The AI data center buildout: the power edge
The AI buildout is the biggest infrastructure project of the decade. But what does it actually do — to power grids, to materials, to jobs, to your electricity bill? We traced the connections, sourced every figure, and tried to explain each link in plain language. Where we don't know, we say so.
The thesis moved from mid-cycle to mid-cycle and levered. Capital is accelerating into the buildout even as the binding constraint shifts from chips to sites: Sesterce’s $11.21B Finnish data-center bet lands in the same country, the same week, that Finland halted Google’s site work over environmental assessments — permission is the co-constraint alongside electrons, and water is joining it. Google’s answer is to leave the ground entirely: a 4-TPU prototype satellite (Project Suncatcher) is testing whether compute can outrun the terrestrial permission grid. The build is now financed on credit as much as on cash: Tencent mulls a $5B AI-infra bond days after SoftBank’s ~$11.1B junk raise, with 2026 AI-debt issuance already at $489B and the credit market ‘splitting in two’ — watch credit, not compute demand, as the risk layer. Governments keep competing for compute (India’s tax break already drew CoreWeave’s 240 MW; Uttar Pradesh now adds a 50-department space+AI program), and defense AI keeps converting to standing contracts (NGC2 to the Pacific, Shield AI’s exportable X-BAT, a $99M Army C2 IDIQ, a $205M DCSA award). The a16z ‘Age of Atoms’ line stands: the cycle’s leadership has moved from bits to atoms, and hyperscaler free cash flow stays depressed through ~2028.
See how this read evolved →The players
Who's driving the buildout — the labs training the models, the chipmaker, the hyperscalers writing the checks, and the power side keeping up. Click a name for context.
AI labs
Chipmakers
Hyperscalers & cloud
Power & grid
By the numbers
Four charts, each connected. Read them left to right, top to bottom — the blue annotation under each one explains how it connects to the next.
The five largest tech firms spent $400B+ in 2025 and are set to raise it ~75% in 2026 — most of it AI infrastructure. [11] IEA
This spending builds the data centers that consume the electricity above — and drives the price spike below.
From $28.92 to the $333.44 cap in three years; the grid's market watchdog pinned ~63% of the jump on data centers. [12] RTO Insider [5] Utility Dive
This cost gets passed to electricity bills — that's where regular people start paying for the AI buildout.
Thousands to build (~800–1,200 at peak); ~25–50 to run it. Billions of capex, dozens of permanent staff. [8] Brookings
The construction boom is real but temporary — and it pulls workers from housing and other projects that also need them.
Numbers to know
Every number that matters for understanding the buildout. Click a row to inspect it — or read the "Why it matters" column first if the metric is unfamiliar.
| Metric | Latest | Trend / projection | Why it matters | Confidence | Source |
|---|---|---|---|---|---|
| Global data-center electricity | ~485 TWh in 2025 (+17% YoY) | → ~950 TWh by 2030 (~3% of world power) | This is the raw demand signal — everything else on this page flows from this number going up. | Official | [11] [1] |
| US share of US electricity | ~4.4% in 2023 | → 6.7–12% by 2028 | Data centers are going from a rounding error to a meaningful fraction of America's total power use in just five years. | Official | [2] |
| Big Tech capital spending | >$400B in 2025 (5 largest tech firms) | → up ~75% in 2026, mostly AI infrastructure | This is the money that funds the buildout. When this goes up, electricity demand, construction jobs, and materials demand all follow. | Official | [11] |
| PJM grid capacity price | $333.44/MW-day (2027/28, at the cap) | from $28.92 in 2024/25; ~$530 uncapped | This is the cost of reserving future electricity in the eastern US. It 12x'd in three years — the clearest signal of data centers competing with everyone else for power. | Press-reported | [12] [5] |
