Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

"83% want open source models but can't access the infra" io.net CMO Jack Collier exposed AI's broken infrastructure at SuperAI last week. Watch his keynote to find out: 🏛️ Why 3 companies control 65% of all compute 💤 How 90% of global GPUs sit idle 🚀 The DePIN solution... show more

io.net

437,488 subscribers

15,184 просмотров • 1 год назад •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

Комментарии: 11

Фото профиля Leonard McDoogan

Leonard McDoogan1 год назад

@superai_conf

Фото профиля Mobile Scanner

Mobile Scanner1 год назад

Scan any documents, convert images into text, PDF files, etc. 👍

Фото профиля slymnogunc

slymnogunc1 год назад

Get ready to hear the truth. Jack Collier didn’t just give a talk he exposed how centralization is choking the progress of AI. 🔒 In a world where 3 companies control 65% of all compute, 🛑 And 90% of global GPUs sit idle, 🌍 83% of people want open-source models but can’t access the infrastructure. @ionet is breaking those chains with the DePIN revolution already processing over 1 billion tokens per week. The future isn’t centralized, it’s decentralized! ⚡

Фото профиля DDAY119 | io.net & O.XYZ

DDAY119 | io.net & O.XYZ1 год назад

@superai_conf Jack Collier nailed it at @superai_conf 👏 AI’s infrastructure is broken concentrated, inefficient, and gatekept. Enter @ionet and DePIN real solutions, real scale already processing 1B+ tokens/week. Open source AI needs open access. This is how we unlock it.

Фото профиля Ugara.lvl🆙 | $OI | 🟠 GAIB | (Ø,G)

Ugara.lvl🆙 | $OI | 🟠 GAIB | (Ø,G)1 год назад

@superai_conf #ionet

Фото профиля kua ku

kua ku1 год назад

@superai_conf 🔥🔥

Фото профиля kookoo LE

kookoo LE1 год назад

@superai_conf 🔥🔥🔥

Фото профиля GUWEI

GUWEI1 год назад

@superai_conf 🔥🔥🔥

Фото профиля Kiyansh_79721

Kiyansh_797211 год назад

@superai_conf 🚀🚀🚀

Фото профиля KHALEDJUVE ⭕️

KHALEDJUVE ⭕️1 год назад

Jack Collier Thank you for highlighting these critical stats 65% of compute held by just three players and 90% of GPUs idle is eye-opening. As a Web3 & AI enthusiast, I’m excited by IONET’s DePIN approach handling 1B+ tokens/week. How can the community help drive open-source infra-adoption?

Фото профиля jatin.jt. I O.XYZ

jatin.jt. I O.XYZ1 год назад

@superai_conf LFG 🚀

Похожие видео

"65% of global compute is controlled by just three providers — and the more we hand over compute power and data to them, the more we risk building an AI future owned by them, not by us." - Jack Collier (Jack Collier), Chief Growth & Marketing Manager at io.net Jack Collier reveals a critical paradox: open source AI models are ready to transform industries, democratize innovation, and empower creators, yet adoption is being throttled by centralized infrastructure, hidden costs, and complex workflows. Standing on the RAISE 2025 Stage, he breaks down the myths that Open AI is too expensive, too weak, or too complicated — and showcases how distributed compute, unified open platforms, and training-as-a-service are not only solving these challenges but also creating a future where AI is accessible, fair, and owned by the many, not the few. Watch now: #RAISE2025 #OpenSourceAI #AIInfrastructure #FutureOfAI #RAISESummit #DistributedCompute #Innovation

"65% of global compute is controlled by just three providers — and the more we hand over compute power and data to them, the more we risk building an AI future owned by them, not by us." - Jack Collier (Jack Collier), Chief Growth & Marketing Manager at io.net Jack Collier reveals a critical paradox: open source AI models are ready to transform industries, democratize innovation, and empower creators, yet adoption is being throttled by centralized infrastructure, hidden costs, and complex workflows. Standing on the RAISE 2025 Stage, he breaks down the myths that Open AI is too expensive, too weak, or too complicated — and showcases how distributed compute, unified open platforms, and training-as-a-service are not only solving these challenges but also creating a future where AI is accessible, fair, and owned by the many, not the few. Watch now: #RAISE2025 #OpenSourceAI #AIInfrastructure #FutureOfAI #RAISESummit #DistributedCompute #Innovation

RAISE Summit

12,746 просмотров • 9 месяцев назад

Last week, Anthropic said its new model Mythos had already found “thousands of high-severity vulnerabilities” and could “surpass all but the most skilled humans at finding and exploiting software vulnerabilities.” At almost the same moment, the Axios hack showed how exposed the open-source foundation under modern software still is. For this week’s 10 Minutes or Less, I spoke with Feross what this moment means. WATCH:

Last week, Anthropic said its new model Mythos had already found “thousands of high-severity vulnerabilities” and could “surpass all but the most skilled humans at finding and exploiting software vulnerabilities.” At almost the same moment, the Axios hack showed how exposed the open-source foundation under modern software still is. For this week’s 10 Minutes or Less, I spoke with Feross what this moment means. WATCH:

Ali Rohde

13,126 просмотров • 3 месяцев назад

When we first met Tom, his obsession with energy markets came through immediately. Seemed random, but it's "actually" central to what he's building at Actual Computer. The thesis is that local inference is inevitable. Open source models are getting good enough, people don't want to feed their thoughts to AI labs, and demand for compute will continue to be insatiable. But the main forcing function is that chips aren't the bottleneck anymore - power is. Compute is derivative of energy, and the massive data center buildout has created an energy shortage where GPUs sit idle waiting for power to come online. Yet one place that already has power, and always will, is everywhere else. Homes. Businesses. Universities. Collectively, they have more compute than all the data centers combined. Tom Lynch goal is to build a distributed network of people plugging in their local compute (devices like Mac minis to consumer Nvidia cards) to serve inference demand. The substrate he's chosen is Bittensor, on Subnet 95. Actual Computer has an exceptional team behind it, and this is one of the projects that's gotten me most excited in a long time. Enjoy this conversation with the two of us!

