ago, demand was just "strong." But now? The entire landscape has shifted into absolute overdrive. She specifically noted that in the last 60 to 90 days, there has been a massive acceleration, especially when looking at the forecasts and demand patterns for 2026. The most insane takeaway is her blunt assessment that "there's not enough compute out there for everything that wants to be done." It is a complete bottleneck of global ambition versus available silicon. What is really flying under the radar here is that this isn't exclusively a GPU story anymore. The desire to ramp compute quickly for heavy AI enterprise workloads is actively pulling massive CPU content right along with it. The demand is practically insatiable across the entire board. The heavyweights are scrambling, the trenches are fighting for allocation, and the hardware supercycle is officially accelerating to a whole new level.

Ian Miles

100,831 次观看 • 5 个月前

"Compute is the currency of the future." "Compute is the new OIL." Great, so how do I invest? You can play with the NVIDIA stock, or spin up an AI-Ready Data Center. Challenges of Spinning up AI-Ready Compute: Cost: Even a small cluster of H100 machines is upward to $10M GPU Scarcity: Unallocated or “underutilized” H100, A100, and RTX4090 GPUs are almost non-existent. Very Specific Physical & Technical Requirements: A data center with the right infrastructure is also very rare. There are only 10 in North America that deploy thousands of GPUs, in the open market. Domain Expertise: Teams of experts with the years of experience required to operate, deploy, and manage AI compute at scale are as rare as the GPUs they deploy. Exabits checks all of these boxes and we are the premier Compute Tokenization Platform. We are backed by REAL COMPUTE used by REAL CUSTOMERS.
0:30

"Compute is the currency of the future." "Compute is the new OIL." Great, so how do I invest? You can play with the NVIDIA stock, or spin up an AI-Ready Data Center. Challenges of Spinning up AI-Ready Compute: Cost: Even a small cluster of H100 machines is upward to $10M GPU Scarcity: Unallocated or “underutilized” H100, A100, and RTX4090 GPUs are almost non-existent. Very Specific Physical & Technical Requirements: A data center with the right infrastructure is also very rare. There are only 10 in North America that deploy thousands of GPUs, in the open market. Domain Expertise: Teams of experts with the years of experience required to operate, deploy, and manage AI compute at scale are as rare as the GPUs they deploy. Exabits checks all of these boxes and we are the premier Compute Tokenization Platform. We are backed by REAL COMPUTE used by REAL CUSTOMERS.

Exabits

13,896 次观看 • 2 年前

ELON: AI CLUSTERS WILL BECOME A NORMAL THING EVERY COUNTRY HAS Compute is becoming a new currency. While building a "frontier model" today is a high-stakes gauntlet that requires a level of technical skill only a few companies possess, he predicts that AI compute clusters are about to become as essential to a nation as a power grid or a military. By 2026, "AI Sovereignty" will be the only thing keeping nations from becoming digital colonies of Silicon Valley or Beijing: “Probably all countries will have their own AI clusters over time. It is currently very difficult to actually build an AI cluster and have it run... because if you are 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 is just going to be a normal thing that every country has.” Source: vitrupo Elon Musk
1:03

ELON: AI CLUSTERS WILL BECOME A NORMAL THING EVERY COUNTRY HAS Compute is becoming a new currency. While building a "frontier model" today is a high-stakes gauntlet that requires a level of technical skill only a few companies possess, he predicts that AI compute clusters are about to become as essential to a nation as a power grid or a military. By 2026, "AI Sovereignty" will be the only thing keeping nations from becoming digital colonies of Silicon Valley or Beijing: “Probably all countries will have their own AI clusters over time. It is currently very difficult to actually build an AI cluster and have it run... because if you are 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 is just going to be a normal thing that every country has.” Source: vitrupo Elon Musk