| California 'who pays' package | 7 bills signed Sep 22, 2026 | data centers must fund their own grid & water upgrades | The clearest 'who pays' rule in the US so far: new rate classes, grid-cost responsibility on operators, and energy/water disclosure — instead of passing costs to households. | Official | [19] |
| Generation waiting to connect | ~2,300 GW in US interconnection queues | ~4.5-year average wait | New power plants need years just to plug into the grid. This bottleneck means demand is growing faster than the system can add supply. | Official | [4] |
| Nuclear restarts for AI | Three Mile Island: 835 MW, 20-yr Microsoft PPA | back online ~2027 | Companies are restarting retired nuclear plants to get clean, reliable power — a move nobody expected five years ago. | Official | [3] |
| Copper demand | ~330k–500k tonnes/yr by 2030 | ~2% of global demand | Every data center and power line needs copper. AI is one of several reasons copper is getting more expensive — alongside EVs and housing. | Proxy / estimate | [6] |
| Power transformers | ~30% US supply deficit (2025) | ~2–4 year lead times | You can't connect a data center (or anything else) to the grid without transformers. There aren't enough, and the wait is years. | Proxy / estimate | [7] |
| Jobs per ~100 MW campus | ~800–1,200 building it (peak) | vs ~25–50 permanent staff | Massive construction boom, tiny permanent workforce. Great for electricians now; the long-term local job impact is much smaller than the construction phase suggests. | Press-reported | [8] |
| Compute contracted | Stargate $500B / 10 GW; Anthropic–Google up to 1M TPUs | tens of $B; >1 GW online in 2026 | Computing power is being bought and sold in advance, like oil futures. The scale of these contracts is what's pulling all the other numbers on this page. | Press-reported | [9] [10] |
| Nvidia → OpenAI | Up to $100B for ≥10 GW (~4–5M GPUs) | first GW H2 2026; Nvidia funds ~$10B equity per GW — 'circular financing' | The chip supplier is funding its biggest customer's purchases. This circular arrangement is either an ecosystem bet or a bubble warning, depending on who you ask. | Press-reported | [13] [14] |
| Largest single AI site | SpaceXAI (ex-xAI) Colossus: ~555k GPUs, ~2 GW, ~$35B (Memphis) | world's largest AI training site; xAI merged into SpaceX → SpaceXAI (2026) | One facility drawing 2 GW — roughly what a mid-sized city uses. This is the physical scale of AI training made concrete. | Proxy / estimate | [15] [18] |
| AI datacenter power (SemiAnalysis) | ~20 GW of new capacity energized in 2026 | +~30 GW expected 2027; US AI power ~3→28 GW (2023→26) | The buildout isn't slowing down — it's accelerating. US AI-specific power capacity grew nearly 10x in three years. | Proxy / estimate | [16] |
| AI-lab revenue vs. capex | Anthropic ~$65B, OpenAI ~$40B run-rate (mid-2026) | against $400B+ (2025) / ~$700B (2026) capex — the 'bubble' gap | The entire AI industry earns less in revenue than tech companies are spending to build it. Whether this gap closes or crashes is the single biggest financial question behind the buildout. | Press-reported | [17] |
| Construction crowding-out | Private data-center construction +$9B vs. all other private construction -$46B | through July 2026 (Commerce Dept, seasonally adjusted, YoY) | The data-center buildout is the bright spot in an otherwise dark construction year — it is visibly pulling electricians, land, and capital away from everything else being built. | Official | [20] |
| Industrial power displacement | Mississippi aluminum smelter (~1,000 jobs) relocated to Oklahoma | after a nearby data center locked up the electricity it needed | The first concrete 'who pays' casualty of power scarcity: a manufacturing project with a thousand jobs lost its electricity to a data center. This is the who-pays fight made physical. | Press-reported | [20] |
| DOE grid modernization for AI | ~$2B for 31 projects across 26 states | +23 GW capacity; PPL peak demand could quintuple by 2032 | Federal dollars are now flowing specifically to squeeze more electricity out of the aging grid for AI demand — Washington is treating grid capacity as an AI bottleneck, not a background utility problem. | Press-reported | [21] |
