U.S. Reverses Climate Bill, China’s Solar Surge & 7-Year Turbine Backlog
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) The biggest energy story of the week, and the year, is the passage into law of the so-called One Big Beautiful Bill that essentially reverses much of the renewable energy and industrial policy enshrined in the Biden-era IRA. This capacity will not immediately be replaced with gas turbines: the turbine majors – GE Vernova, Mitsubishi, and Siemens are sold out, with wait times in some cases up to seven years, and costs reportedly up as much as 50% over the past 10 months.

2.) South Central Texas cooperative Guadalupe Valley Electric Cooperative to work with distributed battery storage developer Base Power to create a 2-MW virtual power plant, with battery systems deployed in new houses built by national homebuilder Lennar.

3.) California’s Turlock Unified School District takes delivery of nine electric Blue Bird buses and charging infrastructure to be charged by mixture of AC and DC chargers taking electricity from solar canopies over the school parking lot as well as utility power. Software from vehicle to x vendor Mobility House will optimize the use of on-site solar.

4.) New Jersey’s State Senate unanimously passes smart solar permitting legislation to expedite and simplify the permitting process for residential solar and battery storage systems with online, automated permitting platform.

5.) China reportedly installs a record 93 GW of solar capacity in May, equal to adding about 100 solar panels every second. To put that figure in perspective, the first three months of Q1 installments in the U.S. totaled 10.1 GW.

Peter Kelly-Detwiler
Google, Gates & the Energy Tech Revolution: Fusion, Fission, and the Future
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) Google signs power purchase agreement PPA with start-up fusion company Commonwealth Fusion Systems for at least 200 MW of energy from Commonwealth’s planned fusion generating station in Virginia. Terms were not disclosed.

2.) Bill Gates- backed modular nuclear reactor company TerraPower raises additional $650 million, with support from Nvidia’s venture capital arm, NVentures. TerraPower is working on an advanced 345MW sodium-cooled reactor, paired with a molten salt-based energy storage system that can boost total output up to as much as 500MW for over five hours.

3.) Modular nuclear start-up X-energy says the NRC could approve construction of its first commercial small modular plant by the end of next year. The first four-reactor, 320-MW plant would deliver both heat and power to Dow’s Seadrift petrochemical facility on the Gulf Coast of Texas.

4.) Arizona state-owned utility Salt River Project comments that projected data center demand is responsible for 60% of its peak load growth and 70% of total energy sales over the coming decade.

5.) Norway’s government said it may impose a temporary ban on new crypto data as early as this fall, to ensure sufficient electricity for other industries.

6.) The New Jersey Board of Public Utilities (NJBPU) rolls out a rebranded energy storage initiative to promote development of 2 GW of energy storage that state is mandated by law to procure by 2030.

7.) ENGIE North America will trial a new precycling provision in PPAs calling for end-of-life recycling of solar panel and project component at four mid-Western projects totaling 375 MW. The initiative should divert almost 48 million pounds of material from landfills. 

Peter Kelly-Detwiler
AWS, Meta & the Military Go Nuclear: New Energy Deals Reshape the Grid
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) Talen Energy and Amazon Web Services sign 1,920-MW power purchase agreement to supply PA data centers from Susquehanna nuclear power plant. Talen and AWS originally sought to expand a 300-MW contract - with Talen serving a co-located AWS data center - to 960 MW, an arrangement rejected by the FERC. The new contract will include the 300 MW w/an additional 1,620 MW in a “front of the meter” framework requiring no federal approval.

2.) Ohio Power Siting Board OK’s a 200 MW gas-fired generation facility to directly serve a new Meta data center behind the meter. Williams will employ three different variants of gas turbines, and 31 total units, including 14 reciprocating engines – none of which will be physically connected to the grid. The $1.6 bn project is expected to be completed by Q3 of 2026.

3.) Advanced geothermal company XGS Energy and Meta Platforms announce deal to develop 150 MW of geothermal energy in New Mexico, connected to local utility PNM’s power grid. A new report identifies over 160,000 MW of untapped geothermal potential in New Mexico.

4.) Enhanced geothermal company Fervo reports successful completion of a new appraisal well three miles in depth, with a projected bottomhole temperature of 520°F. The undertaking took just 16 days, and at depths of 15,000+ feet, it was able to drill 300 feet per hour. This potentially opens up much of the eastern half of the U.S. – where hot rock is deeper - as a potentially viable geothermal resource. National models suggest hundreds of thousands of MWs of geothermal potential between 10,000 and 20,000 feet with temperatures ranging from 400 to 600 °F.

5.) Modular nuclear technology company Oklo says it has received a Notice of Intent to Award by the Defense Logistics Agency Energy on behalf of the Department of the Air Force and the U.S. Department of Defense. Under this agreement, Oklo would design, construct, own, and operate a power plant to provide electricity and heat to Eielson Air Force Base in Alaska.

6.) U.S. energy storage developer Powin files for Chapter 11 relief. With an estimated 9,000 MWh of projects already deployed, Powin cited import tariffs and other political and regulatory uncertainties, especially related to the Investment Tax Credit. Last week, two residential solar companies – Sunnova and Mosaic – also filed for Chapter 11.

7.) U.S. battery swap company Ample announces a partnership in Madrid, Spain with mobility solutions provider Free2move, utilizing up to 100 adapted Fiat 500e vehicles with swappable drop-in battery packs provided by Ample.

Peter Kelly-Detwiler
EV Grid Breakthrough, $2.9M Battery Grant, & Meta’s Massive Nuclear Deal
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) Utrecht Netherlands launches Europe’s first large-scale Vehicle to Grid car-sharing network. "Utrecht energized" connects EVs to the power grid to support renewable energy balancing.The network will start with a fleet of bi-directional 50 Renault 5 E-Tech electric vehicles, with a plan to scale to 500 cars.

2.) Sodium battery manufacturer Unigrid, Inc.  receives $2.9 million grant from California Energy Commission to build 12,000 square foot dedicated sodium battery pilot production facility in San Diego.

3.) Constellation Energy and Meta ink 20-year power purchase agreement for 1,121 MW of energy from Constellation’s Clinton nuclear plant in Illinois. The deal starts June of 2027, when Illinois Zero Emission Credit program expires and will involve re-licensing and output expansion of 30 MW.

4.) Axios reports approximately 1,000 employes, or a third of the overall workforce, have left the U.S. Cybersecurity and Infrastructure Security Agency (CISA), the agency keep our grids and water and gas utilities cyber-secure.

5.) NYISO raises concerns new renewable capacity isn’t being developed fast enough, says it will need additional fossil fuel generation to maintain resource adequacy.