Mario Nawfal

20,502 次观看 • 8 个月前

The entire AI boom has hit a physical wall that no chip can break: America does NOT have enough electricity to run what these companies are building. You don't form an emergency coalition around a problem you've solved. You form one around a problem that's beating you. This morning Nvidia, Google, and a startup called Emerald AI launched something called the AI Energy Management Alliance. 18 members, including the AI lab Anthropic, the utility National Grid, and the power producers AES and NRG. The stated mission: Data centers will become "good citizens" of the grid, gently easing their power use in the rare hours the system is stressed, protecting your electricity bill and keeping the lights on for everyone else. But read what Nvidia said underneath the press release: Power has become "the defining constraint" on the expansion of AI infrastructure in the United States. The most valuable company in history, the one selling every chip the boom runs on, just told you the ceiling on its own business is the number of watts the country can physically deliver. So what is this alliance actually for? It's a lobbying group. And its real ask is this: Right now a new data center can wait a decade or more to connect to the grid, because utilities have to build for the worst hour of the year. The alliance wants governors and regulators to let any data center that promises to throttle its power during peak hours skip that line and plug in faster and bigger. And the grid is already buckling, and you're already paying for it. On PJM, the largest grid in America, serving 67 million people across 13 states, the wholesale cost of power jumped 76% in a single year. The independent market monitor pinned it directly on data centers and said the damage "is not reversible." Capacity prices in that market went from about $29 per megawatt-day to $329, roughly 11 times higher, and data centers drove most of it. This summer PJM had to beg households to cut their usage during a heat wave to avoid rolling blackouts. That's the backdrop the word "affordability" is doing so much work against. Now look at what the alliance actually brings to fix it: Google runs a demand-response program of roughly 1 gigawatt. Emerald and Nvidia have run 6 demonstrations. The first genuinely flexible AI data center they can point to is a single 100 megawatt site in Virginia that isn't even switched on yet. That's the pilot they're holding up against a buildout measured in tens of gigawatts, ordered years in advance, and already raising your rates today. So strip the civics off it and here's what actually happened: The best-funded companies on Earth just formed a coalition whose entire premise is that they need permission to sometimes use LESS electricity. They are announcing that the grid ran out of room before the models ran out of ambition, and they would like to jump the queue while you cover the upgrades. This has nothing to do with strength. Every earlier limit on AI was just a number on a slide, whether it was compute or funding or chip supply, and every one of them got solved with more money. This one can't be solved that way. You cannot venture-fund a power plant into existence in a quarter, and you cannot print a transmission line. Electricity is the first ceiling the AI industry can't buy its way through, and today it admitted that while dressing it up as a favor to your utility bill.
1:11

The entire AI boom has hit a physical wall that no chip can break: America does NOT have enough electricity to run what these companies are building. You don't form an emergency coalition around a problem you've solved. You form one around a problem that's beating you. This morning Nvidia, Google, and a startup called Emerald AI launched something called the AI Energy Management Alliance. 18 members, including the AI lab Anthropic, the utility National Grid, and the power producers AES and NRG. The stated mission: Data centers will become "good citizens" of the grid, gently easing their power use in the rare hours the system is stressed, protecting your electricity bill and keeping the lights on for everyone else. But read what Nvidia said underneath the press release: Power has become "the defining constraint" on the expansion of AI infrastructure in the United States. The most valuable company in history, the one selling every chip the boom runs on, just told you the ceiling on its own business is the number of watts the country can physically deliver. So what is this alliance actually for? It's a lobbying group. And its real ask is this: Right now a new data center can wait a decade or more to connect to the grid, because utilities have to build for the worst hour of the year. The alliance wants governors and regulators to let any data center that promises to throttle its power during peak hours skip that line and plug in faster and bigger. And the grid is already buckling, and you're already paying for it. On PJM, the largest grid in America, serving 67 million people across 13 states, the wholesale cost of power jumped 76% in a single year. The independent market monitor pinned it directly on data centers and said the damage "is not reversible." Capacity prices in that market went from about $29 per megawatt-day to $329, roughly 11 times higher, and data centers drove most of it. This summer PJM had to beg households to cut their usage during a heat wave to avoid rolling blackouts. That's the backdrop the word "affordability" is doing so much work against. Now look at what the alliance actually brings to fix it: Google runs a demand-response program of roughly 1 gigawatt. Emerald and Nvidia have run 6 demonstrations. The first genuinely flexible AI data center they can point to is a single 100 megawatt site in Virginia that isn't even switched on yet. That's the pilot they're holding up against a buildout measured in tens of gigawatts, ordered years in advance, and already raising your rates today. So strip the civics off it and here's what actually happened: The best-funded companies on Earth just formed a coalition whose entire premise is that they need permission to sometimes use LESS electricity. They are announcing that the grid ran out of room before the models ran out of ambition, and they would like to jump the queue while you cover the upgrades. This has nothing to do with strength. Every earlier limit on AI was just a number on a slide, whether it was compute or funding or chip supply, and every one of them got solved with more money. This one can't be solved that way. You cannot venture-fund a power plant into existence in a quarter, and you cannot print a transmission line. Electricity is the first ceiling the AI industry can't buy its way through, and today it admitted that while dressing it up as a favor to your utility bill.