| Private-grid gas power for AI | Atlas Energy: 283 MW Caterpillar gas equipment committed + 328 MW framework | a 'leading frontier AI lab' reimburses equipment costs | Frontier labs are now underwriting their own power plants to escape grid queues — de-risked, behind-the-meter generation is becoming part of the AI stack. | Press-reported | [22] |
| Data-center construction pace | $59.3B/month by May 2026 (Census Bureau) | +129% from two years prior, no plateau | Construction dollars are the physical footprint of the AI boom — and the leading edge of where electricity demand lands next. | Press-reported | [23] |
| On-site fuel-cell economics | Bloom 800V DC claims -$3.6B (27%) non-compute capex on a 1 GW AI campus; $5.5B (9%) 5-yr TCO edge (Bloom-reported) | Oracle's $165B Jupiter campus (2.4 GW fuel-cell deal): force-majeure scare clarified as schedule risk, not cancellation | Grid queues are so long that on-site power is becoming a cost decision, not a green choice — and a single campus's schedule slip can move the stock. | Press-reported | [24] |
| 2030 grid supply-demand gap | ~268 GW shortfall by 2030 | demand 161 GW in 2026 → 490.7 GW in 2030 vs 222.6 GW grid capacity; US deficit >170 GW, concentrated in PJM, ERCOT, MISO | Power is now the hard ceiling on the AI buildout: grids can deliver roughly half of what data centers will want by 2030 — so electricity prices in hotspot regions are where the AI bill lands first. | Press-reported | [25] |
| US data-center power demand trajectory | 31 GW (2025) → 41 GW (2026) → 66 GW (2027) | 8.5% of peak summer US power demand in 2027, up from 4.1% in 2025; only 50–60% of the next 1–2 years' planned capacity expected online on time | Data centers nearly double their share of peak summer power demand in two years — that's the number that turns grid planning into a political fight. New gas plants take 5–7 years; this demand is arriving in two. The on-time-delivery figure (half, at best) is why 'bring your own power' is now a business model, not a gimmick. | Press-reported | [26] |
| Neocloud pre-commitments (Malaysia) | >70% of 21.7 MW pre-committed; >$1.7B expected revenue over 5 years | $2.9B total expected revenue backlog; energization Q1 2027; liquid-cooled GB300 NVL72, part Vera Rubin-ready | A neocloud signing $1.7B of revenue before the power is even on — demand for AI compute in Southeast Asia is so strong that buyers are contracting capacity ahead of energization. This is what the 268 GW supply gap looks like from the demand side: the queue is forming faster than the grid can connect it. | Press-reported | [27] |
| FERC suspends PJM's procurement plan | 5-month suspension from Sep 30, 2026 | large users (incl. data centers) must bear the costs of serving their demand; more current demand forecasts required | The 'who pays for the AI buildout' question just moved from debate to regulatory mechanics in the largest US power market — the federal regulator is telling grid planners to stop socializing data centers' costs onto everyone else's bills. | Press-reported | [28] |
| Ratepayer Protection Act blocked | US Senate 57–43 (Sep 30, 2026), short of 60 votes | House passed it 417–3; would have required regulators to weigh cost-allocation to data centers | Congress stalled, so the fight over AI's electricity bill moves to regulators like FERC — households keep carrying the grid costs for now. With the House vote nearly unanimous, this fight isn't over; it's changing venues. | Press-reported | [29] |
| Single-utility AI power pipeline (Michigan) | DTE Energy: >8 GW total (2.4 GW executed) | Oracle 1.4 GW approved, Google 1 GW pending; Oracle deal needs up to $2B in utility capex | One utility, 8 GW — the physical scale of AI power demand keeps compounding. Note the fine print: minimum monthly charges and collateral explicitly designed to shield existing ratepayers. This is what 'data centers pay their way' looks like in a signed contract. | Press-reported | [30] |