6.) The American Clean Power Association says the U.S installed 7.4 GW of utility-scale solar, wind and energy storage in Q1, 2025, falling just short of Q1 of 2024’s record 8.1 GW. Florida led with 894 MW of new solar capacity.

7.) Powin, the world’s fourth largest battery storage integrators, says it may have to cut up to nearly 250 employees in Oregon and cease operations by late July, owing to “unforeseen circumstances” in the market.

8.) Saudi Arabia’s massive 2,200 MW Neom green hydrogen and ammonia project – the world’s largest - is now 80% complete, according to industrial gas firm Air Products, co-developer and sole offtaker. 

Peter Kelly-Detwiler
The Solar Duck Is Too Fat?! California’s Grid Crisis + Texas & Louisiana Shakeups
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

The U.S. Energy Information Administration reported that California’s grid operator, CAISO, curtailed 3.4 million MWh of utility-scale wind and solar last year, up 29% from 2023. The majority of energy curtailed – 93% - came from solar because the solar duck’s belly is so fat it drags on the ground – especially in the spring when solar output soars and grid demand is low owing largely to the absence of AC load – or because there is inadequate transmission capability to move the juice. The EIA comments that some gas gen must remain running during the solar-saturated daylight hours so that it can ramp quickly enough to meet the evening peak. At times, demand net of solar may soar from close to zero to over 25,000 MW. Exports can help, as can batteries that can time-shift the energy Tummy-tucking the duck during the day and giving it a serious haircut in the evening.

Also in California, CAISO’s Independent System Operator Board of Governors has approved 31 new infrastructure projects worth $4.8 billion, to be built over the next 10 to 15 years. They are meant to help CAISO address the anticipated 76,000 MW of load growth by 2039, brought about by expanding populations of EVs, increased electrification, data center growth and general economic expansion.

As the Texas legislature wrapped up its activity for this year’s session, several pieces of legislation that would have crippled future solar contributions cleared the Senate but failed to make it through the House, to the relief of many concerned about the ability of the Texas grid to meet rapidly burgeoning demand. One would have required all large renewable installations to purchase gas as a back-up – gutting project economics. Another would have set up new fees and setback requirements, and a third would have stipulated that all renewables be matched one-for-one on a capacity basis with dispatchable assets (think gas). These proposed laws may re-surface in two years, but in the meantime, one pro-renewables law that did pass was SB 1202, that speeds up the permitting process for home solar and storage installations. The legislation will permit authorized third parties like licensed engineers to review associated documents and conduct required inspections, with a requirement to submit inspection results to regulators within 15 days. Once the approval documents are submitted, applicants can commence with construction, and regulatory bodies must issue approvals within two business days of initial notification.

Carbon capture and storage projects have already been having a tough go of it, but many projects’ futures just got a lot worse last week with the DOE’s cancelation of $3.7 billion in grants from its Office of Clean Energy Demonstrations, most of which were focused on carbon capture and sequestration. Among companies affected were Calpine, Exxon Mobile, Ørsted, and PPL. Many of these projects had been approved in the November to January interregnum between the Biden and Trump Administrations.

Advanced geothermal company Exceed Geo Energy and the Presidio Municipal Development District have inked a 110 MW geothermal power purchase agreement, which is intended to deliver more capacity and energy than the existing demand. The goal is to lure industries to the region – which sits at the far western edge of the Texas power grid on the border with Mexico, and occasionally suffers from lengthy power outages. Commercial operations are expected to commence in late 2026, with an initial 9.9 MW delivery, eventually expanding to 110 MW. Exceed is also collaborating with Austin Energy on a 5 MW geothermal project.

Swiss solar manufacturer Meyer Burger announced it is shuttering its 1.4 GW Arizona module production facility in Arizona, laying off 282 employees.The company blamed “lack of funds” though it said it’s discussing restructuring with an “ad hoc group of bondholders.” It also deep-sixed a planned solar cell manufacturing plant in Colorado in August. Meyer Burger was notable as the sole US producer of heterojunction technology modules. 

Days after a widespread and sudden power outage over the Memorial Day weekend that put nearly 100,000 customers in Entergy and Cleco’s Louisiana service area, Entergy apologized to its customers in an email saying it was “deeply disappointed by Sunday’s outage event, which occurred when Entergy was directed by our reliability coordinator, MISO (the Midcontinent Independent System Operator), to bring many of our customers offline." MISO called for rolling blackouts in response to low power supplies, resulting from outages at two nuclear units, in order to avoid a larger shutdown or potential grid failure.

Entergy appeared to squarely place the blame on MISO, commenting, “We regret that Entergy was not provided with enough prior notice of the outage to prepare our customers for the potential loss of power.”

Entergy had initially stated that the outage was the result of an unexpected outage at the River Bend nuclear plant, but that plant went offline a full five days before the outage, so there was plenty of time to prepare. Some observers speculate that a breakdown in planning and forecasting between Entergy and MISO was the real underlying issue, but the true cause or causes may take some time to reveal.

Peter Kelly-Detwiler
$20B Nuclear Bet, $1B Fusion Boost, & Solar Grid Wars
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) Ontario Power Generation’s planned first small 300 MW modular reactor in Canada has an official cost of $7.7-billion Canadian (US  $5.5 bn), with $6.1 bn for the GE Hitachi BWRX-300 reactor and an additional $1.6-billion on related infrastructure such as admin buildings and cooling water tunnels that would eventually support three additional BWRX-300s at a later date. The entire project price tag comes to an estimated $20.9-billion.

2.) Modular nuclear company NuScale Power says it is in advanced discussions with several possible customers for its reactors and could deliver an operating power plant in 2030 if they can get a deal signed soon. NuScale says its manufacturing partner Doosan currently has 12 NuScale modules in production.

3.) Fusion start-up Commonwealth Fusion Systems has raised another $1+ billion, with the recent investment led by an unnamed hyperscale data center developer. Commonwealth plans to build its first functioning 400MW fusion reactor within a decade, and last year indicated plans for a grid-scale fusion plant in Virginia by the early 2030s.

4.) Reuters reports U.S. energy security officials found rogue communication devices embedded in Chinese-made inverters as well as batteries, including cellular radios. Unauthorized communications allow adversaries to remotely switch off inverters or change settings, potentially destabilizing grids and resulting in blackouts.

5.) Community solar developer Nexamp plans to develop about 100 new projects across the U.S. for Microsoft across five ISOs to help meet its commitment to be carbon-negative by 2030.

6.) Nexamp says it also inked a deal with Chipotle Mexican Grill for RECs supporting development and operation of 15 new community solar farms generating 75 MW in Illinois, New York, Maryland and Maine.