Ricardo

26,257 次观看 • 12 天前

The Tesla Megapod trademark might be the most important filing nobody is talking about (Save this). Tesla wants to build a product that is basically a self contained AI computer in a box. You truck it somewhere, plug it into power, and it runs AI workloads. The reason this matters is that the AI industry has a problem because you can't build data centers fast enough. Getting land approved, getting power from the grid, getting zoning cleared, that process takes 3–5 years, and demand is moving much faster than that. US data center construction actually fell in 2025 for the first time in years, because developers kept hitting bureaucratic walls. Tesla's answer to that problem is simple. They already own thousands of Supercharger stations across the country and those stations already have land, power, and grid connections, all the hard stuff is done. The idea is to drop a Megapod at those sites and turn idle charging infrastructure into AI compute, almost overnight. What Jason, Chamath, and Friedberg were all circling around is that this drop a box on a concrete pad approach is already how serious companies are deploying AI compute right now. Vertiv and DUG Technology both ship GPU-packed containers that can go live in 90 days and Tesla would just be bringing that idea to a network it already owns. The opportunity is very real whether Tesla is the one to actually execute on it is still an open question.
4:06

The Tesla Megapod trademark might be the most important filing nobody is talking about (Save this). Tesla wants to build a product that is basically a self contained AI computer in a box. You truck it somewhere, plug it into power, and it runs AI workloads. The reason this matters is that the AI industry has a problem because you can't build data centers fast enough. Getting land approved, getting power from the grid, getting zoning cleared, that process takes 3–5 years, and demand is moving much faster than that. US data center construction actually fell in 2025 for the first time in years, because developers kept hitting bureaucratic walls. Tesla's answer to that problem is simple. They already own thousands of Supercharger stations across the country and those stations already have land, power, and grid connections, all the hard stuff is done. The idea is to drop a Megapod at those sites and turn idle charging infrastructure into AI compute, almost overnight. What Jason, Chamath, and Friedberg were all circling around is that this drop a box on a concrete pad approach is already how serious companies are deploying AI compute right now. Vertiv and DUG Technology both ship GPU-packed containers that can go live in 90 days and Tesla would just be bringing that idea to a network it already owns. The opportunity is very real whether Tesla is the one to actually execute on it is still an open question.

Milk Road AI

16,630 次观看 • 3 个月前

Elon Musk explains how data centers in Space could help the US compete with China in AI “Given that Chinese AI companies are doing as well as they are with a relatively small amount of compute, it seems that if they had a lot of compute, there’s a good chance they would be the leaders. The constraints on AI are essentially electricity and AI chips. China is very strong in digital AI, and it has far more electricity than the United States. In fact, China already has more electricity than the United States, Europe and India combined. The US government can prevent US companies from using Chinese models, but it can’t stop the rest of the world from using them. That won’t stop China from becoming the leader in AI. Digital intelligence comes down to how much AI compute, how many chips and how much electricity you have. The current constraint is electricity outside China. In China, the constraint is chips. China is also closer than most people realize to solving the lithography problem, which would enable it to make AI chips in very large volumes. Once we address the power constraint with AI data centers in space, the constraint will once again be chips outside China.”
4:51

Elon Musk explains how data centers in Space could help the US compete with China in AI “Given that Chinese AI companies are doing as well as they are with a relatively small amount of compute, it seems that if they had a lot of compute, there’s a good chance they would be the leaders. The constraints on AI are essentially electricity and AI chips. China is very strong in digital AI, and it has far more electricity than the United States. In fact, China already has more electricity than the United States, Europe and India combined. The US government can prevent US companies from using Chinese models, but it can’t stop the rest of the world from using them. That won’t stop China from becoming the leader in AI. Digital intelligence comes down to how much AI compute, how many chips and how much electricity you have. The current constraint is electricity outside China. In China, the constraint is chips. China is also closer than most people realize to solving the lithography problem, which would enable it to make AI chips in very large volumes. Once we address the power constraint with AI data centers in space, the constraint will once again be chips outside China.”