| Amazon–Constellation 20-year nuclear PPA | 690 MW (incl. ~190 MW new capacity 2030–32); >$3B investment at Calvert Cliffs | retail supply agreement covers Amazon across the 13-state PJM market; underwrites 20-year relicensing and new clean generation at the site | The grid's biggest customer is becoming its builder: Amazon is underwriting new baseload directly instead of waiting in the interconnection queue. Every PPA like this is a bet that AI electricity demand is a decades-long load, not a bubble spike — and it lands a day after FERC told PJM to stop socializing those costs. | Official | [31] |
| PJM backstop auction halted | Sept 30 opening cancelled; effective date pushed to Feb 28, 2027; PJM has set no new timeline | FERC's cost-allocation and collateral concerns; TD Cowen warns of 'acute uncertainty'; Constellation, NRG, Talen shares fell | The 'data centers pay their way' rule is now enforced by the auction simply not happening: until cost allocation is fixed, the largest US grid won't procure the 6.8 GW of backstop power data centers need. Markets priced the uncertainty immediately — the AI buildout's bottleneck just moved from concrete to procedure. | Press-reported | [32] |
| TVA data-center tariff takes effect | Oct 1, 2026: dedicated rate class; ~10% increase phased over 3 years; Capacity Commitment Charge for new/expanding loads >5 MW | nation's largest public utility; IRP sees up to ~32 GW new capacity needs by 2040 alongside a $13B capital package | Another 'who pays' template goes live: 10 million TVA customers won't subsidize data-center growth — the new rate class makes the AI load pay for its own grid, phased in so the charge sticks. This is the design other utilities will copy, because it works politically and financially. | Press-reported | [33] |
| AWS 'Built Together' community package | >$1B over 5 years for US host communities (education, job training, energy upgrades) | 100+ local data-center moratoriums under consideration; AWS drops NDAs with governments, publishes annual energy/water use, and pledges data-center power costs won't land on household bills | The community-opposition counterstrike: $68B of US projects were blocked or delayed in the June quarter, so the industry is now buying consent — free community college, job training, energy upgrades — and, more importantly, promising that data centers' power costs won't show up on your bill. If local opposition is the real bottleneck on the buildout (and it increasingly looks like it), this is the price of unblocking it. | Press-reported | [34] [35] |
| Local opposition to AI data centers (Gallup) | 71% oppose (48% strongly); first time Gallup asked, Mar 2026 | More unpopular than nuclear plants (53%); 50% of opponents cite environmental concerns, ~20% higher utility bills | The permission bottleneck now has a number: AI data centers poll worse than nuclear plants. That’s the demand curve AWS’s $1B “Built Together” package is buying against — and with $68B of projects blocked or delayed in a single quarter, the industry’s binding constraint isn’t chips or power anymore, it’s neighbors. | Press-reported | [36] |
| Michigan PSC approves DTE-Google 20-year power deal | 80% minimum billing demand; 20-year term (Van Buren Township) | early-termination payment covers 15 years of minimum charges; credit/collateral requirements | The ‘data centers pay their way’ template in its most muscular form yet: Google pays for 80% of contracted load even when idle, and walking away early costs 15 years of minimum charges — explicitly structured so other DTE customers don’t subsidize the load. Regulators are now pricing the who-pays risk into the contract itself. Pairs with the DTE >8 GW pipeline metric: this is the deal structure behind the number. | Press-reported | [37] |
| Google's Oct-6 utility double-deal | Black Hills $1.8B (564 MW, Cheyenne) + Constellation 3.59 GW (>$4.3B investment) | Google covers full cost of Black Hills build; Constellation: 890 MW 20-year nuclear PPA + 2,700 MW 15-year supply; CEG +13% on the day | Two utility deals in one day — and both are structured so Google, not ratepayers, funds the new capacity. This is the 'who pays' question answered at scale: the biggest AI load in the world is now directly financing both gas and nuclear generation. Constellation's 13% pop is the market pricing utilities as AI stocks — the hardware→software rotation the Street expects by 2028 hasn't touched the power layer, which remains the binding constraint. | Press-reported | [38] [39] |