7.) Abundance Energy, sonnen, and Energywell are collaborating on a Texas virtual power plant combining on-premise solar power, batteries, and advanced energy software. The VPP will focus largely on the greater Dallas-Fort Worth and Houston areas, offering each residence two 4.8kW/20kWh sonnenCore+20 battery units, and will total 60 MWh, with plans for continued expansion.

8.) SunRun says it has expanded participation in California’s Demand Side Grid Support initiative. It’s aggregating power in its CalReady program from over 56,000 homes with solar-plus-storage systems to support the grid between 4 and 9 p.m., from May through October. Sunrun’s expects output of 250 MW per event in 2025, with the potential to peak at 375 MW.

Peter Kelly-Detwiler
Massive Solar Tariffs, Fast-Track Energy Permits, & Big Wins for EV Batteries
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) After a year-long investigation pre-dating the Trump Administration, the U.S. sets new anti-dumping and countervailing duties on solar imports from Cambodia, Malaysia, Thailand, and Vietnam, finding suppliers guilty of exporting to U.S at below production costs. The tariff levels were as high as 3,521% for Cambodian imports. The U.S. imported almost $13 bn of solar from these countries in 2024, representing over 75% of total imports.

2.) The FERC last week approves PJM’s proposal to establish a price cap and price floor for its next two capacity auctions affecting the 2026/27 and 2027/28 delivery years. This action sets a floor of $175/MW-day and a ceiling of $375, replacing the former $500 ceiling and $0 floor.

3.) The US government announces new emergency permitting process for energy and mining projects on federal lands, with approval times for these projects cut from one to two years down to 14 to 18 days.

4.) The DOI justifies decision as responding to President Trump’s declaration of an energy emergency. Permitting will only apply to fossil fuel projects, geothermal power, coal, uranium, other critical minerals, biofuels, and kinetic hydropower - other renewable energy projects need not apply.

5.) GM now exceeds Tesla in U.S. battery production capacity, according to BloombergNEF, and has dramtically cut battery costs - by $60 per kWh from 2023 with further cost reductions of $30 per kWh expected. The goal is $100 per kWh, down 50% from 2023.

6.) Automotive group Stellantis and US battery developer Factorial Energy validate solid-state EV battery cells, part of a joint effort to create a demonstration fleet of Dodge Charger Daytonas equipped with solid state batteries next year. The two companies say the cells support a state-of-charge increase from 15 % to over 90 % within 18 minutes.

7.) ISO-NE sees record low energy consumption for 4th year in a row, with preliminary data showing demand at 5,318 MW on April 20 Easter Sunday, 2025. On-site solar cut into demand, peaking at around 6,600 MW.

Peter Kelly-Detwiler
Geothermal Surges, AI Powers the Grid & EVs Boom
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) Advanced geothermal company Fervo Energy inks 15-year PPA with Shell Energy North America for 31MW of geothermal power from Fervo’s Cape Station geothermal project in Utah. The deal will help Fervo expand the project capacity from 400MW to 500MW, all of which is now fully contracted.

2.) The Utah state office of the Bureau of Land Management (BLM) runs competitive geothermal lease sale in Utah, netting over USD 5.6 million in bids from four companies, covering 14 parcels totaling 50,961.

3.) Google and Baseload Capital 10 MW PPA for geothermal energy in Taiwan, with Google also directly investing in Baseload Capital to further accelerating the deployment of geothermal energy in the region.

4.) Siemens Gamesa completes work on a 21.5-MW prototype offshore turbine at a test center in Denmark. It is the world’s most powerful installed wind turbine, but three Chinese companies are racing to develop 25-MW turbines.

5.) The Trump administration's decision to halt construction of Equinor's 54-turbine 810 MW Empire Wind 1 farm off New York is raising concerns that fully-permitted developments representing billions in investment are now in peril.

6.) Porsche’s EV line-up has solid first quarter: Of the 71,470 cars delivered, 38.5 % had a plug, with 26% being all electric.

7.) Per Reuters, Chinese battery giant CATL is seeking to buy a controlling stake in EV maker Nio's power unit, with its 3,000+ battery swapping stations in China.

8.) Texas start-up Base Power – a company that installs oversized home batteries at low prices, and often under a subscription service, raises an additional $200 million in a Series B round. The influx of cash will support market growth in Texas as well as construction of a new factory for its battery systems.

9.) The Canadian Nuclear Safety Commission (CNSC) issues power reactor construction license to Ontario Power Generation Inc. (OPG) to build a General Electric Hitachi BWRX-300 small modular reactor at its Darlington New Nuclear Project site in Ontario.

10.) Google’s subsidiary Tapestry partners with PJM to develop AI-powered tools to streamline and accelerate the interconnection process. Tapestry will integrate dozens of PJM’s databases and tools to create a single model of PJM’s transmission network. Tapestry has already been used in Chile, to develop and pilot a planning tool for the nation’s transmission grid. This year, the Chileans will fully integrate the tool into their processes.

Peter Kelly-Detwiler
Texas Energy Crisis Deepens, Nuclear Projects Rise, & Gas Projects Collapse
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) Facing declining reserve margins, Southwest Power Pool is looking to Demand Response to bail it out. SPP projects its reserve margin may fall to 5% by 2029, so it’s resorting to a “comprehensive” demand response policy.

2.) Dandelion – the country’s leader in geothermal heat pumps – is working with home builder Lennar to integrate ground-source heat pumps into 1,500 new Lennar homes in Colorado over the next two years.

3.) Dow and modular nuclear company X-Energy Reactor Company, LLC have submitted a construction permit application to the Nuclear Regulatory Commission for nuclear project in Seadrift, Texas. Approval of the construction permit could take up to 30 months.

4.) The $5 billion low-interest rate Texas Energy Fund, established in 2023 to entice dispatchable gas generation to Texas looks to be in trouble. Engie bailed on its 930-megawatt peaker plant in February, blaming “equipment procurement constraints.” Wattbridge then pulled four projects totaling1.62 GW in late March, citing “risk and costs.” Last week Constellation, canceled its 300 MW Wolf Hollow expansion plan.

5.) The site of Pennsylvania’s Homer City coal plant – 50 miles east of Pittsburg - is being redeveloped to serve a huge data center complex. The new $10 billion facility will employ 7 GE Vernova gas turbines to generate up to 4.5 GW of power – twice the output of the former coal plant shuttered in 2023- and start generating by 2027.

Peter Kelly-Detwiler
EV Battery Wars, AI-Powered Grids & Green Hydrogen Breakthroughs
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) California-based battery swap startup Ample is planning to build a network of stations in Tokyo, each of which can support charging over 100 electric Mitsubishi Fuso delivery trucks.

