The blue economy, it is not just about ports and policies. The blue economy is about families. It's about generations of families that have worked time and time again to build this beautiful economy.A vital piece that we here on the coast feel a closeness to is seafood. Seafood is intertwined with so many aspects of our daily lives. Think about your community celebrations.Everybody's probably been to seafood festivals, crab festivals, oyster festivals, or your family celebrations. You go to seafood boils. T ...

Core Viewpoint - The competition between Broadcom and AMD to challenge Nvidia's dominance in the AI infrastructure market is intensifying, with both companies showing strong stock performance in 2025, particularly AMD with a year-to-date increase of over 70% compared to Broadcom's approximately 45% gain [1]. Summary by Company AMD - AMD is the second-largest player in the GPU market, focusing on the inference segment where cost-per-inference is crucial, and it has a competitive edge against Nvidia's CUDA software [3]. - Microsoft is developing a toolkit to convert CUDA code to AMD's ROCm, enhancing the use of AMD GPUs for inference, and AMD has partnered with OpenAI to deploy 6 gigawatts of GPUs, starting with 1 gigawatt next year, with OpenAI also acquiring a stake in AMD [4]. - In addition to GPUs, AMD is a leading provider of CPUs for computers and data centers, a rapidly growing market where it is gaining market share [5]. Broadcom - Broadcom approaches the AI chip market by designing custom AI ASICs, which are preprogrammed chips optimized for specific tasks, offering better performance and energy efficiency compared to traditional GPUs [6]. - The company has collaborated with Alphabet to develop Tensor Processing Units (TPUs), which have attracted other major data center operators as customers, with potential revenue from three key customers projected to exceed $60 billion by fiscal year 2027, and a $21 billion order from Anthropic for TPUs [7]. - Both AMD and Broadcom are trading at similar valuations, indicating a competitive landscape [8].

[music] Hey everyone, it's great to meet you all. Really great to be here today. My name is Rhythm. This is my co-founder Lyndon.Our third co-founder, Yash, couldn't make it today, but we're all very excited to be here. Um, three of us were previously researchers at OpenAI, and now we're bringing Frontier AI inside of enterprise at applied compute. Today, we're going to be talking about efficient reinforcement learning.As some context on applied compute, we help enterprises build their own intelligence to p ...

AI Infrastructure Challenges & Solutions - Data bottlenecks constrain GPU performance in AI training and inference, leading to wasted resources and reduced productivity [2][4][5][11] - DDN addresses these bottlenecks by optimizing data movement through fast storage systems and integration with AI frameworks and hardware like Nvidia [5][6] - Inference is becoming increasingly important, with spending expected to surpass training systems, posing challenges in model loading, RAG (Retrieval Augmented Generation), and KV cache management [7][8][9] - DDN Core combines Exascaler for training and Infinia for data management to provide a seamless AI experience [13][14] DDN's Value Proposition - DDN's solutions improve data center efficiency by increasing "answers per watt," delivering more compute with less energy consumption [12][13] - DDN handles KV cache, increasing the effective memory of GPU systems and improving productivity by up to 60% in large-scale GPU data centers [9][10] - DDN offers fast-track solutions for enterprises to adopt AI, whether on the cloud or on-premise, through partnerships like the one with Google Cloud [15][16][17] - DDN's platform supports various use cases, including HPC, AI training and inference, research data management, and secure data sharing [19][20] Strategic Considerations - DDN emphasizes the importance of considering data first when building AI at scale, advocating for data desiloing and secure access [28][29] - DDN supports sovereign AI, enabling nations to develop AI models relevant to their specific data, language, and culture while ensuring security and data sharing [20][21][22] - Partnerships are crucial for delivering efficient AI solutions tailored to customer preferences, whether cloud, on-premise, or hybrid [23][24] - AI factories, which integrate data preparation, training, simulation, and production, present complex data challenges where DDN excels [25][26][27]

RT TOPG MERCH (@topgmerch_)You must train to be good enough to beat your opponent on your worst dayWhile he is on his best day https://t.co/9Yr2nTVusL ...

When you’re exhausted.And can’t carry on. You’ve only done what you can already do.That’s when the training STARTS.How long can you fight when you’re too tired to fight? ...