| Google nearing second Constellation nuclear deal | $1B+ for ~890 MW of US plant uprates | expected as soon as this week (Oct 6); days after the 3.59 GW / $4.3B Oct 6 pact | ‘One contract: power + compute’ is now the default procurement pattern — Google is buying plant uprates, not just plants, and the follow-on landed days after the 3.59 GW deal. Pairs with India’s 20-year AI data-center tax break (citation 41): governments and utilities are now bidding against each other to host compute, while ratepayers negotiate where the bill lands. | Press-reported | [40] [41] |
| Sesterce’s $11.21B Finnish AI data-center bet: phase 1 200 MW from 2026 | €10B+ ($11.21B); Jämsä site on a former paper mill; phase 2 600 MW; >1 GW ambition | announced Oct 8, days after Finland halted Google’s site work over EIAs | Capital arrives where permission is tightening — in the same country, the same week. The binding constraint on the $1T+ buildout is increasingly sites, permits and water, not chips. | Press-reported | [44] |
| Black Hills $1.8B for Google’s Wyoming data center + Google’s 3.59-GW Constellation deal | $1.8B (2027–29), 564 MW new generation for Cheyenne; 3.59-GW Constellation contract ($4.3B+) | utility balance sheets becoming AI-infra balance sheets | The buildout’s second-order capex wave: hyperscaler demand is now written into regulated-utility investment plans — power procurement is the new capacity race. | Press-reported | [45] |
| Google’s Project Suncatcher: first 4-TPU prototype satellite launched Oct 1 | Transporter-18/SpaceX; 81-satellite solar-powered compute constellation pitch; cost-competitive with terrestrial DCs by mid-2030s | infra escaping the terrestrial permission grid | The permission-co-constraint thesis answered from orbit: if sites need permits and water, compute can go where neither is required. A prototype, not capacity — tracked as direction, not scale. | Press-reported | [46] |
| CoreWeave’s India entry: 240 MW at AdaniConneX Taloja | 240 MW (option to 480 MW); online mid-2028 | several-billion-dollar lifetime investment; India office opened; days after the 20-year AI-data-center tax break | The tax break is already working: days after India offered a 20-year tax break on overseas revenue from India-based AI data services, CoreWeave committed its first Indian capacity. Sovereign compute pull is no longer a policy paper — it’s a signed power deal. | Press-reported | [42] |
| Finland halts Google data-center site work over environmental assessments | LVV: suspend by Oct 23, 2026 at Muhos and Kajaani | ~530 ha cleared before mandatory EIAs completed; Google acknowledged shortfalls | Permission is now the co-constraint: even where grid capacity is secured (Fingrid had room), construction can’t proceed without environmental clearance. Pairs with Gallup’s 71% opposition and the $68B of blocked/delayed US projects — the buildout’s binding constraint is increasingly local consent, not just electrons. | Press-reported | [43] |
How the numbers connect
Each card below is a connection between two things — with a plain-language explanation of the mechanism, what type of link it is (direct, indirect, correlated, or circular), who benefits, and who pays. The cascade reads top-down by order of consequence; the graph shows the same connections visually.
Click any connection card to open it in the right rail — then dig deeper with Ask.
First order — the raw demand
Every AI model runs on servers, and servers need power 24/7. More data centers means more electricity consumed — this is physics, not a forecast. The scale is what's new: a single large AI campus can draw as much power as a small city.
Data-centre electricity use jumped ~17% in 2025 to about 485 TWh (~1.5% of world power) and is projected to roughly double to ~950 TWh by 2030 (IEA). In the US, data centers were ~4.4% of electricity in 2023, heading to 6.7–12% by 2028 (Berkeley Lab, 2025 update). [11] [1] [2]
Where the data runs out: Projections diverge widely — Berkeley Lab's 2028 US range alone is 6.7–12% — and AI's exact share within data-center load is uncertain.
