Data Center Intelligence - Weekly Roundup (Dec 22-28)
December 31, 2025
The week did not feel like a quiet slide into year end. Big energy and infrastructure deals continued, new AI campuses moved from concept to concrete, and the power and policy conversation around data centers grew sharper. From an FP&A perspective the sector still looks investable, but the bar for understanding power, politics, and counterparties keeps rising.
I have grouped the week into four lenses: industry trends, future expansion, green energy and environmental moves, and government policy.
1. Industry trends
Capital flows, ownership shifts, and customer commitments
Google to acquire Intersect Power for clean energy and data center scale
Google agreed to buy United States renewables developer Intersect Power for about four point seven five billion dollars. Intersect brings several gigawatts of solar and storage projects and a pipeline that explicitly targets power hungry AI and data center demand. Data Center Dynamics+1
Why it matters
This is a tech company buying part of its fuel supply. Instead of only signing power purchase contracts Google is buying the developer and its pipeline. For FP&A teams this signals more vertical integration by large cloud players and a future where energy and infrastructure show up directly on tech balance sheets rather than only in off balance sheet contracts.
Goodman and CPP Investments commit eight billion dollars to a European platform
Goodman Group and CPP Investments launched an eight billion dollar data center development platform targeting France, Germany, the United Kingdom and other core European markets. They plan to use existing Goodman industrial land and logistics corridors as the foundation for multi campus AI ready builds. Baxtel
Why it matters
This is logistics capital scaling into digital infrastructure. For people outside the sector it is a sign that the same firms that built modern warehousing for ecommerce now want to host AI factories. For FP&A teams this provides a new benchmark for committed capital in Europe and reinforces that site control in power rich industrial corridors is a long term competitive edge.
AtlasEdge recycles capital by selling nine non core sites
AtlasEdge agreed to sell a nine site portfolio across several European countries to Templus in a deal valued around sixty five million euros. The sites are smaller edge assets that no longer fit AtlasEdge strategic focus. Proceeds will be recycled into higher growth campuses. Baxtel
Why it matters
This is the equivalent of a hotel chain selling older roadside motels to spend more on full service city hotels. For FP&A teams it underlines that portfolio pruning is now part of the data center playbook. Expect more sales of small legacy assets to fund larger AI oriented developments and adjust valuation approaches accordingly.
Nscale lands an eight hundred sixty five million dollar AI pre lease in North Carolina
Developer Nscale announced a major customer commitment and power partnership in North Carolina. A confidential hyperscale tenant has pre leased a forty megawatt block at a new campus and Nscale has teamed up with WhiteFiber to deliver power and fiber. The overall project value is estimated around eight hundred sixty five million dollars. Baxtel
Why it matters
This is a long term rental agreement for a very power hungry building before the building exists. For FP&A teams these types of pre leases are the cornerstone of underwriting. They support earlier debt and equity commitments but also lock in high density design choices. When you model regional supply you should treat this as effectively off the market before the shell is even complete.
New Era Energy and Digital secures full control of Texas Critical Data Centers
New Era Energy and Digital reached a binding agreement to buy out its partner’s interest in the Texas Critical Data Centers campus in Ector County Texas and closed on an additional two hundred three acres of land. The site is positioned as an AI data center campus with access to significant power. Business Wire+1
Why it matters
This is smaller than the mega deals but signals how energy developers are moving into direct ownership of AI infrastructure. This looks like a power company deciding to build and own part of the factory that uses its electricity. For FP&A it suggests more vertically aligned platforms and more complex risk sharing between generation and data center arms when you review counterparties and capital structure.
2. Future expansion
Land banking, campus concepts, and the geography of AI buildout
Texas AI campus in Ector County moves from concept toward execution
With full ownership and more land in hand New Era is positioning the Texas Critical Data Centers campus near Odessa as a multi phase AI site that can scale with regional transmission and generation upgrades. The location is close to existing energy infrastructure and transmission corridors that historically served oil and gas. Business Wire+1
Why it matters
This is a template for repurposing energy country real estate for AI. For FP&A teams tracking the United States map the Permian region is not only an oil story anymore. Treat Ector County as an early signal that more AI campuses will follow where transmission and substations already exist even if they are far from major cities.
Jet AI announces joint venture for a fifty megawatt campus in Nevada
Jet AI announced a joint venture with Choo Choo Express to develop a planned fifty megawatt data center campus in Moapa Nevada just outside the Las Vegas area. The project is framed explicitly around high performance GPU infrastructure and AI cloud services and is valued around five hundred million dollars. GlobeNewswire+1
Why it matters
This is an AI startup deciding to build its own small factory rather than only renting from larger landlords. For FP&A this raises questions about credit quality versus capacity ambition. When you forecast regional supply treat these smaller sponsor projects as higher variance in timing and scale compared with hyperscaler builds.
