Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

Most AI projects don't fail because models aren't good enough. They fail because inference economics don't scale. During #CXOSpice, Roman Chernin Nebius broke it down: ✅Closed models kill margins ✅Open models unlock control, if you can run them well ✅DataData only becomes a moat when inference is sustainable One... show more

Helen Yu

59,078 subscribers

13,662 Aufrufe • vor 6 Monaten •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

The defining differentiator in AI right now isn't model performance, it's cost per token. Because in this market, efficiency is what turns demand into actual growth. Roman Chernin of Nebius outlines why inference economics are becoming the central challenge, as companies operating on thin margins need infrastructure that can scale without eroding profitability. Token Factory reflects that shift, focusing on orchestration, optimization, and full-stack control to make AI systems economically viable.

The defining differentiator in AI right now isn't model performance, it's cost per token. Because in this market, efficiency is what turns demand into actual growth. Roman Chernin of Nebius outlines why inference economics are becoming the central challenge, as companies operating on thin margins need infrastructure that can scale without eroding profitability. Token Factory reflects that shift, focusing on orchestration, optimization, and full-stack control to make AI systems economically viable.

Six Five Media

294,803 Aufrufe • vor 3 Monaten

My full interview with Roman Chernin, Co-founder & Chief Business Officer of Nebius Nebius just signed a $17 billion deal with Microsoft and a $3 billion deal with Meta. These are two of the biggest tech companies on Earth - and they're coming to Nebius for AI infrastructure. But the company does so much more than that: - Full-stack AI cloud (data centers → software → managed services) - Nebius Token Factory Nebius Token Factory for managed inference & post-training - Partners include Shopify, Higgsfield, Jetbrains My 5 takeaways with Roman Chernin below - thanks for the conversation! Full interview also available on YT ⬇️ Timestamps: 00:00 - Introduction 01:47 - Overview of Nebius $NBIS 03:23 - Roman's background and role at Nebius 04:10 - The $17B Microsoft & $3B Meta deals 05:29 - Why hyperscalers trust Nebius over building in-house 07:06 - Nebius's full-stack approach: Data centers to managed services 08:39 - Customer segmentation: Hyperscalers & Frontier AI Labs vs AI Startups vs Enterprises 14:41 - Token Factory explained: Managed inference & post-training 17:41 - When to switch from closed-source to open-source models 20:59 - Real results: Process achieves 26% cost reduction 25:05 - Why developers love building on Nebius 28:40 - AI agents and the future of infrastructure interaction 32:22 - NVIDIA's Groq acquisition: What it means for inference 35:38 - Why AI adoption will surprise us

My full interview with Roman Chernin, Co-founder & Chief Business Officer of Nebius Nebius just signed a $17 billion deal with Microsoft and a $3 billion deal with Meta. These are two of the biggest tech companies on Earth - and they're coming to Nebius for AI infrastructure. But the company does so much more than that: - Full-stack AI cloud (data centers → software → managed services) - Nebius Token Factory Nebius Token Factory for managed inference & post-training - Partners include Shopify, Higgsfield, Jetbrains My 5 takeaways with Roman Chernin below - thanks for the conversation! Full interview also available on YT ⬇️ Timestamps: 00:00 - Introduction 01:47 - Overview of Nebius $NBIS 03:23 - Roman's background and role at Nebius 04:10 - The $17B Microsoft & $3B Meta deals 05:29 - Why hyperscalers trust Nebius over building in-house 07:06 - Nebius's full-stack approach: Data centers to managed services 08:39 - Customer segmentation: Hyperscalers & Frontier AI Labs vs AI Startups vs Enterprises 14:41 - Token Factory explained: Managed inference & post-training 17:41 - When to switch from closed-source to open-source models 20:59 - Real results: Process achieves 26% cost reduction 25:05 - Why developers love building on Nebius 28:40 - AI agents and the future of infrastructure interaction 32:22 - NVIDIA's Groq acquisition: What it means for inference 35:38 - Why AI adoption will surprise us