Second order — who wins, who pays
All that new electricity has to come from somewhere. Power companies are signing long-term deals to supply data centers — in some cases, spending billions to restart retired nuclear plants. The demand is so large that new power plants are waiting years just to connect to the grid, creating a bottleneck for everyone.
Who wins
Utilities, IPPs, gas & nuclear operators
Who pays
The grid — interconnection queues & strain
Constellation is restarting the 835 MW Three Mile Island Unit 1 ($1.6B) on a 20-year PPA sending 100% of its output to Microsoft. Meanwhile ~2,300 GW of generation & storage sits in US interconnection queues, and the typical project now waits ~4.5 years to connect (Berkeley Lab). [3] [4]
Where the data runs out: Which specific plants are contracted to which campuses is largely undisclosed; queue volume actually fell ~12% year-over-year as withdrawals rose.
Data centers and the power plants feeding them are built from real physical stuff — copper wiring, transformers, turbines. AI isn't the only reason these are expensive (electric vehicles and housing need copper too), but it's adding major new demand to markets that were already strained. The connection is real but shared with other industries.
Who wins
Copper, transformer & turbine suppliers
Who pays
Every other buyer — price & lead-time pressure
AI data centres could consume ~330,000–500,000 tonnes of copper a year by 2030 (~2% of global demand). US power transformers face a ~30% supply deficit in 2025 (Wood Mackenzie), with large-transformer lead times around 2+ years. [6] [7]
Where the data runs out: Copper figures are analyst forecasts that vary; transformer lead times are reported by suppliers, not a clean public series.
You can't build a data center without electricians, welders, and construction crews — hundreds of them, for 18 to 36 months. But once it's built, a massive campus might employ only a few dozen people to run. The construction boom is real but temporary, and it pulls workers away from housing, roads, and other projects that also need them.
Who wins
Electricians & construction trades
Who pays
Sectors that lose those workers; wage pressure
A 100 MW hyperscale campus can employ ~800–1,200 workers at peak over an 18–36 month build — but once open often runs on only ~25–50 permanent staff (Brookings; industry data). [8]
Where the data runs out: Permanent headcount varies by site (≈50–200 for larger facilities); peak construction numbers are self-reported by developers.
Third order — where it lands
When data centers book up the available power supply, the cost of reserving future electricity goes up for everyone. In the eastern US grid (called PJM), the price to reserve capacity 12x'd in three years — and the grid's own market watchdog attributed about 63% of that spike to data center demand. But this is contested: utilities and tech companies disagree on how costs should be split. Either way, the cost eventually shows up in regular people's electricity bills.
Who wins
—
Who pays
Rate-payers; delayed / priced-out projects
PJM's grid capacity price ran from $28.92/MW-day (2024/25) to $269.92 (2025/26) to $333.44 (2027/28) — clearing at the regulatory cap; uncapped, the 2027/28 auction would have hit ~$530. The market monitor tied ~63% of the 2025/26 jump to data centers. [12] [5]
Where the data runs out: Attributing regional price moves to data centres specifically is contested; utilities and data-center firms dispute how costs should be allocated.
Computing power itself is becoming something bought and sold in massive advance contracts, like oil futures. And some of the financing is circular — Nvidia is investing billions in OpenAI, which turns around and buys Nvidia's chips. A supplier funding its own customer's purchases. Critics call it bubble mechanics; supporters call it an ecosystem bet. Either way, the numbers are staggering and the interdependence is real.
Who wins
Compute sellers & chip vendors
Who pays
Buyers — capex, dependency, concentration
Compute is being contracted at unprecedented scale: OpenAI's Stargate targets $500B and 10 GW; Nvidia is investing up to $100B in OpenAI as it deploys ≥10 GW of Nvidia systems — the chipmaker partly financing its own sales ('circular financing'); Anthropic's deal for up to 1 million Google TPUs is worth tens of billions, >1 GW online in 2026. [9] [13] [14] [10]
Where the data runs out: Deal values are announced totals, not audited spend, and Stargate has reportedly faced delays. The Nvidia–OpenAI structure is partly 'circular' — a supplier funding its customer's purchases — which critics flag as a bubble risk. A reported 'SpaceX selling compute to AI labs' deal could not be verified, so it is omitted.
