Hi everybody, I'm Brian Svin of CNBC. Okay, you might have seen this news. You got this huge hundred billion dollar deal with Nvidia and Chat GPT parent company OpenAI.That was on Monday. The total spend on AI in the United States could be closing in on $1 trillion. But here's the thing.We know you love your AI, but it takes a lot of electricity to run it. In fact, they talked about this 10 gawatt deal. Electricity is squishy.It's mushy. I get it. I'm kind of the energy guy here, so I'll tell you this.1 gaw ...

You're operating in a different level of of the curve, right. I think that's one way of putting it. But I do want to get your your reaction to that open air and video deal, because it's all of it that's happening in the market right now.Right. And thank you for having me. It's great to be back.It's both surprising and in scale, but not surprising that it's happening. Number one, I think we're undershooting or underestimating how much compute we need for not only training, but for inference that growth, that ...

How AI Data Centers Could Save California High-Speed Railhttps://t.co/yReFxjwUnJhttps://t.co/yReFxjwUnJ ...

Energy & AI Industry Concerns - Some states are concerned that bringing AI and AI data centers will drive up the cost of energy for ordinary Americans [1] - AI data centers require 99.999% uptime, raising questions about energy prioritization during emergencies like hurricanes [3] Solutions & Strategies - The industry aims to deliver AI and stop the rise in electricity prices through common sense and data-driven approaches [1] - Turning on backup generators at data centers and critical facilities during peak demand can keep the grid robust [5] - Addressing restrictive EPA rules that prevent the use of backup generators is crucial [4] - Securing the grid and lowering the average delivery cost should be the ultimate impact of AI [3] Examples & Agreements - Google signed a deal with Georgia Power and Light to build a huge data center, locking in their rates for 3 years with no increases [2] Grid & Demand - The only time there is a shortage of electrons is at peak demand time [4] - Consumers will always get the energy first [4]

AI Infrastructure & Investment - OpenAI, Oracle, SoftBank 等公司计划开设五个新的数据中心，最终电力容量将达到 7 吉瓦 [1] - 7 吉瓦的电力大约可供 600 万家庭使用 [1] - 行业普遍认为，AI 领域的持续投入巨大，需要创新的融资方式来支持 [2][6] - 行业对 AI 的需求仍然巨大，尤其是在药物研发、癌症研究等尚未解决的问题领域 [3] - 解决这些问题需要大量的训练、GPU 和电力，因此需要大量的资本投入 [4] - ARM 与 NVIDIA 加强合作，下一代数据中心将使用下一代 ARM CPU [6] - 行业正在寻找更具创意的方式来完成融资，因为所需的资本非常庞大 [6] Power & Sustainability - 数据中心对电力的需求巨大，可能需要相当于数百万家庭的用电量 [8] - 本届政府对使用化石燃料和天然气持更加开放的态度，这在短期内有所帮助 [8] - 随着需求的增长，包括小型模块化反应堆（SMRs）和核能在内的所有能源解决方案都在被考虑 [9] - 行业正在考虑包括小型模块化反应堆（SMRs）和核能在内的所有方案，以满足电力需求 [9]

RT Brody Ford (@BrodyFord_)Oracle is set to spend hundreds of billions on data centers, chips, and power. It has to figure out where that cash comes from. $orcl https://t.co/nhO4UOcsf5 ...

RT Brody Ford (@BrodyFord_)scoop: Oracle is looking to borrow $15 billion dollars as it spends big on data centers for OpenAI. $ORCL w/ @Ethansteinberg0 https://t.co/DECTOBmPKc ...

Nvidia Corp. (NASDAQ: NVDA) invests in a company that will spend the money on data centers. ...

Energy Demand & Supply Concerns - The US existing grid may not have enough power to support the massive scale of data centers, leading to concerns about fulfilling energy demand [1][2] - AI data centers are increasing electricity prices due to their growing power needs [2][3] - Bottlenecks exist in bringing natural gas online, hindering the ability to meet the power demands of data centers [5][6] - Fulfilling data center companies' ambitions will require 5-10 years to bring online energy sources like solar power and batteries [6] Nuclear Energy & Regulatory Landscape - The US has taken a pullback on nuclear energy due to environmental and safety concerns, while China is pressing forward with nuclear reactor construction [4] - The average construction time for a nuclear reactor in China is 57 years, raising questions about the US's ability to meet power demands [4] - Both the Trump and Biden administrations have attempted to ease regulatory bottlenecks for bringing new reactors online, but it still takes a long time [8][9][10] - Nuclear companies anticipate new reactors may not be operational until the end of the decade or into the 2030s [8] Cost Considerations & Investment Outlook - The cost to build a US reactor is significantly higher compared to China, potentially creating a log jam [10][11] - Small modular reactors (SMRs) promise to bring scale and reduce costs, but this is currently theoretical [11][12][13] - For investors with long-term horizons, nuclear power investments may be wise, but near-term volatility should be expected [14] - Investment opportunities exist in bridge power solutions, such as companies like Fluence and Amoresco that are deploying batteries and solar on the grid [6][15]

Just a minute. We want to go back to tech. Nvidia coming off its best day since July after announcing plans to invest up to $100 billion in Open AI as part of the data center buildout. Our Mackenzie Sagalas just caught up with OpenAI's CFO and joins us now with more Mac >> Frank. I'm here in Abalene, Texas at the epicenter of OpenAI's ambitious Stargate project. CFO Sarah Frier telling me in an exclusive interview a short time ago that the bottleneck, it's not money, it is power. OpenAI has reviewed upwards ...