Data Centers vs. Renewables: How AI Load Growth Is Reshaping the Interconnection Queue

The numbers that should worry every renewable developer

The U.S. power grid is facing a demand shock it hasn't seen in decades. After 20 years of essentially flat load growth, total electricity demand is now surging — and the primary driver is data centers. According to S&P Global, data center grid power demand reached 75.8 GW in 2026 and is projected to nearly triple to 134.4 GW by 2030. To put that in perspective, 134 GW is roughly equivalent to the entire generation fleet of France.

This isn't speculative. AEP, the utility that serves the heart of PJM's data center corridor from Virginia through Ohio, has publicly cited 24 GW of committed data center load in its service territory alone. Virginia's Loudoun County — home to "Data Center Alley" — already consumes approximately 24 TWh of electricity per year, making it the single most electricity-intensive county in the United States. And it's still growing.

The median wait time for a new large power connection in the U.S. is now 5 years. In Virginia, it's 7 years. For renewable developers, that timeline isn't abstract — it's the same timeline you're competing on for interconnection study slots, transmission capacity, and grid operator attention.

The interconnection queue was never designed to process this volume of load interconnection requests alongside generation requests. The result is a collision that is reshaping how every project in the queue gets studied, scheduled, and built.

Direct competition for interconnection capacity

Data centers and renewable generators don't just share the same grid — they compete for the same interconnection capacity, the same substation upgrades, and the same transmission study resources. When a 500 MW data center campus files for interconnection at the same substation where your 300 MW solar project is queued, the transmission provider must study both requests in the same cluster. The upgrades required to serve the data center's load may consume the very headroom your generation project was counting on.

This competition is most acute in transmission-constrained corridors where data center development is concentrated. In PJM's Dominion zone, the overlap is extreme: the same Northern Virginia substations serving the world's densest concentration of data centers are also the interconnection points for hundreds of megawatts of solar and wind capacity queued for study. When transmission capacity is finite and the queue is processing both load and generation requests, someone loses.

The dynamics are different from traditional load growth. A new residential subdivision adds load gradually over years. A hyperscale data center campus can add 200-500 MW of demand in a single interconnection request, with a load profile that runs 24/7 at high capacity factor. That concentrated, baseload demand profile creates transmission constraints that are fundamentally different from the peaky, distributed load growth utilities have historically planned for.

The cluster study bottleneck

Under FERC Order 2023's reformed cluster study process, all interconnection requests received within a defined window are studied together. Data center load interconnection requests are now entering the same cluster windows as generation requests. This creates a compounding problem: more requests per cluster means longer study timelines, more complex upgrade cost allocation, and more opportunities for restudies when individual projects drop out.

Berkeley Lab's Queued Up 2025 report documented approximately 10,300 projects representing roughly 1,400 GW of generation capacity already sitting in interconnection queues nationally. That was before the current wave of data center load requests. Adding hundreds of large load interconnection requests to already-backlogged clusters doesn't just extend timelines — it introduces a new category of interdependency that cluster studies weren't originally designed to model efficiently.

FERC's co-location framework changes the game

In December 2025, FERC issued a landmark order directing PJM to create new rules for co-located load — the practice of siting data centers directly at or adjacent to generation facilities and serving them behind the meter. This ruling responded to a growing trend of hyperscale operators seeking to bypass the interconnection queue entirely by co-locating their load with existing or new generation resources.

The FERC order addressed several critical questions that had been creating regulatory uncertainty:

Capacity obligations: How should co-located load be treated for capacity market purposes? If a 500 MW generator serves 300 MW of co-located data center load, does the remaining 200 MW count toward the generator's capacity commitment?
Transmission service: Does co-located load that is primarily served behind the meter still require network transmission service? The answer affects both the data center's costs and the transmission revenue that funds grid upgrades.
Interconnection rights: When an existing generator adds co-located load, does this constitute a material modification to the generator's interconnection agreement? If so, the project may need to re-enter the study queue.
Reliability impacts: If the co-located load trips offline, the full output of the generator could suddenly flow onto the transmission system. Transmission planners need to study these contingencies.

For renewable developers, the co-location framework creates both risk and opportunity. The risk is that data center operators may seek to co-locate at substations and generation sites where you already hold a queue position, potentially triggering restudies of your project. The opportunity is that co-location PPAs with data centers can provide the long-term revenue certainty that makes projects financeable — if you can structure the interconnection request correctly from the start.

