In our previous two posts, we covered why M&A has become a preferred path to market for many developers and how to build a defensible shortlist using market screening and asset-level economics. This post covers the final step: validating that the assets that survive your screen will actually hold up economically when the deepest risk variable gets stress-tested.
Network upgrade costs are the single biggest determinant of project success in renewable development. They are also the variable most likely to surface late, after significant capital has been committed, time has been spent, and a process is already underway.
The reason is structural. Cluster study results arrive well into the development process. By the time costs are assigned, most teams have already formed a view on what a project is worth. If the number comes back materially different from what was assumed, it is rarely a pleasant surprise.
This post covers how to quantify that exposure early, how to understand what kind of project you are dealing with before you price it, and how to close the loop from a confirmed site into a preliminary design that validates your CapEx assumptions.
Why network upgrade costs are so hard to underwrite
The biggest risk is often not the initial upgrade cost. It is how much that cost can change as the queue evolves.
When a project enters a cluster study, its upgrade costs are not determined in isolation. They are allocated based on the composition of every other project in that study group. As projects withdraw, which they do at high rates, the allocation shifts. A project that looks expensive early can become attractive as the queue thins. A project that looks cheap can become a liability as peers drop out and a larger share of remaining upgrades falls to you.
The numbers bear this out. At phase one, withdrawal rates typically run between 40 and 50 percent. At phase two, rates typically run between 25 and 35 percent. Each transition reshapes the cost picture for every project still in the cluster.
The practical consequence is that a single data point from a cluster study is not enough. What you need to know is not just what the network upgrade cost is today. It is which direction it is likely to move, and by how much, as the queue evolves.
Bottom line: A network upgrade cost is not a fixed number. It is a moving target that changes as projects enter and leave the cluster. Understanding how that number is likely to evolve is often more important than the initial study result itself.
The three project archetypes
Not all projects behave the same way as the queue changes. After running sensitivity analysis across clusters and scenarios, three patterns emerge consistently. Understanding which type of project you are looking at changes the diligence conversation entirely.
1. The project that gets better.
This is the most commonly mispriced type. At first glance, it carries a high dollar-per-megawatt cost and a significant number of network upgrades. Red flags on paper. But when you look more closely, many of those constraints are only marginally overloaded, at 101 or 102 percent of capacity. That matters because marginal overloads tend to disappear as competing projects withdraw and system loading eases. In practice, as the queue thins, the constraint count drops from 17 to 7 or 8, and total network upgrade costs decline substantially. A project that looks prohibitive based on one data point may be a legitimate acquisition target once the withdrawal trajectory is modeled.
2. The project that gets worse.
This is the one that matters most for due diligence. Project J3436 is a clean example from a recent cluster analysis: at first glance, five constraints, $96,000 per megawatt for a 200 MW solar facility. For a project of that size, that number looks attractive. But under every other scenario modeled, costs more than doubled. The reason: the upgrades driving J3436’s cost are not marginal. They are significant overloads that will not go away as peers withdraw. What changes is the allocation percentage. In phase one, this project is responsible for a small share of those upgrade costs because it is one of many triggering them. As the queue thins and competing projects drop out, the same upgrades remain but J3436’s share of the bill grows materially. A project priced at $96,000 per megawatt becoming a $200,000-plus per megawatt project as it progresses is not a recoverable situation.
3. The stable project.
A 200 MW battery storage facility: 10 constraints, $17 million in total network upgrade costs, $88,000 per megawatt. Similar headline numbers to J3436, but the behavior under sensitivity analysis is fundamentally different. Costs hold across scenarios. Volatility is low. This project has a genuinely low economic risk profile from a network upgrade perspective, and the sensitivity analysis confirms it rather than undermining it.
Bottom line: The headline upgrade cost is often less important than how that cost behaves under different scenarios. A project that appears expensive may improve materially as the queue evolves, while a project that appears inexpensive can become uneconomic as costs are reallocated.