DogeDesigner

246,547 次观看 • 2 个月前

The unwritten rule of AI research: prove your architecture at small scale, then earn the compute to scale it. Jerry Tworek thinks that rule may have quietly killed the transformer's competitors. RL showed almost nothing interesting until it crossed a compute threshold – the capabilities simply don't appear below a baseline. If that's true of RL, it's probably true of other architectures too. Which means there could be entire families of ideas sitting in the discard pile, written off before they were ever given enough compute to show anything at all. That's the bet behind Core Automation: run architecture search at the scale where the answers actually show up.
0:41

The unwritten rule of AI research: prove your architecture at small scale, then earn the compute to scale it. Jerry Tworek thinks that rule may have quietly killed the transformer's competitors. RL showed almost nothing interesting until it crossed a compute threshold – the capabilities simply don't appear below a baseline. If that's true of RL, it's probably true of other architectures too. Which means there could be entire families of ideas sitting in the discard pile, written off before they were ever given enough compute to show anything at all. That's the bet behind Core Automation: run architecture search at the scale where the answers actually show up.

Sequoia Capital

26,746 次观看 • 1 个月前

$NBIS is becoming one of the biggest winners in the new AI economy because it can deliver compute efficiently at scale. The number that mattered most to me was Nebius seeing four or more customers competing for every GPU it brings online (huge for AI cloud market). Management also said they raised prices again and are still selling out across both old and new GPU generations which is one of the cleanest pricing power signals you can get. This is a $100B company in the making.
0:44

$NBIS is becoming one of the biggest winners in the new AI economy because it can deliver compute efficiently at scale. The number that mattered most to me was Nebius seeing four or more customers competing for every GPU it brings online (huge for AI cloud market). Management also said they raised prices again and are still selling out across both old and new GPU generations which is one of the cleanest pricing power signals you can get. This is a $100B company in the making.

Shay Boloor

589,512 次观看 • 4 个月前

Andrew Ng just revealed why the AI companies throwing the most compute at the problem are going to lose. The winner of the intelligence race won’t use the most compute. They’ll waste the least. Ng: “Most of your high-dimensional data lies on a lower-dimensional subspace. It’s just a fact of life.” Here’s what that means in practice. You have a 10,000-dimensional dataset. Every dimension dragged through every calculation. Every training cycle hauling dead weight the model will never use. Ng: “You’re carrying around these 10,000-dimensional examples throughout your whole training process.” That bloat isn’t just inefficient. It’s a tax on every computation you run. Memory bandwidth. Network bandwidth. Computational speed. All of it eaten by dimensions that contribute nothing to intelligence. They contribute noise. The insight that separates the architects from the arms race: that 10,000-dimensional dataset is almost entirely captured by a much smaller subspace. The signal lives in a fraction of the space you’re paying to process. Compress it. 10,000 dimensions down to 1,000. Ng: “You can run your learning algorithm on a much lower-dimensional set of data and it may be much more efficient.” Same hardware. Same budget. A fraction of the friction. Brute force is the strategy of whoever has the deepest pockets. Compression is the strategy of whoever actually understands the problem. The companies that master this don’t just build faster models. They build models that find more truth in less data than anything scaling blindly ever will. Intelligence was never about processing everything. It’s about knowing what to cut.
1:14