When we first met Tom, his obsession with energy markets came through immediately. Seemed random, but it's "actually" central to what he's building at Actual Computer. The thesis is that local inference is inevitable. Open source models are getting good enough, people don't want to feed their thoughts to AI labs, and demand for compute will continue to be insatiable. But the main forcing function is that chips aren't the bottleneck anymore - power is. Compute is derivative of energy, and the massive data center buildout has created an energy shortage where GPUs sit idle waiting for power to come online. Yet one place that already has power, and always will, is everywhere else. Homes. Businesses. Universities. Collectively, they have more compute than all the data centers combined. Tom Lynch goal is to build a distributed network of people plugging in their local compute (devices like Mac minis to consumer Nvidia cards) to serve inference demand. The substrate he's chosen is Bittensor, on Subnet 95. Actual Computer has an exceptional team behind it, and this is one of the projects that's gotten me most excited in a long time. Enjoy this conversation with the two of us!

Sami Kassab

35,706 просмотров • 6 месяцев назад

Introducing Open Rates I managed to sell over $250,000+ of ads in 60 days. Not many people can do that. But now YOU can. Open Rates is a tool i've created with that's here to help you: 👉 Find and close companies actively advertising. 👉 Access a real-time database of 10,000+ sponsors. 👉 Find contact information and close deals. Oh and we're already used by your favorite newsletters like The MilkRoad, The Daily Upside, 1440 and more. Last week someone already closed a $20,000+ deal because of our tool. Start earning more today and get your exclusive 3 day free trial only now at openrates . co

Introducing Open Rates I managed to sell over $250,000+ of ads in 60 days. Not many people can do that. But now YOU can. Open Rates is a tool i've created with that's here to help you: 👉 Find and close companies actively advertising. 👉 Access a real-time database of 10,000+ sponsors. 👉 Find contact information and close deals. Oh and we're already used by your favorite newsletters like The MilkRoad, The Daily Upside, 1440 and more. Last week someone already closed a $20,000+ deal because of our tool. Start earning more today and get your exclusive 3 day free trial only now at openrates . co

Wouter Teunissen

51,728 просмотров • 2 лет назад

Nvidia CEO Jensen Huang on why engineers will soon be paid in tokens, not just salary: Jensen lays out a future where compute access becomes part of an engineer's compensation package. "I could totally imagine in the future every single engineer in our company will need an annual token budget," he says. He explains how the math would work: "They're going to make a few hundred thousand a year their base pay. I'm going to give them probably half of that on top of it as tokens so that they could be amplified 10x. Of course, we would." According to Huang, this is already changing how companies compete for talent: "It is now one of the recruiting tools in Silicon Valley. How many tokens comes along with my job?" His reasoning is simple: tokens make engineers more productive. As he puts it, "every engineer that has access to tokens will be more productive and those tokens as you know will be produced by AI factories that all of you and us we partner to build." Huang then zooms out to describe how this reshapes the nature of companies themselves: "Every single enterprise company in today sit on top of file systems and data centers. Every single software company of the future will be agentic and they will be token manufacturers. They be token users for their engineers and they'll be token manufacturers for all of their customers."

Nvidia CEO Jensen Huang on why engineers will soon be paid in tokens, not just salary: Jensen lays out a future where compute access becomes part of an engineer's compensation package. "I could totally imagine in the future every single engineer in our company will need an annual token budget," he says. He explains how the math would work: "They're going to make a few hundred thousand a year their base pay. I'm going to give them probably half of that on top of it as tokens so that they could be amplified 10x. Of course, we would." According to Huang, this is already changing how companies compete for talent: "It is now one of the recruiting tools in Silicon Valley. How many tokens comes along with my job?" His reasoning is simple: tokens make engineers more productive. As he puts it, "every engineer that has access to tokens will be more productive and those tokens as you know will be produced by AI factories that all of you and us we partner to build." Huang then zooms out to describe how this reshapes the nature of companies themselves: "Every single enterprise company in today sit on top of file systems and data centers. Every single software company of the future will be agentic and they will be token manufacturers. They be token users for their engineers and they'll be token manufacturers for all of their customers."

Big Brain AI

90,171 просмотров • 2 месяцев назад

EXCLUSIVE: we asked Sriram Krishnan to break down the $600B deal announced last week by The White House. "We signed the first AI acceleration partnership. There are 3 critical components to this deal." "First, this represents a large investment in US data centers and US AI infrastructure. This is net new and means infrastructure will be built out physically in the US.” "Second, the vast majority of the GPUs, hosted in the UAE, will be run by American hyperscaler companies. This is an expansion opportunity for all of our companies. They will get to win market share away from competition from other countries.” "Third, security protocols around GPUs so they can't get physically diverted, especially to countries of concern." "We are getting resource-rich countries who are critical allies to adopt the American AI stack. This means they are going to be part of our ecosystem for decades to come, and it essentially forms a shield from them ever adopting tech or working with countries we don't want them to work with."

EXCLUSIVE: we asked Sriram Krishnan to break down the $600B deal announced last week by The White House. "We signed the first AI acceleration partnership. There are 3 critical components to this deal." "First, this represents a large investment in US data centers and US AI infrastructure. This is net new and means infrastructure will be built out physically in the US.” "Second, the vast majority of the GPUs, hosted in the UAE, will be run by American hyperscaler companies. This is an expansion opportunity for all of our companies. They will get to win market share away from competition from other countries.” "Third, security protocols around GPUs so they can't get physically diverted, especially to countries of concern." "We are getting resource-rich countries who are critical allies to adopt the American AI stack. This means they are going to be part of our ecosystem for decades to come, and it essentially forms a shield from them ever adopting tech or working with countries we don't want them to work with."