2.) Chinese battery giant CATL and automaker Nio will launch what they claim is the world's largest EV battery swapping network and will begin coordination in developing battery swapping stations.

3.) Chinese EV maker BYD says its EV platform will enable drivers to charge as fast as drivers refuel gas-powered vehicles. Its Super e-Platform uses flash-charging batteries and new silicon carbide power chips that can charge vehicles at a rate of one megawatt - well over one mile per second.

4.) Itron is collaborating with leading AI chipmaker NVIDIA to accelerate adoption of AI tech at the grid edge. The goal is to combine Itron’s 13 million+ distributed intelligence-enabled endpoints with NVIDIA’s AI tech to create a new level of distributed situational awareness.

5.) California regulators have OK’d new standards for maintaining and operating batteries that will include a requirement for emergency response and action plans. This after a fire at Vistra’s huge Moss Landing battery facility. California’s installed battery capacity now represents over 20% of the state’s peak demand.

6.) The Midcontinent Independent System Operator (MISO) is asking the FERC to approve an Expedited Resource Addition Study process that would speed up its interconnection process for supply assets “that can address urgent resource adequacy and reliability needs in the near term.” In many cases that likely means “gas.” In its December reliability assessment, NERC identified MISO as being most at risk of capacity shortfalls. MISO wants ERAS to come into effect by mid-May, with designated projects to be offered an Expedited Generator Interconnection Agreement within 90 days.

7.) RWE and TotalEnergies signed a 15-year green hydrogen supply deal for RWE’s forthcoming 300 MW electrolysis plant in Nieder-Sachsen, Germany to supply TotalEnergies’ refinery in Saxony-Anhalt, with 30,000 metric tons of green hydrogen. This is the biggest long-term green hydrogen agreement signed in Germany to date.

8.) Amazon is unveiling a carbon credit service – adhering to the most rigorous standards - that looks to restore trust in the voluntary carbon market. This will help companies in its value chain better manage their carbon exposure. Amazon requires participants to have net-zero emissions by 2050 targets, to measure and publicly report their emissions and commit to ongoing decarbonization strategies based on latest climate science.

9.) Skytree and Return Carbon, in partnership with Verified Carbon, are working with EDF Renewables North America to develop Direct Air Capture facilities in Texas. The aim is to develop 500,000 tons per year of negative emissions and store that carbon along the Gulf Coast. Skytree’s DAC technology will be tied to a behind-the meter, but grid-connected EDF wind farm. Return Carbon will bring capital to the game, while Verified Carbon will be responsible for the sequestration. EDF, of course, will supply the power.

Peter Kelly-Detwiler
GM’s Big EV Incentives, Military Geothermal Plans & AI’s Energy Shakeup!
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) GM teams up with PG&E on resi Vehicle-to-Everything pilot, with eligible customers receiving up to $4,500 in incentives for GM Energy home products like the GM Energy V2H Bundle or Home System. Six 2024 EV models eligible and 2025 model years to be added soon. 

2.) U.S. Air Force and DOD designate a team including GE Vernova, Sage Geosystems, the Energy and Geoscience Institute, and the University of Utah to explore deployment of utility-scale geothermal for future renewable and hydrogen energy microgrids at military bases. Sage will provide its geothermal technology while GE Vernova offer capabilities related to power conversion, energy storage, and microgrids.

3.) Distributed energy storage company Base to work with Texas Bandera Electric Cooperative in its residential battery program offering homeowners battery backup systems. Bandera is developing distributed energy storage network to provide grid services, increase resilience, and provide economic value to its members. Battery Storage Subscription Program will have Base providing members with batteries for monthly subscription fee but no upfront cost.

4.) Cybersecurity company Dragos releases case study outlining a cyberattack from Chinese Volt Typhoon hackers on Massachusetts municipal utility Littleton Electric Light and Water Departments. The utility was able to identify and eliminate the threat, but hackers apparently infiltrated the utility about nine months prior to being exposed.

5.) Chinese tech giant Baidu unveils newest AI models - Ernie X1 and Ernie 4.5 that compete with Open AI and Deepseek in terms of performance and cost. Baidu claims its multimodal foundation model Ernie 4.5 "outperforms GPT-4.5 in multiple benchmarks while priced at just 1% of GPT-4.5. This strongly suggests much lower possible future power consumption, with several technologies powering creating new operational efficiencies although energy consumption metrics were not disclosed.

Peter Kelly-Detwiler
Power Wars, EV Revolution & Texas Nuclear Boom!
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) After the Trump Administration slapped tariffs on imports from Canada, including a 10% levy on power, Ontario Premier Doug Ford said he would impose 25% tariffs on power flowing to Michigan, New York and Minnesota. Ford also warned, “I will not hesitate to shut the electricity off completely.” 2023 New York imports totaled 4.4%, while Michigan and Minnesota imported less than .5%.

2.) The Bonneville Power Administration (BPA) announced a draft plan to join the Southwest Power Pool’s Markets+ real-time and day-ahead market instead of CAISO’s competing Extended Day-Ahead Market (EDAM).

3.) Kia is taking pre-orders for the Wallbox Quasar 2 home charger and accompanying hardware to accommodate bidirectional charging from its 76 kWh EV9 electric SUV. The combined vehicle and charger system will enable both home back-up power supply as well as vehicle-to-grid functionality. The charging equipment will sell for $6,440. Pre-orders are limited to residents of seven states.

4.) Also in V2G, Nuvve was recently awarded a contract from the State of New Mexico for up to $400 million over the next 4 years to deliver a comprehensive, turnkey electrification solution to support New Mexico’s zero-emission vehicle adoption and renewable energy goals. This will include electrification of 2,000+ buses and 3,500 state-owned vehicles as well as V2G hubs and turnkey charging solutions.

5.) Utah passed to allow 1.2 kW or smaller portable solar devices to be plugged directly into standard 120 volt wall sockets without requiring any interconnection requests to the utility. If the governor signs, it will take effect on May 7.

6.) Now let’s move to Texas - modular nuclear reactor (MNR) startup Last Energy announced plans to build 30 microreactors in Texas near the Dallas-Fort Worth areas.

7.) Reuters reports that a lengthy period of low wind speeds in Texas has resulted in record high output from the fossil fleet thus far this year. As a result, some fossil generators may shorten or delay planned maintenance outage breaks this spring during the shoulder demand period. This could represent a critical challenge coming into the warmer weather ahead.