Regional hotspots: where the collision is worst

The data center vs. renewables conflict is not uniform across the country. It is concentrated in specific transmission corridors where data center demand and renewable generation potential overlap. Understanding these hotspots is essential for site selection and queue strategy.

RTO / Region	Data Center Corridor	Interconnection Wait	Key Constraint
PJM — Dominion Zone	Northern Virginia (Loudoun, Prince William)	7+ years	Substation capacity exhausted; transmission upgrades backlogged through 2031
PJM — AEP Zone	Central Ohio, Columbus corridor	5–6 years	Rapid load growth from Intel, Google, and AWS facilities; 765 kV backbone nearing limits
ERCOT	Dallas–Fort Worth, Austin, San Antonio	4–5 years	West Texas wind/solar competing with DFW data center load for CREZ transmission
MISO	Iowa, central Illinois, Indiana	4–5 years	Emerging data center corridors in wind-rich zones; DPP studies increasingly complex
CAISO	Silicon Valley, Santa Clara, Sacramento	5–6 years	Severe transmission constraints; limited new substation capacity in Bay Area
ISO-NE	Northern Virginia overflow into NJ/CT; Boston metro	4–5 years	Limited land availability; aging transmission infrastructure

PJM's Dominion zone: ground zero

PJM's Dominion zone deserves special attention because it represents the most extreme case of data center vs. renewable competition. Northern Virginia hosts more than 300 data centers, and the pipeline shows no sign of slowing. Dominion Energy has publicly stated that data center load applications in its territory exceed the utility's ability to serve them within normal planning timelines.

For a renewable developer trying to interconnect a solar project in the Dominion zone, the practical impact is severe. Network upgrade costs have escalated dramatically as transmission planners account for both generation injection and data center load withdrawal at the same substations. Projects that entered the queue expecting $15-20 million in upgrade costs are seeing revised estimates of $40-60 million after data center load is factored into the cluster study. At those cost levels, many projects become uneconomic — which triggers withdrawals, which triggers restudies, which triggers further cost reallocation to the remaining projects.

This cascade effect is the single biggest queue risk in PJM right now. It is not sufficient to evaluate your own project's economics in isolation. You must model the probability that other projects in your cluster will withdraw due to data center-driven cost escalation — and what that means for your share of the upgrades.

ERCOT: the transmission tug-of-war

ERCOT presents a different dynamic. Texas has seen explosive data center growth in the Dallas-Fort Worth metroplex, Austin, and San Antonio, but the transmission constraints manifest differently because ERCOT is an isolated grid. The CREZ (Competitive Renewable Energy Zone) transmission lines built a decade ago to carry West Texas wind to urban load centers are now approaching capacity — and data center operators in DFW are effectively competing with West Texas wind and solar developers for that same transmission capacity.

Unlike PJM, ERCOT doesn't have a formal cluster study process under FERC Order 2023 (ERCOT is not FERC-jurisdictional for interconnection). But the practical effect is similar: more load requests mean more complex planning studies and longer timelines for generation interconnection. Developers targeting ERCOT should pay close attention to ERCOT's long-term transmission planning reports, which now explicitly model data center load growth scenarios.

Impact on queue processing timelines

The most immediate impact of data center load growth on renewable developers is timeline extension. Every stage of the interconnection process — from application review through cluster study, facilities study, and interconnection agreement execution — takes longer when the queue is more congested.

Consider the math. A typical cluster study under the reformed FERC Order 2023 process takes 12-18 months for a moderately complex cluster. When data center load requests are added to the cluster, the transmission planning models must evaluate not just generation injection but also load withdrawal, behind-the-meter configurations, and contingency scenarios where co-located load trips offline. Each additional modeling scenario adds engineering hours and calendar time.

The practical consequence for developers:

Application to cluster study results: 18-24 months (up from 12-15 months pre-data center wave)
Cluster study to facilities study completion: 12-18 months (up from 9-12 months)
Total application to IA execution: 3-5 years in most RTOs, 5-7 years in PJM's congested zones

These aren't edge cases. They are the new baseline. RMI's analysis of PJM's "speed to power" problem documented that the average time from interconnection request to commercial operation in PJM had already exceeded 5 years before the current data center surge. With data center load adding another layer of complexity, the 7-year timeline in congested zones is rapidly becoming the norm rather than the exception.