Elliot Garreffa

29,384 Aufrufe • vor 6 Monaten

Interview with Nebius Co-Founder Roman Chernin Please like & share this video so that all $NBIS investors on X will see it! :) If you prefer watching on YouTube: Timestamps: 00:00 - Why AI Infrastructure Is So Hard to Understand 00:24 - Market Fragmentation and What Actually Differentiates Providers 01:30 - Consolidation, Segmentation, and the Future AI Cloud Landscape 02:56 - What Analysts and VCs Still Get Wrong About AI Infrastructure 05:34 - Nebius Cloud: Product Readiness and Customer Proof Points 07:42 - Why Inference Workloads Are Exploding 09:11 - Training vs. Inference: How AI Models Actually Reach Production 10:10 - Why Inference Market Share May Concentrate Around a Few Winners 12:36 - Customer Use Cases: Coding, Enterprise AI, and Real-World Adoption 14:01 - Why Integrated Training and Inference Matter Strategically 16:01 - Building Scalable AI Infrastructure With High Utilization 18:24 - Token Factory: Inference as a Managed Service 20:24 - Revolut Case Study: AI-Driven Product Enhancements 22:56 - Token Factory Performance Optimization and Competitive Advantage 25:07 - Scale, Capacity, and Efficiency as Growth Drivers 28:36 - Why Inference Capacity Could Become the Next Major Bottleneck 30:10 - How Nebius Benchmarks Performance Across Providers 33:14 - The Future Size and Shape of the Inference Market 36:38 - Value-Based Pricing: Moving Beyond Cost per GPU Hour 40:55 - How Nebius Wins Deals: Quality, Performance, and Customer Experience 44:53 - Autonomous AI Platforms and the Rise of Agent-Based Models 47:28 - Tavily, Agentic Applications, and the Next Layer of the AI Stack 50:45 - Strategic Trade-Offs: Scaling, Product Roadmap, and Customer Relevance 55:40 - Final Thoughts: Adapting to the Next Shift in AI Workloads Nebius Roman Chernin

Interview with Nebius Co-Founder Roman Chernin Please like & share this video so that all $NBIS investors on X will see it! :) If you prefer watching on YouTube: Timestamps: 00:00 - Why AI Infrastructure Is So Hard to Understand 00:24 - Market Fragmentation and What Actually Differentiates Providers 01:30 - Consolidation, Segmentation, and the Future AI Cloud Landscape 02:56 - What Analysts and VCs Still Get Wrong About AI Infrastructure 05:34 - Nebius Cloud: Product Readiness and Customer Proof Points 07:42 - Why Inference Workloads Are Exploding 09:11 - Training vs. Inference: How AI Models Actually Reach Production 10:10 - Why Inference Market Share May Concentrate Around a Few Winners 12:36 - Customer Use Cases: Coding, Enterprise AI, and Real-World Adoption 14:01 - Why Integrated Training and Inference Matter Strategically 16:01 - Building Scalable AI Infrastructure With High Utilization 18:24 - Token Factory: Inference as a Managed Service 20:24 - Revolut Case Study: AI-Driven Product Enhancements 22:56 - Token Factory Performance Optimization and Competitive Advantage 25:07 - Scale, Capacity, and Efficiency as Growth Drivers 28:36 - Why Inference Capacity Could Become the Next Major Bottleneck 30:10 - How Nebius Benchmarks Performance Across Providers 33:14 - The Future Size and Shape of the Inference Market 36:38 - Value-Based Pricing: Moving Beyond Cost per GPU Hour 40:55 - How Nebius Wins Deals: Quality, Performance, and Customer Experience 44:53 - Autonomous AI Platforms and the Rise of Agent-Based Models 47:28 - Tavily, Agentic Applications, and the Next Layer of the AI Stack 50:45 - Strategic Trade-Offs: Scaling, Product Roadmap, and Customer Relevance 55:40 - Final Thoughts: Adapting to the Next Shift in AI Workloads Nebius Roman Chernin

Daniel Koss

203,706 Aufrufe • vor 3 Monaten

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 Aufrufe • vor 6 Monaten

"Inference, if you look at it as a market, will be much, much bigger than cloud computing was pre-ChatGPT." Lightspeed’s Bucky Moore says inference is an underrated investment category in AI, and expects the market to break up into large, specialized platforms for each modality: "The GPU supply crunch that we're seeing right now is largely, as Dylan Patel has said on the show before, due to the fact that not only these consumer products, but also B2B products like Claude Code and Codex are just really taking off and creating insane demand for inference." "We're talking hundreds of billions in spend every year. And if that's true, I think there will be very, very large inference platforms built in each modality." "So there will be an inference platform for real-time video models, there will be an inference platform for open-source and custom language models, there will be an inference platform built specifically for long-running agents." "So I think we're just going to see that industry, which today looks like one industry, break up into many because of how big it is and how much room for specialization there is."