Peter Kelly-Detwiler
Game-Changer for EVs & Energy Policy!
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

1.) The Massachusetts Clean Energy Center (MassCEC), has chosen companies firms to support a vehicle-to-everything (V2X) demonstration program. The pilot will provide a free bidirectional EV charger to 100 participants in the residential, commercial and school-district sectors.

2.) Sticking with electric transportation, it appears Mercedes Benz is making real progress in the race to commercialize solid state battery tech in vehicles. The company reports it has installed a solid-state battery pack, with cells from U.S. company Factorial Energy, into a modified EQS Sedan, starting road trials last month. This battery offers up to 620 miles of range in this configuration.

3.) Implementation of tariffs from Canada and Mexico starts today, March 4th. Canadian electricity imports will see a 10% levy. New England and New York grid operators are not quite sure what this means for them, but in order to be ready, each filed tariffs with the Federal Energy Regulatory Commission last week. ISO-NE and NYISO said they don’t think it’s their job to collect the duties, but they need to understand their responsibilities.

4.) ISO-NE estimated that a 10 percent tariff on Canadian electricity imports could cost $66 million annually.

5.) Cutting basic government-funded energy research, whether in health, energy, or the other sciences, risks impoverishing this country in the future.

6.) Fracking is one such example. U.S. - sponsored research included multiple shale gas projects, including everything from three -dimensional micro-seismic imaging to advanced drill bit technology development, as well as close collaboration with the Gas research Institute. In 15 years, shale gas production went from nothing to about two-thirds of total U.S. gas production.

7.) Today, our money funds materials science development in our national labs, as well as critically important cybersecurity initiatives.

8.) Federal research fosters improvements in advanced geothermal technologies at the FORGE project in Utah, modular nuclear technology research including reactor physics, modeling, simulation, and safety analysis. The list goes on.

Does the federal budget need to be managed? Yes. Does our national debt threaten our future well-being? Yes. And it needs to be addressed.  But the chainsaw approach is short-sighted and counterproductive in a world made up of interconnected systems of systems. 

Peter Kelly-Detwiler
Power Grid Crisis: 10,000 MW Shortfall by 2030?!
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

With interconnection queues are clogged, little new transmission being built, and enormous increases in load growth in some areas of the country, ensuring resource adequacy and managing grid reliability become a growing problem.

PJM is one grid operator facing this challenge, and has recently warned that in a few years it may not have the dispatchable capacity needed to keep the lights on, warning of a shortfall of as much as 10,000 MW of capacity by the 2030/31 year. PJM filed its Reliability Resource Initiative with the FERC last fall. This approach creates a one-time cut to the head of the line fast track review of up to 50 shovel-ready generation or storage assets that meet eligibility scoring criteria related to viability, reliability, and availability.  On February 11, the FERC approved this approach, finding it “just and reasonable and not unduly discriminatory.” Any approved assets will be required to participate in capacity markets for at least a decade.

FERC also approved PJM’s proposal to increase and accelerate access to the transmission system by changing its Surplus Interconnection Service rules, allowing generators to more quickly access existing approved interconnection if they do not exceed the rated capacity of that interconnection point. It also makes it specifically easier for storage resources, allowing for surplus interconnection service “from resources seeking to receive electric energy from the grid and store it for later injection to the grid.” 

So, for example, if a 100 MW gas peaker had access to an interconnection, a battery or solar and batteries could be added to that same delivery point a long as no more than 100 MW was ever delivered from the combined assets.  That’s good, as it improves efficiency of existing assets. But it doesn’t get us what we really need, which is a lot more transmission.

There is now a precedent here which is not a good one for developers of renewables in other areas governed by grid operators – many of which are also facing capacity shortfalls and reliability issues. So, look for potentially more of these types of activities to come.

The CTO of BYD’s battery business stated that his company has already produced its first solid-state cells on a pilot production line last year, with “mass demonstration” of solid-state batteries around 2027. Large-scale introduction of solid-state batteries might only take place after 2030.

Siemens Gamesa has confirmed investment of over $200m to expand its offshore wind blade manufacturing facility in northern France, to be completed by 2026. The expansion will focus on manufacture of 115-metre-long blades for its 14MW turbines. 

Researchers from the University of Cambridge have developed a reactor that pulls carbon dioxide directly from the air and converts it into sustainable fuel, using sunlight as the power source. The solar-powered flow reactor, uses specialised filters to capture CO2 at night, and When the sun comes out, the sunlight heats up the captured CO2. A semiconductor powder absorbs the ultraviolet radiation, initiating a chemical reaction to converts the captured CO2 into solar syngas. 

Peter Kelly-Detwiler
Rethinking Data Center Demand: The Future of AI, Energy Consumption, and Load Projections
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

Just when you get comfortable thinking you know something, you find out that maybe you don’t. In a series of videos late last year, I addressed the issue of exploding data center electricity demand, and the enormous number of applications utilities have received in recent months.  I have been tracking these on a spreadsheet, based on various press releases and articles in the trade press, and thus far I’ve got over 125,000 MW of new projected data center demand. Not all of this demand is AI-related. Some new load will serve your typical data center applications, while some may even be serving crypto loads, now that crypto is in fashion in Washington. 

But there had been some subtle signs that perhaps this new load might not be as big as the headlines suggest. Skepticism was already the word of the day before news came out of China last week that an open-source AI large language model (LLM) there called DeepSeek was nearly as good as some of the proprietary models being built here in the U.S. by some of the biggest players in the space. The news that mattered most – to markets – was that it was not only competitive, but much cheaper, using fewer chips and far less power. DeepSeek reported that its model took only two months and less than $6 mn to build, using a less advanced (and less costly) H800 Nvidia chip.

The one-day carnage on Wall Street was amazing to behold. Leading chip maker Nvidia’s share price fell off a cliff, losing 17% and 600 billion – with a B – of market value. Modular nuclear and fuel cell stocks got savaged as well, shedding up to 25% off their stock prices. 

Over the ensuing week, additional news filtered out that perhaps those numbers weren’t quite so reliable, coupled with accusations that there had been some so-called distilling - transferring knowledge from OpenAI to DeepSeek, or at least some reverse engineering from other AI models. So, it wasn’t like it was built from scratch.

Now come three questions related to the grid and future power consumption:

  1. How much of DeepSeek’s claims will eventually prove to be true, both in terms of the time and resources required to build their LLM, and are there implications for other large language models that essentially use big chips and lots of power to brute force their way through their trainings?

  2. Is the model really that good?

  3. If one can really build AI capabilities more cheaply, does that in fact lead to Jevons Paradox – i.e., the less expensive that computational capacity is, the more of it we will use.