Core Insights - Nvidia reported third-quarter revenue of $57 billion, reflecting a 62% year-on-year increase, with anticipated revenues of around $215 billion for the year and expected to surpass $300 billion next year [2] - The company is positioned as a leader in the AI sector, with its chips powering significant advancements in AI models and data center expansions, leading to high market confidence reflected in its stock trading multiples [2] - Nvidia's margins are impressive, with approximately 50% net margin, 60% operating margin, and 70% gross margin, indicating strong profitability [2] AI Market Dynamics - AI budgets are increasing as businesses view AI as a transformative platform shift, leading to heightened capital expenditures and acceptance of cash burn by investors [3] - The demand for high-end chips has exceeded supply for over two years, with Nvidia at the center of this demand due to its superior chip performance [4] Competitive Landscape - Competitors like AMD are becoming more competitive, and cloud computing companies are focusing on developing custom chips, raising questions about Nvidia's long-term market position [4][14] - Investors are urging Nvidia's clients to demonstrate measurable AI profitability, which remains largely unachieved [4] Nvidia's Competitive Advantage - Nvidia's moat is not solely based on its chips but on its comprehensive system that integrates multiple components necessary for AI operations, including GPUs, interconnects, and software [5][6] - The CUDA platform is a significant factor in Nvidia's competitive edge, providing a tightly integrated ecosystem that is deeply embedded in AI development, making switching costly for developers [9][11] Future Considerations - While Nvidia is expected to maintain its position in the short to medium term, its long-term lead may diminish as the economics of inference favor specialized silicon and competitors develop their own solutions [12][14] - The shift towards cost efficiency over peak performance may lead to a reevaluation of Nvidia's earnings multiple and potential valuation reset if margins decline or competitors gain market share [15]

RT Avi Chawla (@_avichawla)A popular LLM interview question:"Explain the 4 stages of training LLMs from scratch."(step-by-step explanation below) https://t.co/43WiCQuJfc ...

Core Insights - Google's capital expenditures are projected to rise significantly, from an initial estimate of $60 billion to a current projection of $91–93 billion for 2025, marking an increase of almost 50% [3][4] - The funding is primarily directed towards AI infrastructure, including servers, storage, and chips to support various Google services [4] - Google remains a top customer for Nvidia, with anonymous customers accounting for 39% of Nvidia's revenue, indicating strong demand from major cloud providers [5][9] Capital Expenditures - Google's capital expenditures guidance has increased from $75 billion in February to $85 billion mid-year, and now to $91–93 billion [3] - This represents a substantial year-over-year increase of 75% in capital expenditures [9] AI Infrastructure Investment - The investment is focused on AI infrastructure, including servers, storage, and cooling systems, as well as a large quantity of chips [4] - Google is implementing a dual-track strategy by leveraging Nvidia for flexibility while also utilizing its own Tensor Processing Units (TPUs) for efficiency and cost management [8][12] Nvidia's Role - Nvidia is a key supplier for Google, with the top three hyperscalers (Amazon AWS, Microsoft Azure, Google Cloud) commanding over 60% of the global cloud market [5] - Nvidia's sales have increased by 58%, driven by strong demand and pricing power [9] TPU Development - Google is focusing on TPUs, which are designed for efficient AI inference, as opposed to GPUs that are used for training [8][11] - The latest TPU generation, Ironwood (v7), is reported to be over 4 times faster than its predecessor, with significant improvements in computing power [11] Strategic Positioning - Google's strategy aims to optimize its reliance on Nvidia while enhancing its own TPU capabilities, which could lead to cost control and improved margins [14][17] - As TPUs take on more workloads, Google gains negotiating power with Nvidia, potentially reducing costs associated with chip purchases [13][15] Market Dynamics - The AI landscape is shifting towards inference, where TPUs excel, while Nvidia remains essential for flexibility in cloud services [8][10] - Google's strong position in AI across various services like Search, Ads, and YouTube supports the increased use of TPUs [12]

Core Insights - AI datacenters are entering a new phase focused on inference rather than training, which is expected to reshape the competitive landscape and spending patterns in the industry [1][2][8] Shift from Training to Inference - The focus is shifting from training large models to optimizing inference processes, with techniques like distillation and quantization making inference cheaper and more efficient [2][3] - By 2027, inference is expected to dominate incremental compute spending, with a notable shift already occurring in 2025-2026 [3] Beneficiaries of the Shift - Broadcom is highlighted as a key beneficiary due to its custom ASICs that support inference for major companies like Google, Amazon, and Meta [4] - Marvell Technology is also positioned to benefit as inference workloads increasingly rely on Ethernet and PCIe, moving away from expensive training-oriented technologies [5] Hardware and Networking Trends - Celestica is well-positioned as the industry moves towards standardized, cost-effective inference hardware, allowing operators to source from multiple vendors [6] - Arista Networks continues to support high-performance training networks, but the shift towards Ethernet in inference may create new opportunities for networking companies [6] Power Efficiency and Deployment - Inference is significantly less power-hungry than training, often requiring 5-10 times less power, making it easier to deploy in datacenters with limited grid capacity [7] - The trend towards making AI cheaper, faster, and easier to run is expected to drive spending towards companies like Broadcom and Marvell [8]