"Inference, if you look at it as a market, will be much, much bigger than cloud computing was pre-ChatGPT." Lightspeed’s Bucky Moore says inference is an underrated investment category in AI, and expects the market to break up into large, specialized platforms for each modality: "The GPU supply crunch that we're seeing right now is largely, as Dylan Patel has said on the show before, due to the fact that not only these consumer products, but also B2B products like Claude Code and Codex are just really taking off and creating insane demand for inference." "We're talking hundreds of billions in spend every year. And if that's true, I think there will be very, very large inference platforms built in each modality." "So there will be an inference platform for real-time video models, there will be an inference platform for open-source and custom language models, there will be an inference platform built specifically for long-running agents." "So I think we're just going to see that industry, which today looks like one industry, break up into many because of how big it is and how much room for specialization there is."

TBPN

33,278 Aufrufe • vor 4 Monaten

$BREAKING: SambaNova Announces First Close of $1 Billion Series E at $11B Valuation Led by General Atlantic, with participation from Seligman Ventures, T. Rowe Price, & Capital Group SambaNova CEO Rodrigo Liang (Rodrigo Liang) says AI is entering a new era of "premium inference" "The world of inference is running into an economics problem." Enter: new SambaNova "SM50" "A cloud-scale RDU that allows us to deliver premium inference at significantly higher performance than you can get with GPUs, at a fraction of the cost & a fraction of the power." Why does that matter? "When you're running faster with the high-accuracy models, you can charge more at a lower cost, allowing our service providers to generate more margin." Liang says the shift is just beginning: "The world's moving into inference." And after today's funding announcement: "We're using this capital to accelerate our supply chain and make sure we can deliver the racks to customers worldwide." Full interview from Paris coming soon.$

BREAKING: SambaNova Announces First Close of $1 Billion Series E at $11B Valuation Led by General Atlantic, with participation from Seligman Ventures, T. Rowe Price, & Capital Group SambaNova CEO Rodrigo Liang (Rodrigo Liang) says AI is entering a new era of "premium inference" "The world of inference is running into an economics problem." Enter: new SambaNova "SM50" "A cloud-scale RDU that allows us to deliver premium inference at significantly higher performance than you can get with GPUs, at a fraction of the cost & a fraction of the power." Why does that matter? "When you're running faster with the high-accuracy models, you can charge more at a lower cost, allowing our service providers to generate more margin." Liang says the shift is just beginning: "The world's moving into inference." And after today's funding announcement: "We're using this capital to accelerate our supply chain and make sure we can deliver the racks to customers worldwide." Full interview from Paris coming soon.

Molly O’Shea

14,162 Aufrufe • vor 26 Tagen

When the team from Forbes came to our San Jose headquarters, the conversation quickly became about more than just chips. It became a discussion about the future of AI and what it actually takes to run it at scale. As AI adoption accelerates, the challenge is no longer training models, it’s inference: running models efficiently, reliably, and fast enough to power the next generation of agentic applications. As our CEO and Founder Rodrigo Liang put it: “It's much lower power, much smaller footprint, and it's faster than anybody else. What's not to like?” And as Dharmesh Thakker General Partner of Battery Ventures explained: "The technology that SambaNova has is very complimentary to what Intel has because most AI data centers need a combination of CPUs and GPUs." A big thank you to Anton McGonnell and Vasanth Mohan for helping bring this story to life and for the thoughtful conversations throughout the day. We’re excited about what this shift means in practice, from premium inference and agentic AI to more efficient, sustainable data center infrastructure that lets enterprises deploy AI at scale, not just experiment with it. And thank you to Kirsten Taggart and the team at Forbes for spending the time with us and taking a deeper look at what we’re building. 🎥 Watch the full feature below.

When the team from Forbes came to our San Jose headquarters, the conversation quickly became about more than just chips. It became a discussion about the future of AI and what it actually takes to run it at scale. As AI adoption accelerates, the challenge is no longer training models, it’s inference: running models efficiently, reliably, and fast enough to power the next generation of agentic applications. As our CEO and Founder Rodrigo Liang put it: “It's much lower power, much smaller footprint, and it's faster than anybody else. What's not to like?” And as Dharmesh Thakker General Partner of Battery Ventures explained: "The technology that SambaNova has is very complimentary to what Intel has because most AI data centers need a combination of CPUs and GPUs." A big thank you to Anton McGonnell and Vasanth Mohan for helping bring this story to life and for the thoughtful conversations throughout the day. We’re excited about what this shift means in practice, from premium inference and agentic AI to more efficient, sustainable data center infrastructure that lets enterprises deploy AI at scale, not just experiment with it. And thank you to Kirsten Taggart and the team at Forbes for spending the time with us and taking a deeper look at what we’re building. 🎥 Watch the full feature below.