Desert land in Arizona positioned as the next major AI corridor
A detailed feature on Hassayampa Ranch in rural Arizona described a plan to turn about two thousand acres of farmland into a cluster of hyperscale data centers served by existing gas pipelines and proximity to the Palo Verde nuclear plant. The project is still at the early stage but reflects how developers are chasing power near traditional generation assets rather than metro cores. Fortune
Why it matters
This is what it looks like when the cloud leaves the city and moves into the desert near its fuel source. For FP&A teams Arizona has to move up the priority list as a long term AI training and compute corridor. Model a slow initial ramp with the potential for very large step changes in capacity once transmission and water questions are resolved.
Virginia and the Atlantic corridor remain crowded but attractive
Industry roundups this week highlighted continued momentum at large Virginia projects including the nine hundred megawatt CleanArc campus in Caroline County, which aims to deliver three hundred megawatts by early 2027, and ongoing site marketing across the broader Mid Atlantic region. These projects are explicitly framed around AI and cloud loads and depend heavily on grid upgrade schedules. DataCenterKnowledge+1
Why it matters
Think of Virginia as the main airport hub that keeps adding gates even as the airspace gets busier. For FP&A teams do not assume Northern Virginia growth is over simply because constraints are real. Instead treat it as a region where timing and power access will drive very different outcomes for each project rather than one uniform trend line.
Europe and the Nordics see a wave of campus concepts
A European weekly recap described new or refreshed plans for a one point four gigawatt site between Paris and Lille, a sixty six thousand square meter campus in Vaasa Finland, a large project in Valer Norway, and an NTT campus in Nierstein Germany with up to four hundred eighty two megawatts. datacentrenews.substack.com
Why it matters
This is the European version of the United States Midwest and desert story: big AI campuses moving into cooler climates and power rich regions. For FP&A teams this means your five year models for EMEA need to treat France, Germany, Finland, and Norway as potential multi campus clusters, not niche outposts.
3. Green energy and environmental builds
Power strategy, generation mix, and environmental pressure
Older power plants find new life serving AI data centers
A widely cited report explained how rising AI driven electricity demand is making it attractive to keep some aging fossil fuel plants online longer or consider bringing them back into service. The piece tied record capacity prices in PJM and the tight supply demand balance to data center growth in states like Virginia and Pennsylvania. Reuters+1
Why it matters
This is the tradeoff behind the scenes of your AI tools. Cleaner grids are colliding with huge new loads. For FP&A teams this means power price assumptions in core regions need to move higher and stay higher for longer, with an eye on future regulatory moves that could force earlier retirement and push operators toward new sources like nuclear and storage.
Proposal to use retired Navy reactors for a Tennessee data center
A nuclear developer filed an application to use two retired naval reactors to provide around the clock power for a proposed data center project at Oak Ridge Tennessee. The concept would deliver roughly five hundred megawatts of steady electricity under a White House program aimed at pairing advanced generation with AI loads. Energy Connects
Why it matters
This is how far planners are willing to go to chase dependable cleanish power for AI. For FP&A teams this signals a future where some sites blend characteristics of utility assets and commercial campuses. These hybrids will have very different risk profiles, permitting paths, and financing structures than standard merchant power plus data center arrangements.
Distributed energy and prepaid solar plus storage deals gain momentum
Utility analysts highlighted how data centers and other large loads are using upfront lease payments and long term offtake commitments to unlock new solar plus storage projects, especially in power constrained regions. The theme is that data center demand is becoming a catalyst for distributed energy resources rather than only competing with them. Utility Dive
Why it matters
This looks like large tech customers paying for grid upgrades and batteries in advance. For FP&A this broadens the toolkit. Instead of treating power as a fixed utility line item you can now model structured agreements where prepayments reduce long term cost volatility but increase capital intensity in the early years.
Environmental focus sharpens on AI water and emissions impact
Public broadcast coverage this week revisited the water use and emissions impact of AI data centers, stressing that site choices and the power mix can either mitigate or magnify local environmental stress. Examples ranged from desert campuses with high cooling and water loads to Great Lakes projects where freshwater is abundant but community scrutiny is increasing. wqln.org+1
Why it matters
This connects the idea of the cloud to very physical tradeoffs in water basins and regions already under stress. For FP&A teams water risk and emissions intensity should now be explicit inputs in site scoring and long term operating expense assumptions, not footnotes.
Asia Pacific sites secure longer leases and land positions
Industry trackers noted that Keppel obtained a ten year land lease extension for a major Singapore data center campus in a transaction valued around one point three eight billion Singapore dollars and that AirTrunk announced a second Melbourne campus with an investment plan of around five billion Australian dollars. Blackridge Research
Why it matters
This shows that Asia Pacific is not just catching up but locking in long duration sites for AI and cloud. For FP&A teams the lesson is that long land tenure and pre arranged power access are becoming strategic assets of their own. When you value platforms in Singapore and Australia, the security of tenure and ability to expand in place deserve more weight.