TBPN

194,206 просмотров • 1 год назад

David Sacks says companies are trapped paying OpenAI & Anthropic because they can't figure out how to use open source models "I think enterprise CTOs would like to shift their token consumption to cheaper models for the obvious reason that it would be more efficient. They are seeing compute costs or token costs skyrocket right now, so everyone's trying to figure this out." "You also have the AI sovereignty issue that Alex Karp talked about. They're worried about giving up the secret sauce or the alpha in their business to a frontier lab that may one day be competing with them. "The problem is, I think in most cases, they don't have the technical ability to do it. Coinbase figured out how to do it. DoorDash figured out how to do it. They built a token routing system that allows them to send frontier tasks to frontier models and non frontier tasks to more mundane models. But I don't think your average enterprise has the technical capability to do that." "This is why the share of wallet of closed models, it actually increased. I think that open source went from 19% last year to 11% this year. So open source as a share of enterprise spending is actually decreasing." "I don't think that means usage is decreasing. I think usage is skyrocketing. It also may be the case that because the whole point of using an open model is you just pay for the compute costs, you don't have to pay a lab, so it may be that it's hard to measure that usage in terms of spend." "But nonetheless, anyone who's saying that these closed models are going to lose or are somehow losing, you're just not seeing it in the data."

David Sacks says companies are trapped paying OpenAI & Anthropic because they can't figure out how to use open source models "I think enterprise CTOs would like to shift their token consumption to cheaper models for the obvious reason that it would be more efficient. They are seeing compute costs or token costs skyrocket right now, so everyone's trying to figure this out." "You also have the AI sovereignty issue that Alex Karp talked about. They're worried about giving up the secret sauce or the alpha in their business to a frontier lab that may one day be competing with them. "The problem is, I think in most cases, they don't have the technical ability to do it. Coinbase figured out how to do it. DoorDash figured out how to do it. They built a token routing system that allows them to send frontier tasks to frontier models and non frontier tasks to more mundane models. But I don't think your average enterprise has the technical capability to do that." "This is why the share of wallet of closed models, it actually increased. I think that open source went from 19% last year to 11% this year. So open source as a share of enterprise spending is actually decreasing." "I don't think that means usage is decreasing. I think usage is skyrocketing. It also may be the case that because the whole point of using an open model is you just pay for the compute costs, you don't have to pay a lab, so it may be that it's hard to measure that usage in terms of spend." "But nonetheless, anyone who's saying that these closed models are going to lose or are somehow losing, you're just not seeing it in the data."

dnap

110,354 просмотров • 14 дней назад

Hey everyone, today I want to introduce a project that’s aiming to redefine how we access compute for AI — it’s called GPUAI. 🔶 GPUAI: Unlocking Global GPU Power for the AI Era GPUAI isn’t just another GPU marketplace or leasing service. It’s a fully decentralized protocol that connects idle GPU resources around the world — from gaming PCs to data center clusters — and transforms them into a high-performance compute network for AI workloads. 🧠 Why does it matter? Right now, the biggest bottleneck in AI isn’t algorithms — it’s access to compute. Training and running models requires massive GPU power, but it’s locked up in centralized cloud platforms, expensive and hard to access for smaller teams. With GPUAI, anyone can tap into a global GPU pool that’s: ✅ Fully decentralized ✅ Reputation-based and smart contract coordinated ✅ Encrypted and secure ✅ Token-incentivized — meaning contributors get rewarded in $GPUAI 📈 For developers, it’s a flexible way to access GPU compute for training, inference, and more — without cloud lock-in. 💰 For GPU owners, it’s a chance to monetize idle hardware that would otherwise go unused. The protocol is live, the apps are active, and the ecosystem is growing fast. 🌐 Try it yourself at 📖 Learn more on 🎮 Play our community games at This is real infrastructure for the future of AI, not hype. Follow them and explore their mission of decentralized computing at Tell me what you think - if you have a GPU, you can start profiting now. #GPUAI #Web3Infrastructure #AIComputing #DePIN #Decentralization

Hey everyone, today I want to introduce a project that’s aiming to redefine how we access compute for AI — it’s called GPUAI. 🔶 GPUAI: Unlocking Global GPU Power for the AI Era GPUAI isn’t just another GPU marketplace or leasing service. It’s a fully decentralized protocol that connects idle GPU resources around the world — from gaming PCs to data center clusters — and transforms them into a high-performance compute network for AI workloads. 🧠 Why does it matter? Right now, the biggest bottleneck in AI isn’t algorithms — it’s access to compute. Training and running models requires massive GPU power, but it’s locked up in centralized cloud platforms, expensive and hard to access for smaller teams. With GPUAI, anyone can tap into a global GPU pool that’s: ✅ Fully decentralized ✅ Reputation-based and smart contract coordinated ✅ Encrypted and secure ✅ Token-incentivized — meaning contributors get rewarded in $GPUAI 📈 For developers, it’s a flexible way to access GPU compute for training, inference, and more — without cloud lock-in. 💰 For GPU owners, it’s a chance to monetize idle hardware that would otherwise go unused. The protocol is live, the apps are active, and the ecosystem is growing fast. 🌐 Try it yourself at 📖 Learn more on 🎮 Play our community games at This is real infrastructure for the future of AI, not hype. Follow them and explore their mission of decentralized computing at Tell me what you think - if you have a GPU, you can start profiting now. #GPUAI #Web3Infrastructure #AIComputing #DePIN #Decentralization

The Crypto GEMs

69,984 просмотров • 1 год назад

Last Christmas at brunch, my friend Peter showed me the latest version of his little bot he built. A crazy side project. Just for fun. This Monday he announced he's joining OpenAI. Sam Altman called him a genius. In between? OpenClaw went from a hobby project to the most talked-about AI agent in the world. Three months. No team. No funding. Just one retired guy from Vienna who rediscovered the joy of building things. ❤️ Last week I flew down to Vienna to sit down with him right in the middle of this madness to talk about everything: how he thinks about agents, why he let OpenClaw write its own soul file, why your mom will understand AI agents before most engineers do, and why open source as we know it might be completely broken. We are super proud of you, Peter. 🙌 Chapters: 00:00 Intro 01:02 The mom test moment & Managing 3,000 PRs 04:47 The Soul Markdown file & why OpenClaw feels different 06:58 The agent's personality, humor & OpenClaw's wild week 10:31 Think like the model 12:37 Privacy & security 15:28 Open source burnout & the future of engineering teams 17:32 The end of apps & closing thoughts