SambaNova

226,753 Aufrufe • vor 1 Monat

"Mac Minis for example are a very good fit" - Andrej Karpathy Andrej Karpathy shouted out my work on EXO Labs in his keynote at Y Combinator AI SUS! Here's the breakdown: Right now most AI workloads run in the cloud where requests from different users are continuously batched together. These workloads are FLOPS-bound and favors hardware with the best unit economics of $ per FLOP, i.e. enterprise GPUs. The personal computing revolution will shift these workloads to personal devices with lower batch sizes (mostly batch_size=1). batch_size=1 inference is memory-bound, because all of the model parameters need to be loaded into the GPU every time a token is generated. Apple Silicon with its Unified Memory architecture has a lot of memory and memory bandwidth per $ compared to other hardware: - M4 Pro Mac Mini, 24GB @ 273GB/s, $58.33/GB, $5.13/GB/s - H100, 80GB @ 3350GB/s, $625/GB, $14.93/GB/s The unit economics of Apple Silicon are becoming more compelling with every release of the Mac. The future of AI inference looks more like open weights models (OpenAI open weights model soonTM) run at low batch_size on personal devices.

"Mac Minis for example are a very good fit" - Andrej Karpathy Andrej Karpathy shouted out my work on EXO Labs in his keynote at Y Combinator AI SUS! Here's the breakdown: Right now most AI workloads run in the cloud where requests from different users are continuously batched together. These workloads are FLOPS-bound and favors hardware with the best unit economics of $ per FLOP, i.e. enterprise GPUs. The personal computing revolution will shift these workloads to personal devices with lower batch sizes (mostly batch_size=1). batch_size=1 inference is memory-bound, because all of the model parameters need to be loaded into the GPU every time a token is generated. Apple Silicon with its Unified Memory architecture has a lot of memory and memory bandwidth per $ compared to other hardware: - M4 Pro Mac Mini, 24GB @ 273GB/s, $58.33/GB, $5.13/GB/s - H100, 80GB @ 3350GB/s, $625/GB, $14.93/GB/s The unit economics of Apple Silicon are becoming more compelling with every release of the Mac. The future of AI inference looks more like open weights models (OpenAI open weights model soonTM) run at low batch_size on personal devices.

Alex Cheema

103,566 Aufrufe • vor 1 Jahr

In between major AI model breakthroughs, new models still ship, but they’re not fundamentally expanding what you can do with them. Braintrust CEO Ankur Goyal says that’s exactly when open source starts to surge: "When new models come out, people forget about open source and forget about economics because it changes fundamentally what you can do." "Programming is a good example. The workflow today is completely different than it was a year ago. If you tried to use models from a year ago today, you wouldn't be able to do crazy stuff like OpenClaw or whatever people are building nowadays." "When models don't change at that speed, people optimize the performance on the use cases that they have." "If you look at usage on our platform, almost half the token usage is coming from open source models from a very small number of companies that have figured out how to optimize use cases really well."

In between major AI model breakthroughs, new models still ship, but they’re not fundamentally expanding what you can do with them. Braintrust CEO Ankur Goyal says that’s exactly when open source starts to surge: "When new models come out, people forget about open source and forget about economics because it changes fundamentally what you can do." "Programming is a good example. The workflow today is completely different than it was a year ago. If you tried to use models from a year ago today, you wouldn't be able to do crazy stuff like OpenClaw or whatever people are building nowadays." "When models don't change at that speed, people optimize the performance on the use cases that they have." "If you look at usage on our platform, almost half the token usage is coming from open source models from a very small number of companies that have figured out how to optimize use cases really well."

TBPN

16,294 Aufrufe • vor 5 Monaten

✈️ Starting the day in Istanbul. Let's talk AI. The future of AI won't be shaped by size, but by precision. Lately I've been diving into what OpenLedger has been building, and I think we're witnessing one of the most important hard forks in AI: 👉 From giant generalist models 👉 To focused, hightrust AI agents Here's why that shift matters , and why OpenLedger's vision makes perfect sense: ✅ Specialized > Generalized General AI can do many things. But specialized AI? It does one thing extraordinarily well. Tailored models don't waste compute on irrelevant context, every parameter is purposedriven. ✅ Explainability isn't optional anymore In highstakes sectors like finance or healthcare, because the model said so won't cut it. We need transparent reasoning paths. Models must show how they reached conclusions , not just what they concluded. ✅ Trust comes from traceability With OpenLedger's Proof of Attribution, each AI decision is traceable, verifiable, and tamperproof. We're talking onchain records of who contributed what , accountability by design. ✅ Less hallucination, more signal Smaller, specialized models trained on clean, domainspecific data are far less prone to hallucinations. Clear boundaries = higher reliability. ✅ Efficiency is the real scalability Deploying one massive model for everything? Expensive, slow, and unsustainable. Specialized AI is leaner, faster, and far more costeffective. In short, OpenLedger isn't just following a trend. They're laying the rails for the infrastructure layer of verifiable, domainaware AI. And in a space flooded with blackbox models and hype, that clarity hits different.

