Appleton Finn’s Post

⚡ 𝐄𝐧𝐞𝐫𝐠𝐲 𝐂𝐨𝐧𝐬𝐭𝐫𝐚𝐢𝐧𝐭𝐬 𝐀𝐫𝐞 𝐇𝐨𝐥𝐝𝐢𝐧𝐠 𝐁𝐚𝐜𝐤 𝐀𝐈… 𝐒𝐨, 𝐖𝐡𝐚𝐭’𝐬 𝐍𝐞𝐱𝐭? 𝐍𝐮𝐜𝐥𝐞𝐚𝐫 𝐏𝐨𝐰𝐞𝐫?! ⚡ To keep scaling LLMs, you need three main things: data, compute (GPUs), and electricity. If you’re missing one, well… maybe it’s time to consider a different business model. 😅 As LLMs scale, so does electricity consumption, and many companies are hitting a wall due to government restrictions on power usage. Even Mark Zuckerberg recently said: “We would probably build out bigger clusters than we currently can if we could get the energy to do it.” Now, despite efforts to go green, CNBC reports that Microsoft is reportedly signing deals with nuclear plants to power their next-gen foundation models. 💥 What do you think? Is it worth it—regardless of the price the planet might have to pay for it? 🔗 Link to the article: https://lnkd.in/ddXtRBfX ♻️ Repost if you liked this and follow me for a pragmatic (and occasionally funny) take on Artificial Intelligence and Generative AI! #AI #GenerativeAI #Energy #LLM #NuclearPower #Business #Innovation

Which will win by 2028, fission or fusion? AI (LLM) data centres are using a lot more energy. Indeed, genAI will soak-up an even greater percentage of the world’s power consumption as AI chips will take 1.5% of global electricity use over the next five years, according to TechInsights.  So, the latest announcement from Microsoft on how it’s going to power its own AI is interesting.  Most media outlets picked up on the reopening of Three Mile Island (hence the image along with this post) to provide nuclear energy in four years for Azure ML, Copilot, and their successors. Undoubtedly, this is big news and a bold move.  However, I spotted something even more interesting: Microsoft is simultaneously investing in the development of fusion through Helion, which says they’ll be fully operational *also* in four years.  So my question is: which tech is going to be more commercially realistic by 2028, fission or fusion? Over to you…

In the financial/economics worldview, energy and resources are almost infinitely abundant and scarcity is simply fixed by price signals. For the last 200 years this has been to all intents and purposes true. Nvidia currently trades at c. 80x Revenues. Might we speculate that this be the moment that the financial/technical world collides with the physical/resource world? In the face of the immense demands for electricity that are anticipated consequent to the forecast increase in AI, Silicon Valley is waking up to nuclear.  But ask ChatGPT and driven by its creator's biases it suggests that we just need more renewables (combined with more central planning) - becuase that has such a good historical track-record.... "𝑹𝒆𝒏𝒆𝒘𝒂𝒃𝒍𝒆 𝑬𝒏𝒆𝒓𝒈𝒚 𝑰𝒏𝒕𝒆𝒈𝒓𝒂𝒕𝒊𝒐𝒏**: 𝑰𝒏𝒄𝒓𝒆𝒂𝒔𝒆 𝒕𝒉𝒆 𝒅𝒆𝒑𝒍𝒐𝒚𝒎𝒆𝒏𝒕 𝒐𝒇 𝒓𝒆𝒏𝒆𝒘𝒂𝒃𝒍𝒆 𝒆𝒏𝒆𝒓𝒈𝒚 𝒔𝒐𝒖𝒓𝒄𝒆𝒔, 𝒔𝒖𝒄𝒉 𝒂𝒔 𝒔𝒐𝒍𝒂𝒓, 𝒘𝒊𝒏𝒅, 𝒂𝒏𝒅 𝒉𝒚𝒅𝒓𝒐𝒆𝒍𝒆𝒄𝒕𝒓𝒊𝒄 𝒑𝒐𝒘𝒆𝒓, 𝒕𝒐 𝒎𝒆𝒆𝒕 𝒕𝒉𝒆 𝒈𝒓𝒐𝒘𝒊𝒏𝒈 𝒆𝒏𝒆𝒓𝒈𝒚 𝒅𝒆𝒎𝒂𝒏𝒅. 𝑬𝒏𝒄𝒐𝒖𝒓𝒂𝒈𝒆 𝒕𝒉𝒆 𝒂𝒅𝒐𝒑𝒕𝒊𝒐𝒏 𝒐𝒇 𝒓𝒆𝒏𝒆𝒘𝒂𝒃𝒍𝒆 𝒆𝒏𝒆𝒓𝒈𝒚 𝒕𝒉𝒓𝒐𝒖𝒈𝒉 𝒊𝒏𝒄𝒆𝒏𝒕𝒊𝒗𝒆𝒔, 𝒔𝒖𝒃𝒔𝒊𝒅𝒊𝒆𝒔, 𝒂𝒏𝒅 𝒓𝒆𝒈𝒖𝒍𝒂𝒕𝒐𝒓𝒚 𝒎𝒆𝒄𝒉𝒂𝒏𝒊𝒔𝒎𝒔." What a contrast to the owners of AI who want to preserve their wealth and power - and understand that cheap and reliable electricity is a critical underpinning of the AI revolution. What a contrast. Full post on the linked sub-stack