As far as the first claim, that remains to be verified. However, if it’s remotely true, it could dramatically change how much the current energy-intensive, brute force approach is applied to LLM model development in the future. That would bring energy consumption figures way down, though nobody knows by quite how much – this is all too new.

The second claim also may not stand up to further scrutiny. As noted, some anecdotal evidence I have seen suggests that DeepSeek is not really that good at answering some simple questions. And OpenAI has made some claims that need to be verified. What is thrue is that the model is pretty good. A New York Times tech reporter that spent half of the past Monday playing  with the tech came away impressed, noting that it compared well with OpenAI’s ChatGPT and Anthropic’s Claude. It solved some complex math, physics and reasoning problems at twice the speed of ChatGPT, and its responses to computer programming questions were “as in-depth and speedy as its competitors.” It wasn’t quite so good at composing poetry, planning vacations, or coming up with recipes, but so what? If it’s almost as good, at a fraction of the price…well. So, it looks like there’s a “there” there.

The next question then comes down to use, or so-called “inference.” DeepSeek is free, and it was the most frequently loaded App last week. As defined by Perplexity.AI as “Inference involves using the patterns and relationships learned during training to solve real-world tasks without further learning. For instance, a self-driving car recognizing a stop sign on an unfamiliar road is an example of inference.” Provision of that response to my query was also an example of inference (see what I did there?).  

Inference can help with real-time decision making, and it involves a number of steps: 1) Data Preparation; 2) Model Loading; 3) Processing and Prediction, and 4) Output Generation to give you the information or results you seek. Inference is very energy-intensive, so if we use less on LLMs but they get cheaper and more ubiquitous, what does that mean for energy consumption in that arena? We are so early into the adaptation and adoption of these tools that nobody knows.

But as far as the electricity required, we could be in the midst of a typical Gartner hype cycle, such as the one we experienced in the early 90s Dot.com frenzy – when Pets.com’s sock puppet was going to dominate the dog food industry.

Admittedly, 25% of Dominion Energy’s demand in Virginia is dedicated to serving data centers. And AI will clearly have many uses, some of which we can only imagine today. But the LLMs may run into various limits with declining economies of scale that would eventually reduce expected demand. There will also be substantial gains in processing and cooling efficiencies that drive energy requirements down, and we will probably see those results in years to come. Right now, we are in the early days of throwing money, a first version of chips, and data at the opportunity. But checkbooks and coffers are not limitless and a focus on efficiency will eventually follow – it always does.

There will also be companies that don’t survive the race that will probably be dominated by only a few deep-pocketed participants (although scrappy low-budget start-up DeepSeek suggests that perhaps an oligarchy is not inevitable). If this goes the same way the search engine race did, we will be left with only a small number of well-resourced players. This LLM quest may yield similar results, with most companies failing or being consolidated, and If you don’t believe me, you can go Ask Jeeves.

There’s also a big issue related to these headline demand numbers: the data companies may be filing many more applications than they intend to actually develop, because of the way the process for connecting with the utility actually works. Only a small number of utilities actually have rigorous procedures for evaluating the applications to ensure they are likely to get to physical service. The best ones, like seasoned veteran Dominion Energy, require proof of control of land, a financial commitment from the data company to support required engineering studies, and signature of a Construction Letter of Authorization obligating the applicant to pay for all project-related expenditures regardless of whether the project eventually breaks ground. Only then does an Electric Service Agreement (ESA) get signed that makes its way into the forecast. In fact, the Dominion 2030 forecast is for less load than is actually covered by ESAs.

A review of various forecasts in other parts of the country demonstrates that this same level of rigor is not routinely applied. Thus, it is quite likely that data companies are submitting multiple interconnection requests. Many data companies are likely doing what you are I would do if we needed lots of juice as fast as possible. We’d submit multiple applications to numerous utilities, with the hope that at least some of the applications would “get to Yes.”

It’s not possible to gain insight into what exactly is happening at any point in time, since the industry is competitive and maintains a high degree of confidentiality. But it’s very likely that there are numerous place-holder phantom requests. 

The analogue on the bulk power supply side of the industry may be instructive, where over ten thousand generation projects wait in transmission interconnection queues. If recent history is a guide, fewer than 20% of those endeavors will actually get built. 

If utilities further tighten up their load interconnection requirements, and implement more rigorous procedures that require higher up-front financial commitments, we may get a better sense as to how many real applications are out there.xx 

It’s clear that AI has real value to society, and we are beginning to see some use cases emerge, it’s also clear that we are still in the very early days, with rapidly evolving technologies and business models, and many unanswered questions. However, getting past the current hype cycle will take some time. We won’t know the full implications until we start to see some projects proceed, while others are canceled. If you don’t believe me, ask Perplexity.AI.  It tells me, “several factors suggest that only a fraction of the proposed projects will likely be completed.” Amen to that.

Peter Kelly-Detwiler
How AI is Revolutionizing the Grid: Efficiency, Reliability, and Resilience
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

How can AI help make the grid more efficient, reliable, and resilient? Today, we’ll tackle some promising use cases on the supply side, in the bulk power system. 

With transmission, AI can help with predictive maintenance. Operationally, it can help boost the performance of transmission lines by assisting certain grid enhancing technologies (known as GETs) that make more efficient use of existing infrastructure. Dynamic line rating replaces the historical method of limiting capacity based on static ratings in favor of an approach looking at actual ambient conditions. Lower temps and higher wind speeds pull heat from lines, allowing them to move more power, in some cases as much as 50% more.

That helps limit congestion bottlenecks and aids with the interconnection of more generating assets, and is most helpful to wind assets, since logically during the same periods when wind turbine output is high, that same wind is dissipating heat from the lines.

Then there’s topology optimization - opening and closing breakers to route power differently, facilitating higher utilization of assets. AI can help by more quickly assessing a wider variety of scenarios.

Then there’s interconnection, a big problem today. In 2000, it only took two years.  Planners were dealing with fewer and far larger projects – mostly big gas and coal plants, with only about 300 projects in the queue. That number is now over 10,000. AI can help cut time required to evaluate scenarios and increase the number of scenarios that can be assessed.

On the generation side, gas generators can be run more efficiently based on operating conditions rather than prescribed schedules. Algorithms applied to data from sensors can tell grid operators how hard they can run a turbine, and better understand when to take turbine out for maintenance, rather than relying on fixed schedules.

AI also helps generate longer term and more geographically precise weather forecasts which help supply asset operators refine output projections and dispatch strategies, while optimizing utility scale battery storage and dispatch as well.

Within a wind farm, AI can minimize the disruptions in wind flow affecting downwind turbines by steering wakes and optimizing output. This can cut land requirements for future wind plants by an average of 18% and up to 60%.