✈️ Starting the day in Istanbul. Let's talk AI. The future of AI won't be shaped by size, but by precision. Lately I've been diving into what OpenLedger has been building, and I think we're witnessing one of the most important hard forks in AI: 👉 From giant generalist models 👉 To focused, hightrust AI agents Here's why that shift matters , and why OpenLedger's vision makes perfect sense: ✅ Specialized > Generalized General AI can do many things. But specialized AI? It does one thing extraordinarily well. Tailored models don't waste compute on irrelevant context, every parameter is purposedriven. ✅ Explainability isn't optional anymore In highstakes sectors like finance or healthcare, because the model said so won't cut it. We need transparent reasoning paths. Models must show how they reached conclusions , not just what they concluded. ✅ Trust comes from traceability With OpenLedger's Proof of Attribution, each AI decision is traceable, verifiable, and tamperproof. We're talking onchain records of who contributed what , accountability by design. ✅ Less hallucination, more signal Smaller, specialized models trained on clean, domainspecific data are far less prone to hallucinations. Clear boundaries = higher reliability. ✅ Efficiency is the real scalability Deploying one massive model for everything? Expensive, slow, and unsustainable. Specialized AI is leaner, faster, and far more costeffective. In short, OpenLedger isn't just following a trend. They're laying the rails for the infrastructure layer of verifiable, domainaware AI. And in a space flooded with blackbox models and hype, that clarity hits different.

Crypto Sinan

13,636 Aufrufe • vor 11 Monaten

Connecting all intelligence with ORA ⏺️ So, how does it work? Let's dive in 👇 At the heart of ORA is the Onchain AI Oracle (OAO), designed to empower dApp developers by seamlessly integrating verifiable AI inference into their applications without sacrificing decentralization. At the center of ORA is Optimistic Machine Learning (opML). This technology ensures that AI inferences originate from the correct models and that results remain untampered. A robust network of nodes independently verifies each AI inference and can raise challenges if tampering is suspected. Want to be part of this decentralized movement? You can easily run your own node using Tora, our user-friendly node software. Built on Optimistic Machine Learning (opML), our Onchain AI Oracle connects the best open-source AI models to the blockchain. This means any dApp can easily become an AI dApp! With our Initial Model Offering (IMO), developers can tokenize open-source models, effectively overcoming the cold start problem by securing essential funding for their projects. ORA is integrated with several L2s, including Mantle Network, Linea.eth, and Arbitrum. This allows you to build on low-cost chains that best suit your needs. ORA is committed to bringing decentralized, verifiable AI to the blockchain. Are you ready to join us? Use OAO: Read more about OAO: Learn about IMO: Use Tora:

Connecting all intelligence with ORA ⏺️ So, how does it work? Let's dive in 👇 At the heart of ORA is the Onchain AI Oracle (OAO), designed to empower dApp developers by seamlessly integrating verifiable AI inference into their applications without sacrificing decentralization. At the center of ORA is Optimistic Machine Learning (opML). This technology ensures that AI inferences originate from the correct models and that results remain untampered. A robust network of nodes independently verifies each AI inference and can raise challenges if tampering is suspected. Want to be part of this decentralized movement? You can easily run your own node using Tora, our user-friendly node software. Built on Optimistic Machine Learning (opML), our Onchain AI Oracle connects the best open-source AI models to the blockchain. This means any dApp can easily become an AI dApp! With our Initial Model Offering (IMO), developers can tokenize open-source models, effectively overcoming the cold start problem by securing essential funding for their projects. ORA is integrated with several L2s, including Mantle Network, Linea.eth, and Arbitrum. This allows you to build on low-cost chains that best suit your needs. ORA is committed to bringing decentralized, verifiable AI to the blockchain. Are you ready to join us? Use OAO: Read more about OAO: Learn about IMO: Use Tora:

ORA

453,392 Aufrufe • vor 1 Jahr

How to become an AI researcher with Charlie O'Neill Charlie co-founded Parsed to build specialized open-source models that can outperform frontier labs. I first met Charlie when Parsed was acquired by Baseten, and now he leads our model development team. Charlie is one of the smartest people I know, and I had the pleasure of talking to him about: 0:00 Intro 3:13 Leaving Oxford to start a company 6:37 Becoming an AI researcher 15:37 Developing a unique POV as your moat 22:04 Parsed origin story 26:01 Big Token, the case for open-source models 33:40 Post-training, fine-tuning, specialization 46:52 Will open models catch up with closed models? 51:50 AI-led job replacement vs job creation 54:45 How to get into inference engineering This is one of my favorite conversations I’ve had in a long time. Made with Lan behind the scenes. Enjoy!