Accile AI Chronicles: Daily Industry News Amazon, Google and Microsoft Are Investing in Nuclear Power Tech giants like Amazon, Google, and Microsoft are investing heavily in nuclear power as a sustainable energy source for their operations, including artificial intelligence. This move is seen as a way to reduce emissions and ensure a reliable power supply for their energy-intensive businesses. However, concerns about the high costs, safety, and waste management of nuclear power remain. #Amazon #Google #Microsoft #nuclearenergy #AI #sustainability #emissions #energy #investments #techgiants #critique The full article is available here: https://lnkd.in/dh4cAdQM Accile Consulting - AI Automation Specialists Are you looking to enhance your customer experience and keep up with the latest AI advancements? Partner with Accile Consulting to tap into the full potential of AI Automation and revolutionize your business. #AccileConsulting #AIIntegration #Innovation #Partnership #AIAutomation.

The hyper-scalers will probably spend ~$150bn building data centres this year, up from $100bn in 2023, but there are two constraints on building quickly 👇 1) Chips, where NVIDIA and TSMC are capacity-constrained. Also, if you are a model builder, the next generation of LLMs are aiming for a 10x increase in model scale to 300k GPUs. To house one of these models, you need to build an entirely new data center, which could take between 18-24 months of construction time (before you could train your model in 6 to 12 months). These constraints will change the source of maximum competitive advantage in the AI race from research breakthroughs to construction efficiency. Today, five companies have arrived at the starting line in this race toward data center scale-up: Microsoft/OpenAI, Amazon/Anthropic, Google, Meta, Tesla/xAI. 2) Power networks, which have an inconveniently long lead time, as large AI-centric data centres use an implausibly large amount of electricity. So much so that last week both Google and Amazon announced investments in nuclear power. In both cases, the aim is to build smaller, more modular, and more repeatable reactors to get the construction and compliance costs down. Sam Altman, CEO of OpenAI, recently said (https://lnkd.in/dvqAjyHA) that "energy and intelligence are the two key inputs to techno-abundance, and that eventually the cost of intelligence will approximate the cost of energy, so we need to invest in energy solutions as the biggest constraint on the cost of AI." 👉 Google's nuclear clean energy agreement with Kairos Power 👉 Amazon's agreements for nuclear energy projects: https://lnkd.in/daPhhzcm "In this new phase of AI, steel, servers and power will replace models, compute and data as the must-wins for anyone hoping to pull ahead." - Sequoia Capital

Rising electricity demands from AI and EVs are driving nuclear stocks up! 🌍⚡️ With energy consumption set to soar, investors are betting big on nuclear power to meet future needs. Companies like Cameco, Paladin Energy, and Rolls Royce are seeing significant gains. Is this the next big investment wave? 🚀📈 #NuclearEnergy #Stocks #AI #Investing https://lnkd.in/gwfkw548

Data centers are consuming up to 4% of all energy production. "AI" and it's sudden sensationalism have quickly added to the growth of data center energy demands. Existing data centers are reaching capacity and in some jurisdictions, in Ireland for example, are blocking further development. Azure Data Center Capped: https://lnkd.in/eykwnE9b This is largely due to energy infrastructure strain. For some perspective, Google is seeking deals with nuclear power plants to supply data centers. Google Nuclear Power: https://lnkd.in/e8qaq4Ph Ironically, "AI" tools are also a tool to mitigate power consumption. Companies like NetApp (data storage company) have created intelligent infrastructure tools using "AI" which reduce duplicate data, reorganize data to be efficiently stored, etc. Data growth, half of which is usually not accessed again after creation, is a growing issue companies need to address. #Datacenter #Data #AI