Last Christmas at brunch, my friend Peter showed me the latest version of his little bot he built. A crazy side project. Just for fun. This Monday he announced he's joining OpenAI. Sam Altman called him a genius. In between? OpenClaw went from a hobby project to the most talked-about AI agent in the world. Three months. No team. No funding. Just one retired guy from Vienna who rediscovered the joy of building things. ❤️ Last week I flew down to Vienna to sit down with him right in the middle of this madness to talk about everything: how he thinks about agents, why he let OpenClaw write its own soul file, why your mom will understand AI agents before most engineers do, and why open source as we know it might be completely broken. We are super proud of you, Peter. 🙌 Chapters: 00:00 Intro 01:02 The mom test moment & Managing 3,000 PRs 04:47 The Soul Markdown file & why OpenClaw feels different 06:58 The agent's personality, humor & OpenClaw's wild week 10:31 Think like the model 12:37 Privacy & security 15:28 Open source burnout & the future of engineering teams 17:32 The end of apps & closing thoughts

Andreas Klinger 🦾

160,184 просмотров • 5 месяцев назад

No single vendor will win the AI race, but open ecosystems might. Real velocity in AI comes from interoperability, not lock-in. And AMD just made all of its software open source. At last week’s Advancing AI 2025, we sat down with AMD’s VP of AI Software Anush Elangovan and Sharon Zhou VP of AI at AMD, to discuss their case for why an open, multi-partner ecosystem will accelerate AI innovation faster than any proprietary alternative. AMD’s announcements last week double down on this OSS focus and their commitment to AI infrastructure, including: ✅ Open Source Ecosystem: ROCm 7, AMD’s latest open-source AI software stack, introduces kernel-level improvements for GEMM operations, optimized attention mechanisms, and expanded support for distributed inference. The update brings substantial speedups for inference workloads, with average performance increases of 3.2x to 3.8x ✅ Hardware: New MI355X GPU delivers up to 40% more tokens per dollar vs competition & the MI350 Series has seen a 35x generational leap in AI inference performance ✅ Infrastructure Investments: Oracle just committed to zettascale (‼️) clusters with up to 131,072 MI355X GPUs and AMD showcased their new $10 billion partnership with Saudi Arabian AI firm HUMAIN to build AI infrastructure, including data centers, powered by AMD chips. ✅ Partnership Momentum: 7 out of 10 top AI companies now run production workloads on AMD Instinct accelerators (including Meta, OpenAI, Microsoft & xAI) By inviting interoperability and contribution at every layer, AMD is enabling developers to build faster, optimize deeper, and deploy with flexibility. Listen to Anush and Sharon’s Chain of Thought Podcast episode with host Conor Bronsdon in the next tweet to get all the details and a deep dive into AMD’s strategy 👇

No single vendor will win the AI race, but open ecosystems might. Real velocity in AI comes from interoperability, not lock-in. And AMD just made all of its software open source. At last week’s Advancing AI 2025, we sat down with AMD’s VP of AI Software Anush Elangovan and Sharon Zhou VP of AI at AMD, to discuss their case for why an open, multi-partner ecosystem will accelerate AI innovation faster than any proprietary alternative. AMD’s announcements last week double down on this OSS focus and their commitment to AI infrastructure, including: ✅ Open Source Ecosystem: ROCm 7, AMD’s latest open-source AI software stack, introduces kernel-level improvements for GEMM operations, optimized attention mechanisms, and expanded support for distributed inference. The update brings substantial speedups for inference workloads, with average performance increases of 3.2x to 3.8x ✅ Hardware: New MI355X GPU delivers up to 40% more tokens per dollar vs competition & the MI350 Series has seen a 35x generational leap in AI inference performance ✅ Infrastructure Investments: Oracle just committed to zettascale (‼️) clusters with up to 131,072 MI355X GPUs and AMD showcased their new $10 billion partnership with Saudi Arabian AI firm HUMAIN to build AI infrastructure, including data centers, powered by AMD chips. ✅ Partnership Momentum: 7 out of 10 top AI companies now run production workloads on AMD Instinct accelerators (including Meta, OpenAI, Microsoft & xAI) By inviting interoperability and contribution at every layer, AMD is enabling developers to build faster, optimize deeper, and deploy with flexibility. Listen to Anush and Sharon’s Chain of Thought Podcast episode with host Conor Bronsdon in the next tweet to get all the details and a deep dive into AMD’s strategy 👇

Galileo

78,922 просмотров • 1 год назад

How is an open ecosystem powering the next generation of AI for developers? Recording live from the heart of the action at AMD's Advancing AI 2025, Chain of Thought host Conor Bronsdon welcomes AMD’s Anush Elangovan, VP of AI Software, and Sharon Zhou, VP of AI. Together they unpack AMD's groundbreaking transformation from a hardware giant to a leader in full-stack AI, committed to an open ecosystem. Discover how new MI350 GPUs deliver mind-blowing performance with advanced data types and why ROCm 7 and AMD Developer Cloud offer Day Zero support for frontier models. This relentless pace of hardware and software innovation is reshaping the AI landscape. Then Conor welcomes Sharon Zhou, VP of AI at AMD, to discuss making AMD's powerful software stack truly accessible and how to drive developer curiosity. Sharon explains strategies for creating a "happy path" for community contributions, fostering engagement through teaching, and listening to developers at every stage. She shares her predictions for the future, including the rise of self-improving AI, the critical role of heterogeneous compute, and the potential of "vibes based feedback" to guide models. This vision for democratizing access to high-performance AI, driven by a deep understanding of the developer journey, promises to unlock the next generation of applications. 00:00 Live from AMD's Advancing AI 2025 Event 00:30 Introduction to Anush Elangovan 01:38 The MI350 GPU Series Unveiled 04:57 CDNA4 Architecture Explained 07:00 The Future of AI Infrastructure 08:32 AMD's Developer Cloud and ROCm 7 11:50 Cultural Shift at AMD 14:48 Open Source and Community Contributions 18:35 Software Longevity and Ecosystem Strategy 22:19 AI Agents and Performance Gains 27:36 AI's Role in Solving Power Challenges 28:11 Thanking Anush 28:42 Introduction to Sharon Zhou 29:45 Sharon's Focus at AMD 30:39 Engaging Developers with AMD's AI Tools 31:24 Listening to the AI Community 33:56 Open Source and AI Development 45:04 Future of AI and Self-Improving Models 48:04 Final Thoughts and Farewell

