A 5,000 mile wall of superheated water is barreling toward the American coast. The anomaly is poised to expand to cover the entire Pacific coast of North America by late Summer. It is, by most measures, the largest marine heat wave on Earth, turbocharged by climate change.
Against this backdrop, in Washington, the EPA finalized what it called the single largest deregulatory action in American history, stripping the legal foundation from decades of federal climate rules.
And in data centers from Virginia to Wyoming, AI is quietly rewriting the country's electricity demand with consequences that vary by 22x depending on which state's grid is doing the work. Capital is flowing into projects that are decoupling data center demand from the grid. This month, The Green Place maps the collision: a physical world accelerating in one direction, and a policy landscape sprinting in the other.
Thank you for subscribing. Best,
Ted Shabecoff
Climate: The 5,000-Mile Warning
Headline direction: A Wall of Warm Water Is Headed for the American West
A record-setting Pacific marine heat wave is looming just above the horizon. A massive ocean hot spot stretches across 5,000 miles of the Pacific — from Micronesia to the California coast — with waters running 6 to 8 degrees above average.
Climate scientist Daniel Swain has called this “exceptional event,” one that could herald extreme weather later this year. Warmer nearshore waters could make California more vulnerable to tropical cyclones — something cold coastal waters historically prevent — while also driving more humid heat waves over land, a phenomenon Californians are unaccustomed to.
Computer models warn that this broad hot-water region could morph into a full El-Niño along the equator over time, while the distinct California heat wave could persist into fall and winter.
Unfortunately, American grid systems aren’t equipped to handle extreme weather and are already under strain. Extreme temperatures reduce generating capacity and can cause transmission line to lose efficiency.
Layer in the surge in demand from artificial intelligence, which we’ll discuss in this month’s data science spotlight, and its clear that our power systems are not equipped to handle the intense weather events that are in store for us this year.
Capital: The Data Center Is the Green Investment Thesis Now
Headline direction: Climate Tech's Best Bet Is Cleaning Up AI's Power Bill
This past year President Trump cast a large shadow over the green economy, dismantling sustainability policies, and cutting funding for environmental initiatives. But rather than shrink, climate and green investments in the U.S have flatlined.
The green thesis has pivoted from solving the climate crisis to powering AI responsibility. As such, data centers are at the center of the conversation in 2026, and, with no signs of slowdown on AI infrastructure spending there is an urgent need for the cheap and abundant energy that renewables offer.
According to Daniel Goldman, managing partner at Clean Energy Ventures, “Zero-carbon generation is already among the cheapest sources of power, and growing demand for both grid-scale and distributed batteries is accelerating cost reductions faster than expected.”
Total climate tech venture and growth investment rose to $40.5B in 2025, up 8% year-on-year, led by energy security and grid resilience deals — six of the ten largest mega-deals focused on that theme. Meanwhile, US climate tech VC reached $29B in 2025, the third-highest year on record, with founders shifting focus toward unit economics and profitability over growth-at-all-costs amid policy uncertainty.
Across the Atlantic, the European Commission launched a Clean Energy Investment Strategy in March 2026 backed by €75 billion from the EIB over three years, explicitly designed to de-risk projects and mobilize private capital at scale.
On April 28-29, Colombia and the Netherlands co-hosted the first "International Conference on the Just Transition Away from Fossil Fuels" in Santa Marta, Colombia. The event included a “science pre-conference” attended by 400 global academics. One noteworthy participant was absent from the event: the United States.
With American leadership disappearing, the rest of the world has moved on to advance climate action without U.S support. At the CP30 conference in Belem, Brazil, a group of around 80 nations referred to a roadmap away from fossil fuels. In the CP30 text Columbia and the Netherlands announced that they would host a summit to shift the global conversation from if to how to transition away from fossil fuels.
At the Santa Marta conference, governments, parliamentarians and other stakeholders discussed practical steps to move away from fossil fuels.
The gathering was boosted by the outbreak of war in Iran which Fatih Birol, the executive director of the International Energy Agency (IEA), said “ will [lead governments to] review their energy strategies. There will be a significant boost to renewables and nuclear power and a further shift towards a more electrified future.”
At the Summit, Irene Vélez Torres, the environment minister of Colombia similarly commented “it seems that many of us are seeing at the same time that fossil fuels cannot provide energy security, because fossil fuels are subject to scarcity, and scarcity can be manipulated.”
While there were no tangible wins from the summit, the group committed to meeting next year on the Pacific Island of Tuvalu next year and expressed intent to work on regional plans to phase out fossil fuels.
Despite Trump’s pushback on sustainability, it’s clear that consensus amongst global leadership is growing that the clean energy transition should soon be well underway. With new technologies like cryptocurrency and artificial intelligence placing more and more strain on energy systems, the international community must also embrace endeavors to upgrade and green the energy grid.
A resident in Virginia opened their utility bill. They didn't recently buy an EV. They didn't run the AC more frequently than they did last year. But the bill is up $18 for that month.
What’s going on? The answer is sitting in Loudon County, Virginia, in a cluster of windowless buildings. The engine behind Artificial Intelligence is the data center where amidst rows of servers, Graphic Processing Units (GPU) perform thousands of operations simultaneously. Data has silently become one of the fastest growing sources of electrical demand in the U.S. And someone needs to pay for it.
There are typically 4–8 GPUs in AI servers and AI servers need 2–5 kWh total of electricity on average, around the same as a standard home fridge. This powers between 8,333 and ~20,833 Gemini searches. That may not sound like much but the numbers add up quickly. Billions and billions of prompts are sent to models like Open AI GPT-4 and Claude Opus every day and the math compounds fast.
For every watt spent on computation, another 0.4–0.5 watts goes purely to cooling. Chips running at sustained high intensity generate enormous amounts of heat that has to go somewhere. Cooling alone accounts for roughly 38–40% of total data center electricity consumption — a silent multiplier baked into every search, every generated image, every AI-drafted email. And unlike a factory or an office building, data centers run at high utilization 24 hours a day, 365 days a year. There is no off-peak.
According to research by the Lawrence Berkeley National Laboratory, in 2023 U.S data center energy use reached 176TWh in 2023, roughly 4.4% of total U.S. electricity consumption.
This month, we’ve tracked the energy consumption of the artificial intelligence sector in the United States. Virginia alone hosts an estimated 26% of U.S. data center capacity, concentrated in Loudoun County's "Data Center Alley" — the densest cluster of hyperscale infrastructure on earth. In the PJM Interconnection, a grid spanning across 13 states and D.C, data centers drove an estimated $9.3 billion increase in the 2025–26 capacity auction. The average residential customer in western Maryland is expected to absorb $18 per month. In Ohio, $16. The GPUs in Ashburn are on your utility bill whether you use AI or not. The figures, below which I made using Python Jupyter Notebooks, illustrates the staggering trajectory of AI power generation.