☢️ 𝐌𝐈𝐂𝐑𝐎𝐒𝐎𝐅𝐓 𝐁𝐄𝐓𝐒 𝐎𝐍 𝐍𝐔𝐂𝐋𝐄𝐀𝐑 𝐏𝐎𝐖𝐄𝐑 𝐅𝐎𝐑 𝐀𝐈 Microsoft has signed a landmark 20-year deal with Constellation Energy to reopen the Three Mile Island nuclear power plant in Pennsylvania. The goal is to provide clean, reliable energy for Microsoft's data centers and AI operations. 𝐖𝐡𝐚𝐭 𝐚𝐫𝐞 𝐭𝐡𝐞 𝐝𝐞𝐭𝐚𝐢𝐥𝐬? - The agreement will give Microsoft access to carbon-free power for its crucial data centers - This highlights the immense energy requirements of AI and cloud computing - The plant will be renamed the 'Crane Clean Energy Center' and is set to reopen in 2028 - Aligns with recent trends, like Bill Gates' $1 billion nuclear investment in Wyoming 𝐖𝐡𝐲 𝐢𝐬 𝐭𝐡𝐢𝐬 𝐬𝐢𝐠𝐧𝐢𝐟𝐢𝐜𝐚𝐧𝐭? - Tech giants are seeking nuclear solutions to power their growing AI and cloud needs - Microsoft is strategically positioning itself to meet the computational demands of AI - Securing a dedicated nuclear source helps the company maintain a low carbon footprint This bold move by Microsoft underscores the tech industry's increasing reliance on clean, reliable energy sources to fuel the AI revolution and its voracious computing requirements. #ai #artificialintelligence #microsoft #openai #energy #machinelearning

𝗔𝗹𝘁𝗺𝗮𝗻 𝗽𝘂𝘀𝗵𝗲𝘀 𝗺𝗮𝘀𝘀𝗶𝘃𝗲 𝗔𝗜 𝗱𝗮𝘁𝗮𝗰𝗲𝗻𝘁𝗲𝗿𝘀! OpenAI's CEO reportedly asks Biden to back a plan for multi-gigawatt AI datacenters that could consume as much energy as millions of homes 🔋🤖. 𝗪𝗛𝗬 𝗜𝗧 𝗠𝗔𝗧𝗧𝗘𝗥𝗦: These datacenters would turbocharge US AI capabilities, keeping pace with China. But the energy needs are colossal, raising concerns about power grid capacity and environmental impact. 𝗧𝗛𝗘 𝗖𝗢𝗡𝗧𝗘𝗫𝗧: To support AI ambitions, Altman’s plan could involve up to 2.5 million GPUs, requiring more energy than most US nuclear plants produce ⚡. 𝗛𝗘𝗥𝗘’𝗦 𝗧𝗛𝗘 𝗖𝗔𝗧𝗖𝗛: Building even one of these facilities would be a huge challenge, given power shortages and infrastructure demands. Meanwhile, my futurist view is eyeing the day when fusion power solves this problem once and for all 🌍. 𝗚𝗢 𝗗𝗘𝗘𝗣𝗘𝗥: For a full breakdown of Altman’s ambitious AI datacenter plans and their massive energy demands, read the original report from 𝗧𝗵𝗲 𝗥𝗲𝗴𝗶𝘀𝘁𝗲𝗿. https://lnkd.in/gq7S_rWh At Fletter Consulting Group, LLC 𝗗𝗮𝘁𝗮 is one of the core dimensions in our 𝗔𝗜 𝗦𝘁𝗿𝗮𝘁𝗲𝗴𝘆 𝗠𝗼𝘀𝗮𝗶𝗰. Data drives AI innovation, enabling smarter decision-making and deeper insights. Without robust data, even the most advanced AI systems would be ineffective. Data fuels AI’s ability to learn, adapt, and deliver results. To explore how leaders are navigating the AI race, connect with peers in FCG’s 𝗔𝗜 𝗠𝗮𝘀𝘁𝗲𝗿𝗺𝗶𝗻𝗱 𝗚𝗿𝗼𝘂𝗽. #SustainableTech #AIInfrastructure #AIDatacenters