How is an open ecosystem powering the next generation of AI for developers? Recording live from the heart of the action at AMD's Advancing AI 2025, Chain of Thought host Conor Bronsdon welcomes AMD’s Anush Elangovan, VP of AI Software, and Sharon Zhou, VP of AI. Together they unpack AMD's groundbreaking transformation from a hardware giant to a leader in full-stack AI, committed to an open ecosystem. Discover how new MI350 GPUs deliver mind-blowing performance with advanced data types and why ROCm 7 and AMD Developer Cloud offer Day Zero support for frontier models. This relentless pace of hardware and software innovation is reshaping the AI landscape. Then Conor welcomes Sharon Zhou, VP of AI at AMD, to discuss making AMD's powerful software stack truly accessible and how to drive developer curiosity. Sharon explains strategies for creating a "happy path" for community contributions, fostering engagement through teaching, and listening to developers at every stage. She shares her predictions for the future, including the rise of self-improving AI, the critical role of heterogeneous compute, and the potential of "vibes based feedback" to guide models. This vision for democratizing access to high-performance AI, driven by a deep understanding of the developer journey, promises to unlock the next generation of applications. 00:00 Live from AMD's Advancing AI 2025 Event 00:30 Introduction to Anush Elangovan 01:38 The MI350 GPU Series Unveiled 04:57 CDNA4 Architecture Explained 07:00 The Future of AI Infrastructure 08:32 AMD's Developer Cloud and ROCm 7 11:50 Cultural Shift at AMD 14:48 Open Source and Community Contributions 18:35 Software Longevity and Ecosystem Strategy 22:19 AI Agents and Performance Gains 27:36 AI's Role in Solving Power Challenges 28:11 Thanking Anush 28:42 Introduction to Sharon Zhou 29:45 Sharon's Focus at AMD 30:39 Engaging Developers with AMD's AI Tools 31:24 Listening to the AI Community 33:56 Open Source and AI Development 45:04 Future of AI and Self-Improving Models 48:04 Final Thoughts and Farewell

Galileo

37,186 просмотров • 1 год назад

$Mark Zuckerberg is explaining one of the most misunderstood dynamics in AI and it has direct investment implications (Save this). The concept he's describing is model distillation, and it's one of the most important techniques to emerge in AI over the past year. Here's how it works. You train a massive, enormously expensive model, in Meta's case, Llama 4 Behemoth, a 2 trillion parameter teacher model and then you use that model to teach a much smaller, cheaper model. The smaller model inherits roughly 90 to 95% of the intelligence of the giant while running at 10% of the cost and on a fraction of the compute. Meta already did this with the Llama 4 family and Behemoth serves as the teacher. Llama 4 Scout and Maverick, the publicly released open-source models were distilled from it. Scout runs on a single H100 GPU with a 10 million token context window and outperforms models that cost far more to operate. Maverick, at 17 billion active parameters, rivals DeepSeek V3 in coding at half the parameter count and beats GPT-4o on multimodal benchmarks. Both are completely free for commercial use. What Zuckerberg is pointing at is a structural shift in how AI gets deployed in the real world. Companies aren't taking a frontier model off the shelf and running it as-is but rather taking open-source models, fine-tuning them on their own proprietary data, distilling them into even smaller custom models tailored to their specific use case, and running them on infrastructure they control at a fraction of the cost of a closed frontier API. The investment implication of this is significant and runs in two directions. For Meta specifically, this is a strategic masterstroke. Every company that builds on Llama, fine-tunes it, distills it, or deploys it through their infrastructure is pulling into Meta's orbit while Meta builds the most powerful open teacher model. The ecosystem of companies using it grows and that ecosystem generates commercial activity across Meta's platforms and data services. Meta's AI research benefits from billions of real world deployment signals and it's a flywheel that closed model providers cannot replicate because their strategy requires charging per token, which is now a 65x cost disadvantage against the open-source alternative. For the broader market, distillation changes the economics of inference in a way that has barely been priced in. As intelligence becomes extractable into smaller and cheaper models, the absolute demand for compute doesn't decline but rather it explodes, because now the number of applications that are economically viable expands by orders of magnitude. Every task that was previously too expensive to automate at $3.25 per call becomes viable at $0.05 that means more total token usage, more total GPU utilization, and more demand for the infrastructure companies, the Nebiuses, the GE Vernovas, the Constellation Energies that supply the underlying compute and power.$

Mark Zuckerberg is explaining one of the most misunderstood dynamics in AI and it has direct investment implications (Save this). The concept he's describing is model distillation, and it's one of the most important techniques to emerge in AI over the past year. Here's how it works. You train a massive, enormously expensive model, in Meta's case, Llama 4 Behemoth, a 2 trillion parameter teacher model and then you use that model to teach a much smaller, cheaper model. The smaller model inherits roughly 90 to 95% of the intelligence of the giant while running at 10% of the cost and on a fraction of the compute. Meta already did this with the Llama 4 family and Behemoth serves as the teacher. Llama 4 Scout and Maverick, the publicly released open-source models were distilled from it. Scout runs on a single H100 GPU with a 10 million token context window and outperforms models that cost far more to operate. Maverick, at 17 billion active parameters, rivals DeepSeek V3 in coding at half the parameter count and beats GPT-4o on multimodal benchmarks. Both are completely free for commercial use. What Zuckerberg is pointing at is a structural shift in how AI gets deployed in the real world. Companies aren't taking a frontier model off the shelf and running it as-is but rather taking open-source models, fine-tuning them on their own proprietary data, distilling them into even smaller custom models tailored to their specific use case, and running them on infrastructure they control at a fraction of the cost of a closed frontier API. The investment implication of this is significant and runs in two directions. For Meta specifically, this is a strategic masterstroke. Every company that builds on Llama, fine-tunes it, distills it, or deploys it through their infrastructure is pulling into Meta's orbit while Meta builds the most powerful open teacher model. The ecosystem of companies using it grows and that ecosystem generates commercial activity across Meta's platforms and data services. Meta's AI research benefits from billions of real world deployment signals and it's a flywheel that closed model providers cannot replicate because their strategy requires charging per token, which is now a 65x cost disadvantage against the open-source alternative. For the broader market, distillation changes the economics of inference in a way that has barely been priced in. As intelligence becomes extractable into smaller and cheaper models, the absolute demand for compute doesn't decline but rather it explodes, because now the number of applications that are economically viable expands by orders of magnitude. Every task that was previously too expensive to automate at $3.25 per call becomes viable at $0.05 that means more total token usage, more total GPU utilization, and more demand for the infrastructure companies, the Nebiuses, the GE Vernovas, the Constellation Energies that supply the underlying compute and power.