AI can also help advanced geothermal projects that extract heat from solid rock miles underground and use it to generate power. Machines and algorithms can tell operators where to drill, physically guide the drill bit through rock, predict reservoir behavior and determine how much heat to extract from a given area over a specified duration.

Some of these applications are already happening with AI related to machine learning. But as the large language models become increasingly powerful and more sophisticated, the ability to develop generative AI – to understand the patterns of existing data and then generate new data to improve decision-making will take us to the next level.

If these AI-driven datacenters are going to stress the grid with all this new demand, we might as well get as much value out of these new capabilities as we possibly can. 

Peter Kelly-Detwiler
Power Grab Part 5: AI’s Impact on Utilities and the Future of Power Infrastructure
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

In past sessions, we looked at AI-driven load, key value propositions for AI, growth projections, large language model training issues, and possible impacts on wholesale power prices. In this section, we’ll review impacts on distribution utilities. 

In competitive markets, the risk of poor economic outcomes is borne by shareholders. With vertically integrated utilities, financial impacts of decisions – good or bad - are passed on to ratepayers.

Let’s summarize.

  1. AI data centers are huge.

  2. They want power now. 

  3. Utilities are being asked to rapidly build supply and supporting infrastructure.

  4. There is much uncertainty as to where and how AI will actually generate a profit, and whether LLMs can keep growing at current rates.

  5. There is also much uncertainty concerning future electricity use. Rapidly growing use of liquid cooling, for example - putting servers in dielectric fluids - can dramatically reduce cooling loads. Then there are the chips. IBM in September announced a breakthrough in chip efficiency and market leader Nvidia continues to make gains here as well.

Utilities face requests to build gigawatts of new generation assets and transmission infrastructure with timeframes that may not line up with AI load being served. Gas plants, e.g., may have 30-40 year lifespans over which amortized. 

If AI business models change and demand for those assets evaporates, then ratepayers will pay the consequence. 

The AEP Ohio rate case highlights the issues around transmission. AE wants to manage risk, requesting large upfront financial commitments. The data companies are pushing back. Walmart and other end users are weighing in, since they, too, may be affected. New data load could crowd out other economic development, while increasing rates.

One study out this week suggests an increase of as much as 70 percent in the next decade.

Utility regulators will have to be informed in ways they’ve never had to in the past, instituting new safeguards, with tens or hundreds of billions of dollars of ratepayers’ money at stake.

Peter Kelly-Detwiler
Power Grab: Data Centers and the Grid Part 4 - AI & Wholesale Power Markets
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

Updates since last week:

Vistra is in discussions with two large companies concerning sale of power from existing gas and nuclear facilities, with plants in ERCOT and PJM eliciting interest. 

PG&E and real estate developer Westbank plan to develop a district energy system combining three new data centers – using up to 200 MW - and 4,000 residential units in San Jose. 

Modular nuke start-up Oklo has LOIs and will partner with two major data center providers to deliver up to 750 MW.

Sharon AI and industrial gas company New Era Helium Corp announced LOI to form JV for design, development, and operation of a 90MW data center in New Mexico’s Permian Basin.

Bloom announces deal to sell up to 1 GW of fuel cells to AEP Ohio.

Gas pipeline company Energy Transfer increases estimate of gas  in discussion w/datacenters from 3 bcf/day to 10 bcf.

Today, we’ll discuss potential implications of AI load for other electricity users in wholesale markets. Next week, we’ll cover distribution utilities.

One more thing: just to cast some high-level doubt into the equation. Recent comments and indicators from the industry hint that gains from language learning models may be slowing down. This is worth watching: it implies simply throwing stronger chips and more energy at the problem may not be enough. 

But assuming the AI language learning model training effort will continue at the current pace. If so, what are likely impacts to wholesale markets?

Some markets will be affected much more than others. ISO-NE and NYISO likely won’t see that much impact as it’s hard to site many datacenters or new power plants to supply them.

PJMis another story. In January, PJM had tripled its forecasted growth numbers from 2023, well before many new utility announcements. Dominion’s service territory is crowded, but new load is going to other regions: New Jersey’s PSEG just filed an updated forecast to PJM in late October, w/datacenter load growing from 343 MW to 1,196 by 2030. Exelon in late October requested of PJM a Large Load forecast adjustment of 2,600 MW for 2029.

The mid-Continental ISO – MISO – is not seeing quite the same pressure, but its CEO noted this month that 2.5 GW of new datacenter load showed up, and over 4,600 MW of data center load is noted in MISO’s July 30th Existing Large Load and New Load Additions Update

Texas is seeing huge demand growth. Oncor alone reports 59,000 MW of requests in a system peaking at 85,000 MW this summer. 

SPP and California? SPP is not a big datacenter market – yet. But it’s growing. North Dakota, may see massive growth from just two undisclosed companies, starting at 500 to 1,000 MW and growing 10x from there, w/up to $250 bn in investment. Meanwhile, California’s power prices may be just too high, and land too expensive to attract much AI load. 

The pressure is most likely going to be centered on the competitive markets in Texas and PJM.

But power prices will be affected everywhere because of equipment limitations, esp. transformers and switchgear, and related raw materials.

Then add the fact that interconnection queues are slowing down, and supply is simply unavailable. So, prices are headed up – more in some markets than others. Consulting group Bain & Company projects average annual cost increases of 1% annually. These increases will be more concentrated in areas of rapid growth, but nowhere will this dynamic be deflationary.

In the next and final session, we’ll talk about the distribution utilities, the enormous uncertainties facing them, and the inherent risk of overbuilding in an extremely uncertain AI world.

Peter Kelly-Detwiler
Power Grab: Data Centers and the Grid Part 3 - Supply Strategies
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

"Previously, we examined the enormous projections for datacenter growth and the energy required to serve that potential demand. We then discussed issues related to chips, power draw, availability of data, and other issues affecting future growth prospects.

Today, we’ll look at potential supply options for datacenters.

Of all factors limiting datacenter growth, power is the big one. Today, most of that energy is used for training – the computers crunching through data to make connections and create intelligence.

Processing and inference are estimated to consume about 70-80% of the power used. In training, most of all the energy is used in the initial run model – which can take weeks or longer. 20-30% of that power use is dedicated to cooling.

The amount of energy used for “inference” - the outputs or decisions from the trained models – is growing rapidly. Inference usage adds up. Questiond to Chat or Perplexity or a host of other platforms burn through about 10x the energy in standard Google search. Training is initially much larger than inference in terms of electricity use, but the latter grows over time.