How Enverus Interconnect® quantifies this in days, not months
The analysis above used to require a consulting engagement, a detailed queue model built from scratch, and several weeks of back-and-forth with the ISO. Interconnect® runs the same analysis in two days.
A single cluster study result tells you what network upgrade costs look like today. It does not tell you how those costs are likely to change as the queue evolves.
That distinction is what makes scenario analysis valuable.
How sensitive is this project to queue withdrawals?
What the platform produces is a full-scale interconnection model for every project in a cluster: constraint identification, cost allocation, and project-level network upgrade costs across standard and custom withdrawal scenarios. You can see where your project ranks within its cluster, by fuel type and by size, and drill into every individual constraint driving the cost.
How likely are those withdrawals to occur?
The sensitivity framework goes beyond standard withdrawal rate scenarios. Interconnect paired with data from Enverus PRISM® (including long-term LMP forecasts, land feasibility scores, and permitting data), provides the ability to build custom sensitivities that reflect the specific characteristics of the projects competing with yours. A project with weak land feasibility signals in PRISM, for example, can be modeled as a likely early withdrawal. The result is a sensitivity analysis calibrated to your cluster, not a generic one built on average assumptions.
How reliable are the results?
Interconnect is powered by SUGAR™, which the ISOs themselves use to run cluster studies. That is not a coincidence. It is what makes the accuracy benchmark meaningful: recent validations against published MISO and SPP cycle results show a 98 to 99 percent match on network upgrade costs between Interconnect outputs and ISO-published outcomes.
Bottom line: The objective is not simply to estimate network upgrade costs. It is to understand how sensitive those costs are to changes in the surrounding queue and how that sensitivity affects project value.
Closing the loop: from confirmed site to preliminary design
Network upgrade cost sensitivity tells you whether a site is viable. It does not tell you whether the design you are assuming captures the value you are underwriting.
Most developers enter M&A with preliminary design assumptions that were built at a different stage of the project, under different site conditions, with different equipment pricing. Those assumptions feed directly into the CapEx estimate in the economic model. If they are off, the model is off, and you find out after you have entered exclusivity.
RatedPower closes that loop. Starting from the PRISM land parcel, with buildable area already defined across 25 configurable constraints, RatedPower produces a preliminary design in minutes: equipment selection, layout configuration, and 0 to 30 percent electrical design documentation, including drawings, bill of quantities, and energy yield results.
The output is not a finished engineering package. It is the right level of detail for the diligence stage: enough to validate your CapEx assumptions, identify site-specific constraints that a desktop screen would miss, and anchor your economic model in a realistic design before you finalize valuation.
Bottom line: A site is not fully underwritten until the design assumptions behind the valuation have been tested. Preliminary design helps confirm whether the economics you are modeling can realistically be built.
The full workflow in sequence
Put together, the three posts in this series describe a single connected workflow:
Start with a market hypothesis, not an asset. Use PRISM to screen the nationwide opportunity set by load growth, forward LMP, generation composition, and substation headroom, and narrow to a shortlist of candidates worth evaluating. Run asset-level economics on those candidates using CEMS data, nodal pricing, and pre-built DCF models, without a manual data build or a banking engagement. Then, before committing to a process, use Interconnect to model network upgrade cost exposure across withdrawal scenarios and understand what kind of project you are looking at. Once the site clears, use RatedPower to produce a preliminary design that validates your CapEx assumptions before you finalize valuation.
What used to require three vendors, multiple consulting engagements, and six or more weeks of elapsed time runs in a single workflow. That is not a minor operational improvement.
The developers who move fastest are not necessarily taking more risk. Increasingly, they are reducing uncertainty earlier in the process.
In a market where timing has become a competitive advantage, the ability to move from hypothesis to conviction in days rather than months changes which opportunities you can actually compete for.
Enverus brings together the data, economics, and design tools renewables developers need to move from market screen to investment conviction, without the manual builds, fragmented vendors, or months of elapsed time. Explore the Enverus platform for renewable developers to see the full workflow.
Want to see how Interconnect models network upgrade cost exposure for a project in your current pipeline?