Andrew Ng just revealed why the AI companies throwing the most compute at the problem are going to lose. The winner of the intelligence race won’t use the most compute. They’ll waste the least. Ng: “Most of your high-dimensional data lies on a lower-dimensional subspace. It’s just a fact of life.” Here’s what that means in practice. You have a 10,000-dimensional dataset. Every dimension dragged through every calculation. Every training cycle hauling dead weight the model will never use. Ng: “You’re carrying around these 10,000-dimensional examples throughout your whole training process.” That bloat isn’t just inefficient. It’s a tax on every computation you run. Memory bandwidth. Network bandwidth. Computational speed. All of it eaten by dimensions that contribute nothing to intelligence. They contribute noise. The insight that separates the architects from the arms race: that 10,000-dimensional dataset is almost entirely captured by a much smaller subspace. The signal lives in a fraction of the space you’re paying to process. Compress it. 10,000 dimensions down to 1,000. Ng: “You can run your learning algorithm on a much lower-dimensional set of data and it may be much more efficient.” Same hardware. Same budget. A fraction of the friction. Brute force is the strategy of whoever has the deepest pockets. Compression is the strategy of whoever actually understands the problem. The companies that master this don’t just build faster models. They build models that find more truth in less data than anything scaling blindly ever will. Intelligence was never about processing everything. It’s about knowing what to cut.

Dustin

215,643 次观看 • 7 个月前

Demis Hassabis confirmed every frontier AI lab is working on recursive self-improvement and in the same sentence said the safety risk of removing humans from the loop entirely keeps him up at night. That combination should stop you. The CEO of Google DeepMind just confirmed that the thing most people treat as a theoretical future risk is already the active focus of every serious lab on earth right now. He explained why it works in coding and math. The feedback loop is fast. You can verify whether an answer is correct almost instantly. You can generate synthetic training data from it. The loop closes quickly and cleanly. Then he said where it breaks down. In biology, chemistry and physics. Any domain where verifying a hypothesis requires a physical experiment in the real world. The loop does not close in seconds. It closes in weeks or months. Geoffrey Hinton said in his Nobel lecture that recursive self-improvement is the development he fears most and that once started it may not be possible to stop. Hassabis is not pushing back on that. He is describing the guardrails labs are building around a process they are already running. Every lab has to think carefully about the safety of a process where no human is in the loop. He said that as a constraint they are navigating right now. The question they are sitting with is how much of it to let run without a human watching. (Watch the full interview on YouTube at Two Minute Papers channel)
2:00

Demis Hassabis confirmed every frontier AI lab is working on recursive self-improvement and in the same sentence said the safety risk of removing humans from the loop entirely keeps him up at night. That combination should stop you. The CEO of Google DeepMind just confirmed that the thing most people treat as a theoretical future risk is already the active focus of every serious lab on earth right now. He explained why it works in coding and math. The feedback loop is fast. You can verify whether an answer is correct almost instantly. You can generate synthetic training data from it. The loop closes quickly and cleanly. Then he said where it breaks down. In biology, chemistry and physics. Any domain where verifying a hypothesis requires a physical experiment in the real world. The loop does not close in seconds. It closes in weeks or months. Geoffrey Hinton said in his Nobel lecture that recursive self-improvement is the development he fears most and that once started it may not be possible to stop. Hassabis is not pushing back on that. He is describing the guardrails labs are building around a process they are already running. Every lab has to think carefully about the safety of a process where no human is in the loop. He said that as a constraint they are navigating right now. The question they are sitting with is how much of it to let run without a human watching. (Watch the full interview on YouTube at Two Minute Papers channel)

Ihtesham Ali

68,231 次观看 • 3 个月前

“It’s about $1 million a megawatt all-in, between infrastructure and compute, to build a Bitcoin mining site. The AI site, just the infrastructure without compute, is $10 to $15 million a megawatt.” $MARA CEO Fred Thiel points out the math of Bitcoin miners pivoting to AI: An AI data center costs 10-15x more per megawatt to build than a Bitcoin mine but it earns far more per unit of power. And because the tenants are companies like Google, banks will actually finance it. The pricier build is the easier one to fund. Don’t miss the full interview streaming now.
5:46

“It’s about $1 million a megawatt all-in, between infrastructure and compute, to build a Bitcoin mining site. The AI site, just the infrastructure without compute, is $10 to $15 million a megawatt.” $MARA CEO Fred Thiel points out the math of Bitcoin miners pivoting to AI: An AI data center costs 10-15x more per megawatt to build than a Bitcoin mine but it earns far more per unit of power. And because the tenants are companies like Google, banks will actually finance it. The pricier build is the easier one to fund. Don’t miss the full interview streaming now.