So where to get the power? Globally, there’s a preference for the U.S. with its large grid, stable economy, rule of law, and access to communications cables. Europe’s grid is old, and space is at a premium, so some datacenters will locate in Malysia, Singapore, Brazil, anywhere they can get power. China is its own case and growing rapidly.

AI increasingly becomes a national security issue - since the release of ChatGPT two years ago, the U.S. Department of Defense has awarded $670 mn to over 300 companies working on AI-related projects.

In the U.S., supply strategies are:

1) The grid.

2) Existing assets, such as nuclear. Amazon Web Services accessed 300 MW of Talen’s nuclear plant before the FERC rejected plans for an additional 660 MW. Constellation and Microsoft are planning on a 20-year 835 MW agreement to resuscitate Three Mile Island Unit 1. Nextera recently is in talks with the feds and conducting engineering assessments as they eye restarting a 600 MW nuke in Iowa. But only so many existing and recently closed nukes to go exist.

3) Some may opt for fuel cells. In 2024 Bloom Energy expanded a 6.5 MW supply agreement to an Intel datacenter and announced another 15-year 20MW deal with Amazon Web Services. But total quantities will be relatively small in the big picture. 

4) Advanced/enhanced geothermal will help: Google has a 3.5 MW deal with developer Fervo Energy in Nevada and committed to buying 115 MW of energy from Fervo under a contract with utility NV Energy. Sage Geosystems announced a 150 MW agreement with MetaThese first large projects won’t come online for a few more years, and the industry isn’t going to see 10s of GWs anytime soon.

5) Modular nuclear? Small reactor company Oklo inked a 500 MW deal with colocation company Equinix.  Google signed a deal with start-up Kairos Power for 500 MW between 2030 and 2035. Amazon Web Services announced an investment in modular reactor company X Energy, and an agreement for 320 MW with an option to expand to 960 MW. Delivery date: early mid-2030s.

W/exception of NuScale, no players even have design approval from the Nuclear Regulatory Commission.

The logical outcome is to bypass the grid entirely and go right to the gas pipeline. Numerous gas pipeline companies report discussions with datacenter operators for direct gas hook-ups to support behind-the-meter generation. Energy Transfer is in discussions with datacenters for new demand in excess of 3 Bcf/day.

The picture is still murky since it’s early days, but all things equal, new AI load whether directly supplied or grid-dependent will likely raise prices for everybody. In our next session, we’ll discuss why.

Peter Kelly-Detwiler
Power Grab: AI Data Centers and the Electric Grid - Part 2
Podcast Streaming Services

Energy Future: Powering Tomorrow’s Cleaner World

Peter Kelly-Detwiler

Energy Future: Powering Tomorrow's Cleaner World invites listeners on a journey through the dynamic realm of energy transformation and sustainability. Listen to this podcast on:

“Let’s start with some new data points from last week:

First, Exelon’s CEO says it has seen what he terms “high probability” datacenter load jump from 6 to 11 GW this year

Google’s CEO indicated that over 25% of new code in the company is generated by AI and then reviewed by engineers. AI substitution for human labor is a huge part of the AI value proposition.

The Financial Times estimates spending on AI datacenters for the big four - Alphabet, Amazon, Meta, and Microsoft will exceed $200 billion this year.

The FERC rejected a request from Amazon Web Services to expand a contract involving power supply to its data center co-located at Talen Energy's Susquehanna nuclear power plant in PA. 

Now, let’s talk compute: the goal here for competitors is to improve the quality of their language learning models so they can deal with more complex logic and increase overall accuracy. They do this by training on lots of data with increasingly powerful machines. 

Analyst Epoch AI notes that annual compute capability has recently been growing at a rate of 4X. Will this growth will continue at that torrid pace, and what the implications are for our power grids?

Epoch AI looks at this issue by examining four underling factors: 1) power availability – our sweet spot that we will talk a lot more about later; 2) global chip capacity; 3) the “latency wall,” delays in increasingly complex computations; and 4) the availability of data to train on. Let’s look at 2 through 4 - we will deal with power in its own session. 

Chips are in high demand. These game processing units – GPUs – bring power and parallel processing to the game, performing highly complex calculations at rapid speeds. GPUs keep getting better, but are in high demand and also expensive. Nvidia’s newest Blackwell chip cost about $10 billion to design and create, and buyers are paying $30,000 to 40,000 per GPU. That same Blackwell chip draws between 700 W and 1.2 kW depending on the configuration and cooling strategy. 

Nvidia owns about 80% of GPU market share, followed by AMD, and the industry cannot keep up with current demand. But Google, Amazon, Meta, and Microsoft are all at work developing their own chips, so that strain may eventually ease.

Next, let’s look at the “latency wall.” It takes a certain amount of time (latency) for an AI model to process each datapoint, and that latency increases as model sizes grow. Models train by separating data into batches, and each AI training run takes as long as is needed to process a batch. The more batches processed, the longer the run takes. Today’s latencies aren’t that big – batches can be processed quickly. But as future training runs get larger and models get bigger, this could become an issue, and efficiencies might fall off. This scaling issue may limit future growth rates.

Finally, let’s look at data. AI datacenters train on data. Everything we ever posted to LinkedIn, Facebook or Insta. Youtube videos. Scientific papers, Movies, TV shows, stupid clips on TikTok. All of it. To understand data, we must understand the concept of a token - the smallest element into which text data can be broken down into in order for the AI model to process it. One word is usually a single token. With images, audio clips or video,s computers typically break them into smaller patches for tokenization (one picture or one second of video might represent 30 tokens).

It’s estimated that the web holds about 500 trillion words of unique text, which may grow 50%. by 2030. Add in images, audio, and video and you might get to 20 quadrillion tokens for computer training by the end of the decade. BUT, projections are that with ever faster computers and more efficient algorithms we might actually run out of data to train on, even as soon as 2026. Then, machines may learn to generate their own synthetic data. Or they could find other ways to learn. Nobody really knows. This uncertainty leads to a critical question for utilities. What if we build all this infrastructure, and then by 2030, there’s less to do with it? The phrase “stranded assets” should come to mind.

Meanwhile, chips become increasingly more efficient, requiring less electricity for processing and addressing the waste heat. Nvidia says its GPUs used in training have seen a 2000x reduction in energy use over 10 years. Until now, such gains have allowed data centers to do more and their appetite appears endless. But if future gains continue, how does that affect future datacenter power needs? Nobody truly knows. What we do know now is that the power grab continues unabated, and data centers are looking at all kinds of supply strategies to get the juice wherever they can. And that’s the topic we will focus on in the next session.

Peter Kelly-Detwiler