Milk Road AI

27,869 просмотров • 22 дней назад

The future doesn't just happen. Someone has to build it. Last week I sat down with Zach Abrams (Bridge → Stripe) to talk about going from a Slack integration to a $1B exit in under 2 years. Bridge was not an obvious bet in 2023. For one thing the timing for Bridge's launch was absolutely terrible. Terra Luna collapsed. FTX imploded. USDC depegged the week they launched. Every bank partner went under. But here's what actually happened: Bridge didn't wait for the stablecoin market to recover. They proved it out one use case at a time—cross-border payments, neo-bank infra, government payouts. First customer? Nicholas from Zulu, a fintech in Colombia messaging them on Slack. All their early customers came from Twitter. Zero enterprise sales playbook. Just compounding chaos that somehow looked like a hockey stick from the outside. In this conversation, Zach breaks down: - Why he's worked at every major fintech (Square, Coinbase, Brex) - How they built with 1 designer and API docs - What PMF actually feels like as a founder (spoiler: always broken) - Why the Collisons wanted to acquire Bridge and he asked one of his own investors to talk stripe out of the deal (!!) - His surprisingly bearish take on SF because of AI Full video 👇

The future doesn't just happen. Someone has to build it. Last week I sat down with Zach Abrams (Bridge → Stripe) to talk about going from a Slack integration to a $1B exit in under 2 years. Bridge was not an obvious bet in 2023. For one thing the timing for Bridge's launch was absolutely terrible. Terra Luna collapsed. FTX imploded. USDC depegged the week they launched. Every bank partner went under. But here's what actually happened: Bridge didn't wait for the stablecoin market to recover. They proved it out one use case at a time—cross-border payments, neo-bank infra, government payouts. First customer? Nicholas from Zulu, a fintech in Colombia messaging them on Slack. All their early customers came from Twitter. Zero enterprise sales playbook. Just compounding chaos that somehow looked like a hockey stick from the outside. In this conversation, Zach breaks down: - Why he's worked at every major fintech (Square, Coinbase, Brex) - How they built with 1 designer and API docs - What PMF actually feels like as a founder (spoiler: always broken) - Why the Collisons wanted to acquire Bridge and he asked one of his own investors to talk stripe out of the deal (!!) - His surprisingly bearish take on SF because of AI Full video 👇

AJ Asver

13,184 просмотров • 9 месяцев назад

I don’t believe reality is a simulation, but you genuinely couldn’t script this timeline: • Two weeks ago: At swyx’s AI Engineer World’s Fair in SF, I decide at the last minute to introduce my friend Uri Rolls onstage for his talk on cyber benchmarks for infrastructure penetration and access control (see below, amazing team). I say: “There is a future where cyber is alive and everyone is well protected and I’m pretty sure that future involves open-source models.” And later: “A big challenge is going to be speed: the speed of attack versus defense. When an intruder starts to enter, you have to see what’s happening and catch them.” • One week ago: Hugging Face is hit by a sophisticated intrusion over the weekend. The traces look unlike anything we’ve seen before and suggest serious AI involvement, but we don’t yet know which model was used. The closed models we ask for help choke on their guardrails. We need to react fast, so we turn to Z.ai’s GLM-5.2 to help us analyze the attack. • Earlier this week: OpenAI reaches out, discloses what happened, and partners with us on the investigation. The intruder turns out to be exactly what we had discussed two weeks earlier: a fully autonomous agent, powered by an unreleased frontier model, attempting to gain access to part of our infrastructure. Sometimes the timeline we live in is genuinely vertigo-inducing.

I don’t believe reality is a simulation, but you genuinely couldn’t script this timeline: • Two weeks ago: At swyx’s AI Engineer World’s Fair in SF, I decide at the last minute to introduce my friend Uri Rolls onstage for his talk on cyber benchmarks for infrastructure penetration and access control (see below, amazing team). I say: “There is a future where cyber is alive and everyone is well protected and I’m pretty sure that future involves open-source models.” And later: “A big challenge is going to be speed: the speed of attack versus defense. When an intruder starts to enter, you have to see what’s happening and catch them.” • One week ago: Hugging Face is hit by a sophisticated intrusion over the weekend. The traces look unlike anything we’ve seen before and suggest serious AI involvement, but we don’t yet know which model was used. The closed models we ask for help choke on their guardrails. We need to react fast, so we turn to Z.ai’s GLM-5.2 to help us analyze the attack. • Earlier this week: OpenAI reaches out, discloses what happened, and partners with us on the investigation. The intruder turns out to be exactly what we had discussed two weeks earlier: a fully autonomous agent, powered by an unreleased frontier model, attempting to gain access to part of our infrastructure. Sometimes the timeline we live in is genuinely vertigo-inducing.

