AI Energy Teams & Solar Developers - PPAs & Grid Strategy

Why Solar + AI is the defining energy story of this decade Here's a number that should stop you mid-scroll: training a single large AI model can consume as much electricity as 5 average American cars produce in CO2 over their entire lifetimes. Meanwhile, solar just became the cheapest source of electricity in human history-cheaper than coal, gas, and nuclear, in most of the world. These two facts are not unrelated. They are, in fact, the central tension of the next decade of energy. The demand shock nobody planned for When hyperscalers like Microsoft, Google, and Amazon built their 2020–2025 capacity plans, ChatGPT didn't exist. A standard Google search uses roughly 0.3 watt-hours of electricity. A ChatGPT query uses nearly 10 times that. Multiply that across billions of daily interactions, and you begin to understand why data center electricity demand in the US is projected to double by 2030 after being essentially flat for a decade. Virginia alone - home to the world's largest concentration of data centers - is staring down a power demand growth curve that its grid was simply never designed to handle. Dominion Energy, the state's primary utility, has had to tear up its forecasts multiple times in the last two years. Why solar is the only answer that scales fast enough New nuclear takes 15 years and billions in overruns. Natural gas locks in decades of emissions and volatile fuel costs. Wind is constrained by geography. Solar, by contrast, can go from signed contract to generating electricity in 18–24 months at utility scale and the cost has dropped 90% in the last 15 years. This is why Amazon signed agreements for over 100GW of renewable energy. Why Microsoft struck the largest single clean energy deal in history. Why Google has been running on matched renewable energy since 2017. These aren't PR moves. They're infrastructure decisions driven by hard economics: a data center locked into expensive grid power is a data center that loses on operating cost to every competitor who isn't. What this means for the industry The companies that will define the next era of both AI and clean energy aren't just the ones building the best models or the most efficient panels. They're the ones solving the hardest problem at the intersection: how do you deliver reliable, round-the-clock power to a facility that never sleeps, using a source that only generates during daylight? Battery storage, grid interconnection strategy, power purchase agreement structuring, behind-the-meter generation - these are now core competencies for any serious AI infrastructure operator. And for solar developers, the hyper-scaler customer is transforming the economics of the entire industry. Over the next several weeks, I'll be digging into each piece of this puzzle- from how utility-scale solar actually works, to why some solar companies are thriving while others are going bankrupt, to what skills and roles will matter most in this converging industry. To start the conversation: Do you think AI companies are doing enough to address their energy footprint, or is the "we buy renewable credits" approach just greenwashing? Would love to hear from people working in either industry. Sources: IEA World Energy Outlook 2024: solar cost and deployment data Goldman Sachs Power Up report: data center demand projections Dominion Energy 2024 Integrated Resource Plan MIT Technology Review: AI energy consumption analysis Lawrence Berkeley National Laboratory: US Data Center Energy report