Where the data runs out
Surfaced, not hidden — every place this map is thinner than we'd like.
- Electricity demand: Projections diverge widely — Berkeley Lab's 2028 US range alone is 6.7–12% — and AI's exact share within data-center load is uncertain.
- Generation & the grid: Which specific plants are contracted to which campuses is largely undisclosed; queue volume actually fell ~12% year-over-year as withdrawals rose.
- Materials: Copper figures are analyst forecasts that vary; transformer lead times are reported by suppliers, not a clean public series.
- Labour: Permanent headcount varies by site (≈50–200 for larger facilities); peak construction numbers are self-reported by developers.
- Rate-payers: Attributing regional price moves to data centres specifically is contested; utilities and data-center firms dispute how costs should be allocated.
- Compute deals: Deal values are announced totals, not audited spend, and Stargate has reportedly faced delays. The Nvidia–OpenAI structure is partly 'circular' — a supplier funding its customer's purchases — which critics flag as a bubble risk. A reported 'SpaceX selling compute to AI labs' deal could not be verified, so it is omitted.
Sources
- [1]Energy and AI — Energy demand from AI — IEA
- [2]United States Data Center Energy Usage Report: 2025 Update — Lawrence Berkeley National Laboratory
- [3]Constellation to Launch Crane Clean Energy Center (Three Mile Island Unit 1 restart; 20-year Microsoft PPA) — Constellation Energy
- [4]Queued Up: 2025 Edition — Characteristics of Power Plants Seeking Transmission Interconnection — Lawrence Berkeley National Laboratory
- [5]Data centers 'primary reason' for high PJM capacity prices: market monitor — Utility Dive
- [6]AI data center copper demand & 2030 forecasts — Fastmarkets
- [7]Power transformers and distribution transformers will face supply deficits of 30% and 10% in 2025 — Wood Mackenzie
- [8]New evidence on data center employment effects — Brookings
- [9]OpenAI's first data center in $500 billion Stargate project is open in Texas — CNBC
- [10]Google and Anthropic announce cloud deal worth tens of billions of dollars (up to 1M TPUs) — CNBC
- [11]Data centre electricity use surged in 2025, even with tightening bottlenecks — IEA
- [12]PJM Capacity Auction (2027/28) Clears at Max Price, Falls Short of Reliability Requirement — RTO Insider
- [13]OpenAI and NVIDIA announce strategic partnership to deploy 10 GW of NVIDIA systems — OpenAI
- [14]Nvidia plans to invest up to $100 billion in OpenAI as part of data center buildout — CNBC
- [15]Colossus (data center) — xAI's Memphis supercluster (~555k GPUs, ~2 GW, ~$35B) — Wikipedia
- [16]AI Datacenter Energy Dilemma — Race for AI Datacenter Space — SemiAnalysis
- [17]Anthropic revenue run rate surpasses $65 billion pre-IPO (OpenAI ~$40B) — Axios
- [18]Musk's xAI, SpaceX combo is the biggest merger of all time, valued at $1.25 trillion (xAI later rebranded SpaceXAI) — CNBC
- [19]California tightens oversight of data center energy use, grid costs — Utility Dive
- [20]The AI Build-Out Is Becoming the Biggest Economic Bet in U.S. History (Brookings $10.3T, crowding-out +$9B/-$46B, Mississippi smelter) — Dow Jones Newswires / WSJ (Putzier & Lahart)
- [21]Energy Department will spend $2 billion to squeeze more electricity from the aging power grid — Associated Press (via Renewable Energy World)
- [22]Atlas Energy backs AI data center power expansion (283 MW + 328 MW) — TipRanks
- [23]How Much Are AI Companies Spending on Data Centers? ($59.3B/month construction pace, Census Bureau) — Motley Fool
