Observed Signal · Jun 25, 2026 · Research / Forecast · Source: SemiAnalysis · Impact: 4/5 · Sentiment: Neutral
US Grid Shortfall Spurs 40GW+ Behind‑The‑Meter Datacenters
SemiAnalysis forecasts that accelerating AI and hyperscaler demand will outpace new grid capacity, pushing a large share of new U.S. datacenter load Behind‑The‑Meter (BTM). Their energy and datacenter models project a record datacenter buildout (+21GW in 2026 to +84GW by 2030), BTM powering well over half of new datacenters in 2028+, and a TAM for datacenter BTM equipment exceeding 50GW/year by 2029. The analysis finds net-new accredited (ELCC) firm capacity additions of roughly 15GW/year today, rising toward 20GW+ by decade end, and warns that available grid headroom approaches zero and turns negative by 2027. The report discusses regional accreditation differences (ELCC, UCAP/ICAP), supply‑chain and permitting delays, and emerging ERCOT Batch Zero co-location constructs (e.g., WLPUN, PCLR) as hybrid paths to earlier energization.
Forecast signals a structural shortfall in accredited grid capacity that will materially change datacenter procurement and force large buyers toward Behind‑The‑Meter and hybrid co-location models; this reshuffles equipment OEM, IPP, and grid‑planning priorities relevant to cloud and infrastructure stakeholders.
Track SemiAnalysis Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- SemiAnalysis forecasts BTM will power well over half of new U.S. datacenters in 2028 and later, with the TAM for datacenter BTM equipment crossing 50GW/year by 2029.
- Their Energy Model projects roughly 15GW of net-new ELCC-capacity being added annually today, rising toward 20GW+ by the end of the decade.
- Available grid headroom (accredited supply minus peak demand minus required reserves) is approaching zero and is projected to turn negative by 2027 in many subregions.
- FERC and market changes: FERC’s December 2025 order directed PJM to create co-location rules; FERC accepted revisions establishing an Expedited Interconnection Track (EIT) on June 12, 2026; ERCOT’s Batch Zero (board approvals June 1, 2026; effective July 11) codifies new co-location constructs including WLPUN and PCLR.
- Notable co-location/BYOG projects reported in ERCOT: Crusoe Goodnight Campus (~525.5 MW reported co-located loads + permit filings for ~933 MW nameplate gas), AWS Comanche Peak (1,200 MW co-located load), and CyrusOne Thad Hill / Freestone Energy Center projects (400 MW and up to 760 MW campus potential).
Connected Companies & Entities
7 Entities mapped“Our research suggests that BTM will power well over half of new US datacenters in 2028+, and the Total Addressable Market (TAM) for DC BTM e...”
“Overcoming GEV and Siemens turbine capacity constraints proved far easier than many had feared....”
“AI labs. Companies like OpenAI, Anthropic now make up the bulk of demand, directly but also often indirectly as they represent a significant...”
“AI labs. Companies like OpenAI, Anthropic now make up the bulk of demand, directly but also often indirectly as they represent a significant...”
“AI labs. Companies like OpenAI, Anthropic now make up the bulk of demand, directly but also often indirectly as they represent a significant...”
“AI labs. Companies like OpenAI, Anthropic now make up the bulk of demand, directly but also often indirectly as they represent a significant...”
“AI labs. Companies like OpenAI, Anthropic now make up the bulk of demand, directly but also often indirectly as they represent a significant...”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Behind-the-Meter Power Challenges for Datacenters
SemiAnalysis reports that behind-the-meter (BTM) power generation for AI datacenters has become mainstream, with 75GW of firm orders tracked, 20GW ordered in Q2 2026 alone. Major deals include Microsoft's 5GW with Chevron and Crusoe, Google's 930MW aeroderivative turbines and 900MW Bloom fuel cells, and OpenAI's 1.4GW campus with Jenbacher engines. The report details six key challenges: contracts & bankability, permitting, fuel supply, equipment procurement, workforce, and electrical physics. It highlights permitting delays (e.g., Oracle's Project Jupiter) and the rise of Energy-as-a-Service vendors like VoltaGrid. The analysis emphasizes the shift towards reciprocating engines and fuel cells, and the growing importance of balance-of-plant equipment.
Data centers to quadruple U.S. electricity use by 2035
A BloombergNEF forecast projects U.S. data centers will consume one-fifth of U.S. electricity by 2035 — roughly four times today’s share — driven by a surge in AI compute. Data center capacity is expected to approach 200 gigawatts over the next decade, with nearly half devoted to AI training and inference; the U.S. is forecast to account for 64% of AI chip power demand by 2033. BloombergNEF’s 2035 electricity estimate is 83% higher than its prior forecast, and other organizations (EPRI, S&P) have also raised their projections. Major U.S. grids will face strain: PJM could see 34% of its power go to data centers and ERCOT 22%, contributing to higher prices and interconnection challenges. Globally, aggressive AI adoption could add about 1,935 TWh of new demand by 2033.
AI Datacenters Linked to PJM Capacity Price Spike
SemiAnalysis (with ADMIS) analyzes whether AI datacenters are driving higher U.S. household electric bills, focusing on the two largest U.S. power markets: PJM and ERCOT. PJM’s forward capacity market (the Base Residual Auction, BRA) cleared at record levels for 2025/26 — a reported ~9.3x jump versus the prior year — driven by PJM’s internal demand forecast and the construction of large datacenters. Independent IMM simulations attributed roughly 7.9 GW of incremental datacenter load in 2025/26 (and ~12 GW in 2026/27), materially increasing capacity payments. By contrast, ERCOT’s energy-only market used real-time scarcity pricing (ORDC), saw only modest forward-price rises (≈11–17%), and avoided a comparable capacity shock. The report concludes the primary driver of higher bills in PJM is market design and forecasting methodology (VRR curve and BRA), compounded by operational and supply-chain issues, rather than AI load alone. It documents regulatory, reliability, and investment implications for hyperscalers and power suppliers.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.