How to become an AI researcher with Charlie O'Neill Charlie co-founded Parsed to build specialized open-source models that can outperform frontier labs. I first met Charlie when Parsed was acquired by Baseten, and now he leads our model development team. Charlie is one of the smartest people I know, and I had the pleasure of talking to him about: 0:00 Intro 3:13 Leaving Oxford to start a company 6:37 Becoming an AI researcher 15:37 Developing a unique POV as your moat 22:04 Parsed origin story 26:01 Big Token, the case for open-source models 33:40 Post-training, fine-tuning, specialization 46:52 Will open models catch up with closed models? 51:50 AI-led job replacement vs job creation 54:45 How to get into inference engineering This is one of my favorite conversations I’ve had in a long time. Made with Lan behind the scenes. Enjoy!

Madison Kanna

129,409 Aufrufe • vor 20 Tagen

$🇨🇳 🇺🇸 CHINA'S AI STACK IS GETTING DANGEROUSLY COMPETITIVE. Huawei's Ascend chips are now powering DeepSeek at a fraction of Western inference costs. The result: DeepSeek just made its 75% API price cut permanent on DeepSeek-V4-Pro. This is not a promotion. This is the new floor. The full vertical stack outside US sanctions and outside Nvidia's margins. DeepSeek can keep cutting. Chinese tech gets cheaper every quarter. A massive global undercut is no longer a question of if, only when. Western AI labs are running on $30,000 Nvidia chips at full markup. DeepSeek is running on Huawei silicon at cost. The unit economics aren't even in the same conversation. Every export ban that pushed China toward Huawei is paying off, for China.$

🇨🇳 🇺🇸 CHINA'S AI STACK IS GETTING DANGEROUSLY COMPETITIVE. Huawei's Ascend chips are now powering DeepSeek at a fraction of Western inference costs. The result: DeepSeek just made its 75% API price cut permanent on DeepSeek-V4-Pro. This is not a promotion. This is the new floor. The full vertical stack outside US sanctions and outside Nvidia's margins. DeepSeek can keep cutting. Chinese tech gets cheaper every quarter. A massive global undercut is no longer a question of if, only when. Western AI labs are running on $30,000 Nvidia chips at full markup. DeepSeek is running on Huawei silicon at cost. The unit economics aren't even in the same conversation. Every export ban that pushed China toward Huawei is paying off, for China.

CryptoGoos

43,154 Aufrufe • vor 2 Monaten

$NBIS cofounder Roman Chernin describes how their recent acquisitions of Eigen AI and Clarifai were all about speed, incredible talent, and acceleration: "The philosophy is very simple. We need to build so many things, and we need to move so fast, that we're always looking for people who can accelerate us. It should be exceptional talent, and/or something that has a great adoption." "Our two recent acquisitions [were] two teams that work on inference optimization. A big part of our business is how efficiently we convert GPUs into tokens. And these two teams — Eigen AI and Clarifai — one is focused on model optimization, the engine of inference. How you run specific models and all the techniques around spec decoding, quantization, and so on." "And the other is system optimization. All the routing, KV caching, and orchestration across the big cluster of compute and so on." "We have a very strong internal team working on inference. But we felt that we needed to move faster, bring more capabilities. Because the market is so fast."

$NBIS cofounder Roman Chernin describes how their recent acquisitions of Eigen AI and Clarifai were all about speed, incredible talent, and acceleration: "The philosophy is very simple. We need to build so many things, and we need to move so fast, that we're always looking for people who can accelerate us. It should be exceptional talent, and/or something that has a great adoption." "Our two recent acquisitions [were] two teams that work on inference optimization. A big part of our business is how efficiently we convert GPUs into tokens. And these two teams — Eigen AI and Clarifai — one is focused on model optimization, the engine of inference. How you run specific models and all the techniques around spec decoding, quantization, and so on." "And the other is system optimization. All the routing, KV caching, and orchestration across the big cluster of compute and so on." "We have a very strong internal team working on inference. But we felt that we needed to move faster, bring more capabilities. Because the market is so fast."