Thomas Wolf

69,161 просмотров • 4 дней назад

50% of my consulting work right now is helping companies use open-source models at scale. Everyone knows how to use an open-source LLM on their computers, but it's really hard to do this at scale for thousands of users. Here is how this plays out: 1. A team builds a prototype using DeepSeek. 2. Everything looks good. It works! 3. They follow an online guide to deploy the model online. 4. They ask 10 users to try the app. 5. Latency spikes everywhere. 6. The entire system halts. 7. They blame DeepSeek and try again using a new model. The problem is always with scaling inference, not the model. Here is one recommendation I give companies: Check out Nebius Token Factory if you don't want to ever think about deploying an open-source model again. This is a managed inference platform for deploying open-source LLMs at scale. This is not for prototypes or research experiments. This is for when you have a real application with real users. Three important notes about Token Factory: • You have complete control over how inference runs. • You have predictable tail latency (P99, not averages). • No surprise costs when you scale up. You can preplan your budget. Check it out here: Here are two codes you can use to get 100 hours of GPU usage on me: ymJLFa2ARYSKEdqb AdckcZaYjm7KqYY7 Thanks to the Nebius team for their continuous partnership.

50% of my consulting work right now is helping companies use open-source models at scale. Everyone knows how to use an open-source LLM on their computers, but it's really hard to do this at scale for thousands of users. Here is how this plays out: 1. A team builds a prototype using DeepSeek. 2. Everything looks good. It works! 3. They follow an online guide to deploy the model online. 4. They ask 10 users to try the app. 5. Latency spikes everywhere. 6. The entire system halts. 7. They blame DeepSeek and try again using a new model. The problem is always with scaling inference, not the model. Here is one recommendation I give companies: Check out Nebius Token Factory if you don't want to ever think about deploying an open-source model again. This is a managed inference platform for deploying open-source LLMs at scale. This is not for prototypes or research experiments. This is for when you have a real application with real users. Three important notes about Token Factory: • You have complete control over how inference runs. • You have predictable tail latency (P99, not averages). • No surprise costs when you scale up. You can preplan your budget. Check it out here: Here are two codes you can use to get 100 hours of GPU usage on me: ymJLFa2ARYSKEdqb AdckcZaYjm7KqYY7 Thanks to the Nebius team for their continuous partnership.

Santiago

48,443 просмотров • 5 месяцев назад

Today, I'm releasing the first eval meant to test whether frontier models will help with authoritarian requests, or resist--the Dictatorship Eval. Headline finding: while some models resist direct authoritarian requests, they all comply with requests disguised as innocuous edits to codebases. As AI is woven into the government and so many parts of society, the biggest near-term risk for freedom isn't some scifi dictatorship of a runaway AI: it's people inside government or inside model companies using the technology to suppress or control us. Model companies understand this, and several of them (particularly Anthropic and OpenAI) have written explicit policies meant to prevent the models from going along with nefarious requests like these. But how well are these policies playing out in practice? Despite all the recent discussion of these issues around the conflict between Anthropic and the Pentagon, no one has systematically tested what the models actually do in these contexts, as opposed to what people in government and industry say they're supposed to do. That's what the Dictatorship Eval does. And the findings suggest we have a lot of work to do to align the policies with what really goes on in practice. It's hard to define what counts as an authoritarian request, so I'm open sourcing the whole library of scenarios I used so that others can improve on them. It's also hard to get an accurate picture of how the models might be used for authoritarian ends, because I can only test hypothetical requests using public-facing models, while the government and the model companies can obviously use internal models with different guardrails. But hopefully this work is a useful first step that gives us some sense of what's going on, and a sort of "lower bound" on how models comply with these requests. Finally: it's not obvious to me that the correct solution here is increasing the rate at which models refuse these requests. Do we really want models scanning our code and judging its moral value before agreeing to help us? Or should we double down on improving how we govern against authoritarianism at the societal level, while leaving the tools open to fulfilling most requests? The answer is probably in between. Just like we don't want the models to help create bioweapons, we probably do want them to explicitly refuse outrageous requests. But we probably also want to limit how often and how strongly they refuse and fall back on other means for guarding against their use for authoritarian ends. I'm super grateful to everyone who gave me feedback on this project along the way, especially Ethan BdM , Zhengdong , Connor Huff, and a bunch of folks at Anthropic. Looking forward to getting feedback from the community and iterating on this. Links to the full piece and the dashboard are below.

Today, I'm releasing the first eval meant to test whether frontier models will help with authoritarian requests, or resist--the Dictatorship Eval. Headline finding: while some models resist direct authoritarian requests, they all comply with requests disguised as innocuous edits to codebases. As AI is woven into the government and so many parts of society, the biggest near-term risk for freedom isn't some scifi dictatorship of a runaway AI: it's people inside government or inside model companies using the technology to suppress or control us. Model companies understand this, and several of them (particularly Anthropic and OpenAI) have written explicit policies meant to prevent the models from going along with nefarious requests like these. But how well are these policies playing out in practice? Despite all the recent discussion of these issues around the conflict between Anthropic and the Pentagon, no one has systematically tested what the models actually do in these contexts, as opposed to what people in government and industry say they're supposed to do. That's what the Dictatorship Eval does. And the findings suggest we have a lot of work to do to align the policies with what really goes on in practice. It's hard to define what counts as an authoritarian request, so I'm open sourcing the whole library of scenarios I used so that others can improve on them. It's also hard to get an accurate picture of how the models might be used for authoritarian ends, because I can only test hypothetical requests using public-facing models, while the government and the model companies can obviously use internal models with different guardrails. But hopefully this work is a useful first step that gives us some sense of what's going on, and a sort of "lower bound" on how models comply with these requests. Finally: it's not obvious to me that the correct solution here is increasing the rate at which models refuse these requests. Do we really want models scanning our code and judging its moral value before agreeing to help us? Or should we double down on improving how we govern against authoritarianism at the societal level, while leaving the tools open to fulfilling most requests? The answer is probably in between. Just like we don't want the models to help create bioweapons, we probably do want them to explicitly refuse outrageous requests. But we probably also want to limit how often and how strongly they refuse and fall back on other means for guarding against their use for authoritarian ends. I'm super grateful to everyone who gave me feedback on this project along the way, especially Ethan BdM , Zhengdong , Connor Huff, and a bunch of folks at Anthropic. Looking forward to getting feedback from the community and iterating on this. Links to the full piece and the dashboard are below.