Natalie Brunell ⚡️

45,375 次观看 • 2 个月前

Greg Brockman, President of OpenAI, said there is not enough compute in the world to satisfy AI demand, and OpenAI itself cannot launch products it has already built because it cannot find the infrastructure to run them (Save this). OpenAI is spending $50 billion on compute in 2026 alone and it still is not enough. That is the setup but here is the trade. Nebius is one of the most asymmetric infrastructure plays in public markets right now, and most people have never heard of it. Q1 2026 revenue came in at $399 million, up 684% year over year, with AI cloud revenue specifically growing 841% in a single quarter. The company entered 2026 with an exit ARR of $1.25 billion and is targeting $7 to $9 billion by year end, a number that would make it one of the fastest revenue ramps in the history of public infrastructure companies. The contracted backlog sits at $50 billion anchored by a $17.4 billion agreement with Microsoft through 2031 and a $27 billion five-year deal with Meta. They are decade-scale infrastructure commitments from the two largest enterprise AI spenders on earth, signed before the demand curve has even reached its steepest point. Nvidia took a direct equity stake in Nebius, one of only two neoclouds it has invested in alongside CoreWeave. That relationship is not just financial but rather means Nebius gets preferential access to GPU allocation at a moment when every lab and every hyperscaler is competing for the same constrained supply. Contracted power capacity now exceeds 3.5 gigawatts, with expansion plans targeting 5 to 6 GW by mid-2029. And power is the other binding constraint in AI infrastructure, you cannot build a data center without it and Nebius has already secured the capacity that competitors are still fighting to acquire. At full ramp, analysts project revenue in the $15 to $25 billion range by 2029, against a current market cap the contracted backlog alone already dwarfs. Come join Milk Road Pro and get our full Nebius deep-dive, the exact price levels we are watching, how we are sizing the position against the backlog and power capacity timeline, and our full AI thesis. link below!
1:00

Greg Brockman, President of OpenAI, said there is not enough compute in the world to satisfy AI demand, and OpenAI itself cannot launch products it has already built because it cannot find the infrastructure to run them (Save this). OpenAI is spending $50 billion on compute in 2026 alone and it still is not enough. That is the setup but here is the trade. Nebius is one of the most asymmetric infrastructure plays in public markets right now, and most people have never heard of it. Q1 2026 revenue came in at $399 million, up 684% year over year, with AI cloud revenue specifically growing 841% in a single quarter. The company entered 2026 with an exit ARR of $1.25 billion and is targeting $7 to $9 billion by year end, a number that would make it one of the fastest revenue ramps in the history of public infrastructure companies. The contracted backlog sits at $50 billion anchored by a $17.4 billion agreement with Microsoft through 2031 and a $27 billion five-year deal with Meta. They are decade-scale infrastructure commitments from the two largest enterprise AI spenders on earth, signed before the demand curve has even reached its steepest point. Nvidia took a direct equity stake in Nebius, one of only two neoclouds it has invested in alongside CoreWeave. That relationship is not just financial but rather means Nebius gets preferential access to GPU allocation at a moment when every lab and every hyperscaler is competing for the same constrained supply. Contracted power capacity now exceeds 3.5 gigawatts, with expansion plans targeting 5 to 6 GW by mid-2029. And power is the other binding constraint in AI infrastructure, you cannot build a data center without it and Nebius has already secured the capacity that competitors are still fighting to acquire. At full ramp, analysts project revenue in the $15 to $25 billion range by 2029, against a current market cap the contracted backlog alone already dwarfs. Come join Milk Road Pro and get our full Nebius deep-dive, the exact price levels we are watching, how we are sizing the position against the backlog and power capacity timeline, and our full AI thesis. link below!