What this means for renewable developers

Data center demand is simultaneously the greatest threat and the greatest opportunity facing renewable developers in 2026. The threat is clear: competition for interconnection capacity, longer study timelines, higher upgrade costs, and a greater probability of restudies. But the opportunity is equally real.

The threat: capacity competition

In regions where data centers are consuming transmission capacity, renewable generators face a structural disadvantage. Data center operators are often willing to pay for transmission upgrades that would be uneconomic for a generation project, because their revenue model (cloud computing, AI inference) supports much higher per-MW infrastructure costs than a solar PPA. When a data center operator agrees to fund $100 million in transmission upgrades to serve a 500 MW campus, those upgrades may consume the substation capacity that your solar project was counting on — or they may trigger a restudy that reallocates costs in a way that makes your project uneconomic.

Developers need to factor data center pipeline data into their site selection process. Before committing to a site, research the data center development pipeline in the same transmission zone. If major hyperscale campuses are planned or under construction near your target substation, your upgrade cost estimates should include a risk premium for data center-driven congestion.

The opportunity: PPAs and co-location

On the other side of the ledger, data centers represent the single largest source of new renewable energy PPA demand in the market. Microsoft, Google, Amazon, and Meta have collectively committed to purchasing hundreds of gigawatts of renewable energy to meet their net-zero targets. These buyers are willing to sign long-term PPAs at prices that make projects highly financeable — if the projects can actually get through the interconnection queue and reach commercial operation.

Co-location is an emerging model where a renewable generator and a data center are sited together, with the data center consuming some or all of the generation output behind the meter. FERC's December 2025 order on co-located load in PJM is creating a regulatory framework for this model, but the rules are still evolving. Developers considering co-location should:

Engage transmission planning counsel early to structure the interconnection request correctly
Model both the "co-located load stays online" and "co-located load trips" scenarios for capacity and transmission impact
Ensure site control instruments cover the footprint for both the generation facility and the co-located load facility
Track FERC's ongoing rulemaking on co-located load treatment, as the rules may change before your project reaches IA execution

Site control implications in a data center world

Here's where the data center load story intersects directly with site control compliance. In regions where data centers are consuming transmission capacity, the timeline from application to IA execution is stretching. That extended timeline creates compounding risk for your site control portfolio.

Option expiration risk escalates

Most option-to-lease agreements in the renewable development industry have terms of 2-5 years, with extension provisions that vary by market and landowner sophistication. When interconnection timelines were 2-3 years from application to IA execution, a 3-year option with a 2-year extension was generally sufficient to carry a project through the queue. At a 5-7 year timeline, those same options expire mid-process — often at the worst possible moment, between cluster study completion and facilities study, when your financial exposure to the project is at its highest.

In data center hotspots like PJM's Dominion zone, developers are now negotiating initial option terms of 5-7 years with 3-5 year extensions — and landowners are demanding higher annual option payments to compensate for the longer commitment. The cost of site control in these regions has increased 40-60% over the past three years, driven directly by the longer timelines that data center congestion has created.

Upgrade cost reallocation changes your coverage math

When a data center-driven restudy reallocates upgrade costs across the remaining projects in a cluster, the financial exposure per MW changes. That change can cascade into your site control calculus. If your original project economics assumed $30/MWh upgrade costs and a restudy pushes that to $50/MWh, you may need to reduce your project footprint to maintain viability — which means parcels that were part of your original site control package are no longer needed, while parcels in a different configuration may now be required.

This is why having airtight site control becomes even more critical in data center-congested regions. The cost of sitting in queue for extra years while data centers consume capacity means your options may expire, your financial exposure escalates, and your project configuration may need to change in ways that your original site control portfolio doesn't support.

The land competition angle

In some markets, data centers and solar farms are also competing for the same land. In PJM's Dominion zone, for example, agricultural land in Prince William and Fauquier counties is being targeted by both solar developers and data center developers. A landowner who signed an option-to-lease with a solar developer may receive an offer from a data center developer at 3-5x the annual rent — creating pressure to terminate the solar lease in favor of the data center. While a well-drafted option agreement protects against this, developers with weak option language or verbal extensions face real risk of losing site control to data center competition.