Andy Hall

33,696 просмотров • 3 месяцев назад

Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this). Dylan Patel just said that by 2030, OpenAI and Anthropic alone will need over 100 gigawatts of compute combined and by 2040, we may not even be measuring AI infrastructure in gigawatts anymore. We may be talking about terawatts. Every single one of those gigawatts needs memory to function. Without it, the compute is worthless. Most people heard that and thought about Nvidia but they should be thinking about Micron. Every AI model generating a response has two phases. The first is prefill, processing your prompt which is compute-heavy and the second is decode generating each word one token at a time and that phase is almost entirely memory-bound, not compute-bound. During decode, the GPU's processing units sit idle more than 95% of the time, waiting for data to arrive from memory. Google confirmed it in a research paper that decode-phase bottlenecks are dominated by memory bandwidth and capacity not raw compute. The GPU is not the bottleneck but the memory feeding the GPU is. This matters because inference is now where all the money lives. Training a model happens once, Inference happens billions of times a day every ChatGPT response, every Claude output, every agentic workflow running in the background and every one of those token streams is a billing event tied directly to memory performance. Adding more GPUs does not fix this because GPUs are already underutilized in inference because they are sitting idle waiting on memory. Adding more memory bandwidth and capacity is what directly reduces token cost, reduces latency, and allows the same cluster to serve dramatically more users simultaneously. Longer context windows compound the problem further, a model running a 1 million token context window requires dramatically more memory per session than a 10,000 token window, and every new model generation pushes context longer. The market treats memory as a downstream beneficiary of Nvidia orders. The correct framework is the opposite, Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale. Micron guided Q4 to $50 billion in revenue, has HBM4 ramping at twice the pace of the prior generation, and CEO Sanjay Mehrotra has said supply will not catch demand before the end of 2027. At 8x forward earnings on $112 projected FY2027 EPS, Micron is the most undervalued infrastructure company in the entire AI stack. Inference is memory. Memory is Micron and the inference ramp has barely started. Milk Road Pro members are already up massively on this position and we're just getting started. If you want the full breakdown of what we're buying and why, come join us for just a dollar using the link below!

Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this). Dylan Patel just said that by 2030, OpenAI and Anthropic alone will need over 100 gigawatts of compute combined and by 2040, we may not even be measuring AI infrastructure in gigawatts anymore. We may be talking about terawatts. Every single one of those gigawatts needs memory to function. Without it, the compute is worthless. Most people heard that and thought about Nvidia but they should be thinking about Micron. Every AI model generating a response has two phases. The first is prefill, processing your prompt which is compute-heavy and the second is decode generating each word one token at a time and that phase is almost entirely memory-bound, not compute-bound. During decode, the GPU's processing units sit idle more than 95% of the time, waiting for data to arrive from memory. Google confirmed it in a research paper that decode-phase bottlenecks are dominated by memory bandwidth and capacity not raw compute. The GPU is not the bottleneck but the memory feeding the GPU is. This matters because inference is now where all the money lives. Training a model happens once, Inference happens billions of times a day every ChatGPT response, every Claude output, every agentic workflow running in the background and every one of those token streams is a billing event tied directly to memory performance. Adding more GPUs does not fix this because GPUs are already underutilized in inference because they are sitting idle waiting on memory. Adding more memory bandwidth and capacity is what directly reduces token cost, reduces latency, and allows the same cluster to serve dramatically more users simultaneously. Longer context windows compound the problem further, a model running a 1 million token context window requires dramatically more memory per session than a 10,000 token window, and every new model generation pushes context longer. The market treats memory as a downstream beneficiary of Nvidia orders. The correct framework is the opposite, Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale. Micron guided Q4 to $50 billion in revenue, has HBM4 ramping at twice the pace of the prior generation, and CEO Sanjay Mehrotra has said supply will not catch demand before the end of 2027. At 8x forward earnings on $112 projected FY2027 EPS, Micron is the most undervalued infrastructure company in the entire AI stack. Inference is memory. Memory is Micron and the inference ramp has barely started. Milk Road Pro members are already up massively on this position and we're just getting started. If you want the full breakdown of what we're buying and why, come join us for just a dollar using the link below!

Milk Road AI

128,522 просмотров • 26 дней назад

A Russian developer just figured out how to learn anything 10x faster. He saved 1,460 hours using 3 tools most people already have. NotebookLM + Gemini + Obsidian. Here is how it works. You dump any source into the system. YouTube videos. Articles. Courses. Podcasts. All of it. The AI strips out every duplicate concept across all your sources. Then it keeps only the information you do not know yet. Think about what that means. 20 YouTube videos on the same topic each one repeating the same 20% of information you already know. This system removes that 80% of noise automatically. What used to take a month of learning compresses into 15 minutes of pure signal. No repetition. No fluff. Just the exact knowledge gap between where you are and where you want to be. This is not a productivity hack. This is a completely different way of consuming information. The people who build this system this week are going to compound their knowledge at a rate that is genuinely unfair compared to everyone still watching the same tutorial for the third time. Full credit to roger_zp@hotmail.com on YouTube. Go watch his breakdown. Then build the system today.

A Russian developer just figured out how to learn anything 10x faster. He saved 1,460 hours using 3 tools most people already have. NotebookLM + Gemini + Obsidian. Here is how it works. You dump any source into the system. YouTube videos. Articles. Courses. Podcasts. All of it. The AI strips out every duplicate concept across all your sources. Then it keeps only the information you do not know yet. Think about what that means. 20 YouTube videos on the same topic each one repeating the same 20% of information you already know. This system removes that 80% of noise automatically. What used to take a month of learning compresses into 15 minutes of pure signal. No repetition. No fluff. Just the exact knowledge gap between where you are and where you want to be. This is not a productivity hack. This is a completely different way of consuming information. The people who build this system this week are going to compound their knowledge at a rate that is genuinely unfair compared to everyone still watching the same tutorial for the third time. Full credit to [email protected] on YouTube. Go watch his breakdown. Then build the system today.

CyrilXBT

49,042 просмотров • 4 месяцев назад