Milk Road AI

14,578 次观看 • 3 个月前

If intelligence is the log of compute… it starts with a lot of compute! And that’s why we’re scaling our GPU fleet faster than anyone else. Just last year, we added over 2 gigawatts of new capacity – roughly the output of 2 nuclear power plants. And today we’re going further, announcing the world's most powerful AI datacenter, located in southeastern Wisconsin. Fairwater is a seamless cluster of hundreds of thousands of NVIDIA GB200s, connected by enough fiber to circle the Earth 4.5 times. It will deliver 10x the performance of the world’s fastest supercomputer today, enabling AI training and inference workloads at a level never before seen. For AI training workloads, you need compute at exponential scale. That’s why we designed the datacenter, GPU fleet, and network together as one integrated system. This ensures a single job can run from day 1 at exponential scale across thousands of GPUs. Fairwater uses a liquid-cooled closed-loop system for cooling GPUs that requires zero water for operations after construction. And we’re matching all of the energy that is consumed with renewable sources. And of course, it is just one of several similar sites we’re lighting up across our 70+ regions. We have multiple identical Fairwater datacenters under construction in other locations across the US, in addition to our AI infrastructure already deployed in over 100 datacenters around the world, powering model training, test-time compute, RL tuning, and real-time inference at global scale. Too often during times like this, people go with the current and only later wonder, how did we get here? With Fairwater, we're charting a new path: doing the hard engineering work, bringing compute, network, and storage into one highly scaled cluster, and designing closed-loop energy systems to meet real-world computing needs. And partnering with local communities to ensure it's thoughtfully done in a way that is sustainable, creates new jobs, and expands opportunity. We are thrilled to see this take hold in Wisconsin, and we are just getting started.
0:21

If intelligence is the log of compute… it starts with a lot of compute! And that’s why we’re scaling our GPU fleet faster than anyone else. Just last year, we added over 2 gigawatts of new capacity – roughly the output of 2 nuclear power plants. And today we’re going further, announcing the world's most powerful AI datacenter, located in southeastern Wisconsin. Fairwater is a seamless cluster of hundreds of thousands of NVIDIA GB200s, connected by enough fiber to circle the Earth 4.5 times. It will deliver 10x the performance of the world’s fastest supercomputer today, enabling AI training and inference workloads at a level never before seen. For AI training workloads, you need compute at exponential scale. That’s why we designed the datacenter, GPU fleet, and network together as one integrated system. This ensures a single job can run from day 1 at exponential scale across thousands of GPUs. Fairwater uses a liquid-cooled closed-loop system for cooling GPUs that requires zero water for operations after construction. And we’re matching all of the energy that is consumed with renewable sources. And of course, it is just one of several similar sites we’re lighting up across our 70+ regions. We have multiple identical Fairwater datacenters under construction in other locations across the US, in addition to our AI infrastructure already deployed in over 100 datacenters around the world, powering model training, test-time compute, RL tuning, and real-time inference at global scale. Too often during times like this, people go with the current and only later wonder, how did we get here? With Fairwater, we're charting a new path: doing the hard engineering work, bringing compute, network, and storage into one highly scaled cluster, and designing closed-loop energy systems to meet real-world computing needs. And partnering with local communities to ensure it's thoughtfully done in a way that is sustainable, creates new jobs, and expands opportunity. We are thrilled to see this take hold in Wisconsin, and we are just getting started.

Satya Nadella

2,026,666 次观看 • 1 年前

Mark Zuckerberg says he is not an AI researcher -- so here is the one number he obsesses over to win the superintelligence race: "I'm not an AI researcher. So it's not like I'm in there telling them what research ideas to do." "The things that I can do as CEO of the company are: I can make sure that we have the very best people and talent density. I can make sure that we have by far the highest compute per researcher, and that we do whatever it takes to go build out all that capacity." "We're building this Prometheus cluster, which is going to be the first gigawatt-plus single cluster, a contiguous cluster for training, coming online next year." "We're building a cluster in Louisiana that is going to be 5 gigawatt, and we're building several more clusters like that that are just going to be multiple gigawatts." "We're talking about many hundreds of billions of dollars of capital. So you both need to have a good business model that can support it, and you need to believe in it." The tell: the richest lab in the world is not optimizing headcount. It is optimizing compute-per-researcher, and betting hundreds of billions that whoever owns the biggest contiguous cluster owns the frontier. Talent gets you into the room. Watts decide who leaves with the model.
1:06