TBPN

30,486 Aufrufe • vor 2 Monaten

Mark my words, Nebius will be the first Trillion dollar Neo-cloud company and here is why (Save this). Roman Chernin, CEO of Nebius just said on 20VC that Nebius raised prices and demand didn't move. When a company can raise prices and still have more demand than supply, that's the opportunity. Chernin also explained why he is deliberately not charging the maximum. As AI shifts from training, a one time cost to inference, which is the ongoing cost of serving every user and every query, compute pricing becomes the cost structure of the entire AI economy. If Nebius prices customers out, those customers cannot grow, and Nebius cannot grow with them. That is the compounding flywheel built directly into the revenue model. The numbers are already confirming it. Q1 2026 revenue came in at $399 million, up 684% year over year. The AI cloud segment grew 840% and represented 98% of total revenue. Adjusted EBITDA flipped positive to $129.5 million. And Nebius signed a long-term agreement with Meta worth up to $27 billion over five years, a hyperscaler outsourcing its own AI compute stack to a neocloud, which tells you that even companies with $50 billion capex budgets cannot build fast enough. Goldman Sachs says the consensus is underestimating 2027 hyperscaler capex by $500 billion. Every dollar hyperscalers cannot provision themselves flows to neoclouds like Nebius. As that gap widens, Nebius captures the overflow with 3 gigawatts of contracted power already secured and a CEO who just told you raising prices did not dent demand. Our subscribers are already up massively on Nebius and come join Milk Road Pro for our full breakdown, how to size Nebius against the broader neocloud opportunity, and our full AI thesis. Link below!

Mark my words, Nebius will be the first Trillion dollar Neo-cloud company and here is why (Save this). Roman Chernin, CEO of Nebius just said on 20VC that Nebius raised prices and demand didn't move. When a company can raise prices and still have more demand than supply, that's the opportunity. Chernin also explained why he is deliberately not charging the maximum. As AI shifts from training, a one time cost to inference, which is the ongoing cost of serving every user and every query, compute pricing becomes the cost structure of the entire AI economy. If Nebius prices customers out, those customers cannot grow, and Nebius cannot grow with them. That is the compounding flywheel built directly into the revenue model. The numbers are already confirming it. Q1 2026 revenue came in at $399 million, up 684% year over year. The AI cloud segment grew 840% and represented 98% of total revenue. Adjusted EBITDA flipped positive to $129.5 million. And Nebius signed a long-term agreement with Meta worth up to $27 billion over five years, a hyperscaler outsourcing its own AI compute stack to a neocloud, which tells you that even companies with $50 billion capex budgets cannot build fast enough. Goldman Sachs says the consensus is underestimating 2027 hyperscaler capex by $500 billion. Every dollar hyperscalers cannot provision themselves flows to neoclouds like Nebius. As that gap widens, Nebius captures the overflow with 3 gigawatts of contracted power already secured and a CEO who just told you raising prices did not dent demand. Our subscribers are already up massively on Nebius and come join Milk Road Pro for our full breakdown, how to size Nebius against the broader neocloud opportunity, and our full AI thesis. Link below!