IRA safe harbor pressure adds urgency

The Inflation Reduction Act's "begin construction" safe harbor deadline of July 3, 2026 adds a time dimension that makes the data center queue conflict particularly dangerous for renewable developers. To qualify for ITC or PTC tax credits under the IRA, projects must demonstrate that construction has begun — either through physical work of a significant nature or the 5% safe harbor — by the deadline.

Projects stuck behind data center load in the interconnection queue face a painful choice. If your project hasn't received its interconnection agreement by the safe harbor deadline, you can either:

Begin construction without an IA — committing capital to a project that may never receive grid access, or may receive it with upgrade costs that make the project uneconomic
Miss the safe harbor window — potentially losing access to ITC/PTC credits that represent 30-50% of the project's economic value
Withdraw from the queue — forfeiting deposits and upgrade cost allocations, and reapplying in a future cycle where data center competition may be even more intense

None of these options are good. The developers who are best positioned are those who entered the queue early, have their site control locked down with long-term options, and are in regions where data center load growth is manageable. For everyone else, the safe harbor deadline is a forcing function that may drive a wave of queue withdrawals in Q2 and Q3 2026 — which will trigger restudies, which will extend timelines further.

Strategic implications for queue positioning

Given the structural shift that data center load represents, renewable developers should reassess their queue strategy with the following considerations:

Prioritize regions with lower data center density. MISO's western footprint (Iowa, Minnesota, the Dakotas), SPP, and parts of CAISO outside the Bay Area offer strong renewable resource potential with significantly less data center competition for interconnection capacity. The upgrade costs and timelines in these regions reflect a more traditional queue dynamic.
Model data center pipeline risk in your interconnection cost estimates. Before filing an interconnection request, research the data center development pipeline in your target transmission zone. If 500+ MW of data center load is planned or under construction near your point of interconnection, add a 30-50% risk premium to your base upgrade cost estimate.
Extend option terms and secure robust extension provisions. In data center hotspots, negotiate initial option terms of at least 5 years with 3-5 year extensions. Include provisions that automatically extend the option term if interconnection studies are delayed beyond projected completion dates. The incremental cost of longer option terms is small relative to the risk of option expiration mid-queue.
Evaluate co-location economics. If your project is in a data center corridor, consider whether a co-located data center PPA could improve the economics sufficiently to absorb higher upgrade costs. The co-location model is still evolving, but early movers who structure their interconnection requests correctly will have a significant advantage.
Monitor cluster study composition. Track not just your own project's study progress, but the composition of your cluster. If large data center load requests are entering the same cluster window, the probability of restudies and cost reallocation increases significantly.
Automate site control monitoring. In a 5-7 year queue timeline, manual tracking of option expirations, SNDA deadlines, and coverage ratios is not sustainable. A single missed expiration in year 4 of a 6-year queue process can trigger a cascade failure that kills the project.

The bottom line

Data center load growth is not a temporary phenomenon. AI training and inference workloads are on an exponential growth curve, and every major hyperscale operator is building out capacity as fast as the grid will allow — and in many cases, faster than the grid can accommodate. The 75.8 GW of data center demand in 2026 is a waypoint, not a peak.

For renewable developers, this means the interconnection queue is permanently more competitive, more complex, and more time-consuming than it was even two years ago. The developers who will thrive in this environment are those who treat site control as a strategic asset rather than an administrative checkbox — because in a queue where timelines stretch to 5-7 years and data center load can reshape your cluster study results at any point, the strength of your site control portfolio is the foundation everything else is built on.

The margin for error has collapsed. The cost of delay has escalated. And the competition for grid access is only intensifying.

Sources

FERC Order 2023 Compliance in 2026: Why This Is the Year — Full market analysis of the 2026 compliance landscape across all RTOs.
PJM Site Control Requirements in 2026: The Complete Developer Guide — Cycle 1 thresholds, Manual 14H attestation, and multi-state permitting.
Interconnection Queue Withdrawal Penalties by RTO — What it costs to exit the queue at each stage, across all seven markets.

Is data center load growth affecting your queue position?

Zonevex tracks your site control coverage against RTO milestones, flags expiring instruments, and helps you stay ahead of an increasingly competitive queue — so you don't lose your position while waiting.

Book a Demo →