4. Government policy and community response
Rules around power, costs, and where AI is allowed to grow
FERC clarifies rules for direct data center power plant connections
The Federal Energy Regulatory Commission issued an order that clarifies how very large loads such as data centers can connect directly to power plants in PJM and similar markets, including guidance on cost sharing and rate treatment. This follows growing interest in pairing AI campuses with dedicated generation. AP News+1
Why it matters
This is the referee updating the rulebook so big factories of compute can plug straight into their own power stations. For FP&A this opens a path to integrated generation plus data center projects but also adds regulatory complexity. You will need to model separate but linked cash flows and be ready for public debate about who benefits from these arrangements.
Georgia regulators approve a gas heavy plan tied partly to data centers
The Georgia Public Service Commission approved a long term plan from Georgia Power that includes five new methane gas plants and about ten gigawatts of new capacity, with anticipated data center growth cited as a driver. Environmental groups criticized the decision and warned about customer protections. AP News
Why it matters
This is a state effectively betting on more gas to keep up with AI and cloud demand. For FP&A teams it signals that some regions will accept higher emissions in exchange for reliability and growth. When you evaluate projects in those states you should consider both the probability of ample future capacity and the risk of political backlash that could later add costs or restrictions.
PJM capacity prices hit records as data center demand soars
PJM Interconnection reported record high prices in its latest capacity auction, driven in part by surging demand from data centers in Mid Atlantic and Midwestern states and a slower than needed pace of new generation. Capacity prices jumped about one thousand percent over the past two years. Reuters
Why it matters
Capacity prices are the fees paid to power plants just to be ready to serve demand. When these prices spike, customer bills eventually follow. For FP&A teams this is a strong data point that wholesale power costs in PJM are structurally higher going forward. Any long term model that assumes flat or gently rising energy costs in these regions is likely underestimating risk.
Pennsylvania advances a model tariff for large load customers
Pennsylvania regulators continued to advance a statewide model tariff for large load customers such as data centers, with public comment running through December 22. The proposal would require these customers to shoulder more of the infrastructure costs they trigger and to contribute to low income bill assistance programs. GovTech+2PSATS Main+2
Why it matters
The idea is simple. If a huge new facility forces the utility to build expensive upgrades, that facility should pay most of the bill. For FP&A teams this marks an emerging pattern. Expect more states to adopt similar models. Future projects will need to carry larger dedicated grid upgrade costs in their capital plans, but may enjoy clearer rules and faster approvals in return.
Local governments move to limit or proactively regulate data centers
County and borough level actions continued this week. Frederick County Maryland adopted a map that limits where data centers can locate, while Mayfield Borough in Pennsylvania passed a zoning ordinance aimed at proactively regulating data centers before specific proposals arrive. Facebook+1
Why it matters
This is local government trying to get ahead of a wave of proposals rather than being surprised by them one at a time. For FP&A and site selection teams it confirms that entitlement risk is now a moving target at the township and county level. Even in supportive states you should build scenario plans that account for local caps, buffer zones, and new permitting steps that appear mid project.
What this week means for FP&A and executive planning
Across these stories the through line is clear. AI is still pulling enormous capital and creativity into digital infrastructure, but every additional megawatt now has to clear more conditions.
For FP&A leaders the takeaways are:
- Energy is now strategy not backdrop
Large tech buyers are acquiring developers, restarting nuclear plants, and backing novel generation projects. Your models need to integrate power procurement, grid upgrades, and long term price risk into the core business case, not as a line item. - Geography is splitting into metro interconnect hubs and energy native corridors
Northern Virginia, Singapore, and other hubs remain essential, but Arizona deserts, Texas energy country, and Nordic power basins are rising as long term AI training locations. Forecasts should treat these as complementary rather than mutually exclusive. - Policy and community risk are becoming quantifiable costs
Tariff proposals, county maps, and local ordinances are no longer background noise. They shape capital intensity, timelines, and even whether projects can proceed. Build scenario ranges around regulatory outcomes the same way you do for occupancy and pricing. - Ownership and counterparties are diversifying
Asset managers, utilities, and newer AI platforms are joining the mix. Underwriting needs to go beyond brand names and focus on balance sheet depth, contract structures, and alignment between generation and load.
The next five years will reward the teams that can connect these dots inside a single model and update it continuously as power markets and policies shift. FP&A is moving from a scorekeeper role to the operating system that keeps land, power, capital, and community realities in sync.
“The content is based on public information and personal analysis. This is not financial or investment advice.”