- [24]Bloom Energy's 800V DC Bet: A $3.6 Billion Answer to AI's Power Bill (Oracle Jupiter $165B campus) — ad-hoc-news.de
- [25]Global Power Demand Capacity for Data Centers Expected to Rise 31% YoY in 2026; Grid Supply Gap to Widen in 2028, Says TrendForce — TrendForce
- [26]U.S. Hyperscaler Capital Spending Seen Reaching $1.2 Trillion in 2027 — TokenPost
- [27]Bitdeer AI Further Expands AI Cloud Services with Off-take Commitments of more than 70% of Its 21.7MW A201 Malaysia Data Center — Bitdeer (via GlobeNewswire)
- [28]FERC revises PJM data-center power plan: large users must bear the costs — Digital Watch Observatory
- [29]US Senate blocks data-center power bill (Ratepayer Protection Act, 57-43) — NewsNation
- [30]DTE Energy's Michigan data-center pipeline tops 8 GW (Oracle 1.4 GW, Google 1 GW) — Data Center Dynamics
- [31]Constellation and Amazon Announce 20-Year Power Purchase Agreement Adding 190 Megawatts of Nuclear Capacity at Calvert Cliffs — Constellation Energy
- [32]FERC chair slams PJM 'mess' as country's largest grid operator delays power auction — Utility Dive
- [33]TVA raises rates for data centers, adding upfront fee for new and existing facilities — WAFF (Gray TV)
- [34]Amazon pledges $1B to data center communities, warns that local opposition threatens U.S. AI lead (Built Together) — GeekWire
- [35]AI faces $6trn test to justify data centres, Bain says ($68B US projects blocked/delayed, June quarter) — Moneyweb (Bloomberg)
- [36]Gallup: 71% of Americans oppose AI data centers in their communities (more unpopular than nuclear plants) — Kiplinger
- [37]Michigan PSC approves DTE-Google 20-year power deal (80% minimum billing demand, ratepayer protections) — BitcoinVersus
- [38]Black Hills plans $1.8 billion investment to power Google's data center (564 MW for Cheyenne, 2027–2029; Google covers full cost) — Reuters
- [39]Google enters massive 3.6-GW power deal with Constellation Energy (3,590 MW; 20-year 890 MW nuclear PPA + 2,700 MW 15-year supply; >$4.3B investment) — Reuters
- [40]Alphabet/Google nearing second Constellation nuclear deal ($1B+, ~890 MW of US plant uprates; expected as soon as this week) — Bloomberg (via CNBC TV18)
- [41]Why big tech is throwing cash into India in quest for AI supremacy (20-year tax break for AI data services) — The Wall Street Journal
- [42]CoreWeave locks in 240 MW in India with AdaniConneX data center deal (option to 480 MW, mid-2028) — Asianet Newsable (secondary)
- [43]Finland orders halt to work on Google data centre sites over environmental concerns (LVV, Muhos/Kajaani) — Al Jazeera (AFP)
- [44]France’s Sesterce plans $11 billion AI investment in Finland — Reuters
- [45]Black Hills plans $1.8 billion investment to power Google’s data center; Google signs 3.59-GW Constellation contract — Reuters via Economic Times
- [46]Google launches satellite to test feasibility of building data centers in space (Project Suncatcher) — NPR
Keep exploring
Power & grid is one edge of the buildout. Follow connected perspectives into chips, materials, capital, policy, jobs, and more — or see what we're adding next.
Every figure is sourced and confidence-tagged; consequences are read second- and third-order (who wins, who pays). Understanding, not investment advice. Found something wrong or missing? Tell us.
How to cite this page
SundayPyjamas Impact Foundation. "The AI data center buildout: the power edge." Impact Suite. Accessed November 2, 2026. https://sundaypyjamas.org/tracker/data-center-buildout.