Mark Zuckerberg says he is not an AI researcher -- so here is the one number he obsesses over to win the superintelligence race: "I'm not an AI researcher. So it's not like I'm in there telling them what research ideas to do." "The things that I can do as CEO of the company are: I can make sure that we have the very best people and talent density. I can make sure that we have by far the highest compute per researcher, and that we do whatever it takes to go build out all that capacity." "We're building this Prometheus cluster, which is going to be the first gigawatt-plus single cluster, a contiguous cluster for training, coming online next year." "We're building a cluster in Louisiana that is going to be 5 gigawatt, and we're building several more clusters like that that are just going to be multiple gigawatts." "We're talking about many hundreds of billions of dollars of capital. So you both need to have a good business model that can support it, and you need to believe in it." The tell: the richest lab in the world is not optimizing headcount. It is optimizing compute-per-researcher, and betting hundreds of billions that whoever owns the biggest contiguous cluster owns the frontier. Talent gets you into the room. Watts decide who leaves with the model.

Karl Mehta

255,725 次观看 • 2 个月前

Sam Altman says the AI infrastructure race is not about bigger chatbots. It is about shifting the planet's intellectual output from human brains to AI brains. "If people knew what we could do with compute, they would want way, way more." "The thing I am personally most excited about is to use AI and lots of compute to discover new science." "A recent cool example here is we built the Sora Android app using Codex... in less than a month." "You can imagine that way much further, where entire companies can build their products using lots of compute." Then he gives the scale: "Let's say that an AI company today might be generating something on the order of 10 trillion tokens a day out of frontier models." "We're going to have these models at a company be outputting more tokens per day than all of humanity put together, and then 10 times that, and then 100 times that." The bottleneck is moving from software talent to compute supply, inference speed, product integration, and where all that machine intelligence gets pointed.
3:31

Sam Altman says the AI infrastructure race is not about bigger chatbots. It is about shifting the planet's intellectual output from human brains to AI brains. "If people knew what we could do with compute, they would want way, way more." "The thing I am personally most excited about is to use AI and lots of compute to discover new science." "A recent cool example here is we built the Sora Android app using Codex... in less than a month." "You can imagine that way much further, where entire companies can build their products using lots of compute." Then he gives the scale: "Let's say that an AI company today might be generating something on the order of 10 trillion tokens a day out of frontier models." "We're going to have these models at a company be outputting more tokens per day than all of humanity put together, and then 10 times that, and then 100 times that." The bottleneck is moving from software talent to compute supply, inference speed, product integration, and where all that machine intelligence gets pointed.

Karl Mehta

16,619 次观看 • 3 个月前

HRT, one of the largest quantitative trading firms in the world, just revealed the scale of their AI operation. "two of the biggest ingredients you need to build deep learning models are compute and data. HRT has both in incredible orders of magnitude." "we have vast quantities of data, on the order of petabytes, just in terms of raw tick data we've written down since the beginning of electronic trading." "we have a very high compute-per-researcher ratio." then the line that puts it in perspective: "there probably aren't too many places on Earth you can do cutting-edge deep learning research at the level we do it, just because of how few places have access to as much compute." "it's basically the hyperscalers of the world and then we're in the next tranche." a trading firm putting itself in the same compute tier as Google, Microsoft, and Meta. not for chatbots, for trading.
2:08

HRT, one of the largest quantitative trading firms in the world, just revealed the scale of their AI operation. "two of the biggest ingredients you need to build deep learning models are compute and data. HRT has both in incredible orders of magnitude." "we have vast quantities of data, on the order of petabytes, just in terms of raw tick data we've written down since the beginning of electronic trading." "we have a very high compute-per-researcher ratio." then the line that puts it in perspective: "there probably aren't too many places on Earth you can do cutting-edge deep learning research at the level we do it, just because of how few places have access to as much compute." "it's basically the hyperscalers of the world and then we're in the next tranche." a trading firm putting itself in the same compute tier as Google, Microsoft, and Meta. not for chatbots, for trading.

Goshawk Trades

14,780 次观看 • 3 个月前