Milk Road AI

15,677 Aufrufe • vor 1 Monat

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

128,678 Aufrufe • vor 1 Monat

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools

Toha Khan

123,375 Aufrufe • vor 5 Monaten

Nebius will be a trillion dollar company (Save this). The neocloud market, purpose-built AI cloud infrastructure, separate from legacy hyperscalers generated roughly $25 billion in revenue in 2025, up 223% year over year. Synergy Research projects it will approach $400 billion by 2031, compounding at 58% annually one of the fastest sustained growth rates ever recorded for an infrastructure category of this scale. The CEO's explanation for why they win is worth understanding in detail. GPU compute is scarce and that part everyone knows but Nebius is not simply renting GPUs by the hour and marking them up, which is what most neocloud imitators do. They have built their own physical capacity for inference, optimized the full technology stack from the software layer all the way down to the rack hardware and recently acquired a company called Agen specifically to push inference latency even lower and throughput even higher. The CEO frames the core problem directly that in 2026, every product you build is powered by tokens, AI intelligence and while you can get those tokens from OpenAI or Anthropic via a simple API call, the moment you want to run open source models, specialized vertical models, or anything other than the two dominant frontier labs, you run into a wall. You can download the weights from Hugging Face and assemble the pieces. But getting those workloads to run at scale, at the economics you need, with the reliability your product requires, is an extraordinarily complex engineering challenge that most companies cannot staff or afford to solve in-house. That is the problem Nebius is solving, and that is why their inference product called Token Factory exists. The financial results are among the most dramatic growth numbers reported by any public company this year. In Q1 2026, Nebius posted $399 million in revenue, a 684% increase from the same quarter a year earlier. In the span of twelve months, the company swung from a $104 million net loss to $621 million in net income. Cash from operations went from negative $184 million to positive $2.26 billion in the same period meaning this is not growth funded by burning investor capital, it is growth that is now generating its own fuel. For the full year 2026, Nebius is guiding for an annualized revenue run rate of $7 billion to $9 billion, with pipeline creation tracking to surpass $4 billion. The contracted backlog sits at $49 billion, anchored by a $27 billion agreement with Meta, a deal worth up to $19.4 billion with Microsoft, and a public endorsement from Jensen Huang at NVIDIA's GTC conference in 2026. The current market cap is approximately $56 billion. A company with $7 to $9 billion in annualized revenue, growing at 684%, turning cash-flow positive, sitting on $49 billion in contracted backlog, operating in a market compounding at 58% annually toward $400 billion, that company has a credible path to 20x from its current valuation if execution holds. That is the trillion dollar case, and it does not require any heroic assumptions and it requires Nebius to keep doing what it is already demonstrably doing. Milk Road Pro called this one early. Our analysts added Nebius to the portfolio when it was still flying under the radar, and we are sitting on a massive gain on that position right now. If you want to see what else we are building conviction on before the rest of the market catches up, come join us at Milk Road Pro using the link below!

Nebius will be a trillion dollar company (Save this). The neocloud market, purpose-built AI cloud infrastructure, separate from legacy hyperscalers generated roughly $25 billion in revenue in 2025, up 223% year over year. Synergy Research projects it will approach $400 billion by 2031, compounding at 58% annually one of the fastest sustained growth rates ever recorded for an infrastructure category of this scale. The CEO's explanation for why they win is worth understanding in detail. GPU compute is scarce and that part everyone knows but Nebius is not simply renting GPUs by the hour and marking them up, which is what most neocloud imitators do. They have built their own physical capacity for inference, optimized the full technology stack from the software layer all the way down to the rack hardware and recently acquired a company called Agen specifically to push inference latency even lower and throughput even higher. The CEO frames the core problem directly that in 2026, every product you build is powered by tokens, AI intelligence and while you can get those tokens from OpenAI or Anthropic via a simple API call, the moment you want to run open source models, specialized vertical models, or anything other than the two dominant frontier labs, you run into a wall. You can download the weights from Hugging Face and assemble the pieces. But getting those workloads to run at scale, at the economics you need, with the reliability your product requires, is an extraordinarily complex engineering challenge that most companies cannot staff or afford to solve in-house. That is the problem Nebius is solving, and that is why their inference product called Token Factory exists. The financial results are among the most dramatic growth numbers reported by any public company this year. In Q1 2026, Nebius posted $399 million in revenue, a 684% increase from the same quarter a year earlier. In the span of twelve months, the company swung from a $104 million net loss to $621 million in net income. Cash from operations went from negative $184 million to positive $2.26 billion in the same period meaning this is not growth funded by burning investor capital, it is growth that is now generating its own fuel. For the full year 2026, Nebius is guiding for an annualized revenue run rate of $7 billion to $9 billion, with pipeline creation tracking to surpass $4 billion. The contracted backlog sits at $49 billion, anchored by a $27 billion agreement with Meta, a deal worth up to $19.4 billion with Microsoft, and a public endorsement from Jensen Huang at NVIDIA's GTC conference in 2026. The current market cap is approximately $56 billion. A company with $7 to $9 billion in annualized revenue, growing at 684%, turning cash-flow positive, sitting on $49 billion in contracted backlog, operating in a market compounding at 58% annually toward $400 billion, that company has a credible path to 20x from its current valuation if execution holds. That is the trillion dollar case, and it does not require any heroic assumptions and it requires Nebius to keep doing what it is already demonstrably doing. Milk Road Pro called this one early. Our analysts added Nebius to the portfolio when it was still flying under the radar, and we are sitting on a massive gain on that position right now. If you want to see what else we are building conviction on before the rest of the market catches up, come join us at Milk Road Pro using the link below!

Milk Road AI

28,622 Aufrufe • vor 2 Monaten

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 Aufrufe • vor 22 Tagen

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.

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 Aufrufe • vor 1 Monat