TL;DR
An announcement is an option, not an outcome. It is cheap to issue, it confers a right rather than an obligation, and it can expire worthless. No published national system measures how often those options are exercised. One manufacturing sector has a facility-level ledger, and it exists because private foundations paid for it. What that sector shows is not a rate other sectors can borrow. The finding is that the base rate is unmeasured, and here is the funnel it would have to survive.
Ask almost any community what its economic-development organization accomplished last year and the answer will include a dollar figure and a job count, both of them announced.
The announcement becomes the unit of account. That is understandable. It is the only number available on the day the community is asked to judge.
Between the press release and the ribbon cutting sit financing, permitting, incentives, site control, engineering, procurement, construction, labor availability, market conditions, corporate capital allocation and, sometimes, the next quarterly earnings call. A project can survive all but one of those and still disappear.
The United States has an extensive system for counting announced manufacturing investment. It has a robust system for counting aggregate manufacturing construction spending. It has no system connecting the two to completed productive capacity.
The numerator gets the headline. The denominator determines the odds.
The bias is not dishonesty. It is availability.
An announcement is verifiable on the day it happens. A realization rate requires following a project for five to ten years, across ownership changes, scope changes, financing changes and, in the semiconductor cohort, completion dates running as late as 2033. [16]
Anyone reporting annually will report the number that exists annually. That is the announcement.
The effect compounds. A metric that is easy to produce becomes the metric that is expected. A metric that is expected becomes the metric compared across communities. A metric compared across communities becomes the basis on which staff are evaluated and budgets are set.
None of that requires anyone to prefer the opacity. It only requires that the alternative measurement never gets built. An economic-development director who wanted to report realization rates instead of announcements would find there is no standard to report against and no peer figure to be compared to. The absence is upstream of the incentive, not downstream of it.
Consider the economic-development equivalent of a baseball lineup. [20]
The single. An existing manufacturer adds a production line. Perhaps five million dollars. Perhaps thirty jobs. It rarely makes the governor's press conference. But it is connected to an existing business, an existing workforce, an existing supply chain and, often, an existing building.
The double. A manufacturer expands into an additional building or a new site. Twenty-five to one hundred million. Fifty to two hundred and fifty jobs. Not transformational. Frequently very valuable.
The triple. A major greenfield investment. Hundreds of millions. A new employer, a new supply chain, significant infrastructure requirements. A genuine win.
The home run. A billion-dollar project. Hundreds or thousands of jobs. Major tax-base implications. National attention.
The grand slam. The two-billion, five-billion, ten-billion-dollar project. The one that gets the rendering. The one that gets the governor. The one that gets a number so large that nobody asks what share of projects of that type reach operation.
The grand slam is real. Communities do land them.
The mistake is making it the business model.
The Clean Investment Monitor, a joint project of Rhodium Group and the Massachusetts Institute of Technology's Center for Energy and Environmental Policy Research (CEEPR), is the closest thing American manufacturing has to a project-level ledger.
Through the third quarter of 2025, roughly six of every ten announced dollars in clean electricity and manufacturing had actually been invested. In industrial decarbonization, meaning hydrogen and carbon management, the figure was roughly one dollar in ten. [8]
Two sectors, one database, one method, six-fold apart.
CIM gives a reason for the gap rather than leaving it as a statistic. Industrial decarbonization technologies are early stage. They depend on a mix of public and private funding, they need policy support to reach commercialization, and they are frequently contingent on securing an offtake agreement for the product before anyone builds anything. [8]
That is a mechanism a practitioner can use, because it is checkable in the room. A project whose financing assumes a grant not yet awarded, or a buyer not yet signed, sits at a different place in the funnel than one whose financing is closed. The two press releases will not tell you which is which.
The consequence in that sector is worth stating plainly. From 2018 through the third quarter of 2025, 17 billion dollars of announced industrial decarbonization projects were canceled. Fifteen billion was invested in the facilities that got built. More was canceled than was constructed. [8]
The facility count tells a related and separate story. Of 380 clean-technology manufacturing facilities announced after the Inflation Reduction Act (IRA) became law on August 16, 2022, 161 were operational as of March 31, 2025. CIM's own framing is "nearly half." [1]
Dollars and facilities are not the same denominator and cannot be substituted. A facility can absorb enormous capital and remain a construction site. A ten-billion-dollar project can contribute billions in real invested dollars while producing nothing.
Three questions live inside every announcement. Did the company announce it? Did the company spend money on it? Did the factory produce something?
Economic-development reporting frequently collapses all three into one. They are not one.
Since the first quarter of 2018, CIM has tracked 279 billion dollars of announced US clean-technology manufacturing investment across 657 distinct projects. 29 billion of that, across 52 projects, has been formally canceled. Widen the definition to include facilities halted before site selection and facilities retired, and the figure rises to 46 billion across 95 projects. [2]
Ten percent by the narrow measure: 29 billion against 279 billion, meaning roughly a tenth of everything announced across those eight years was formally canceled at some point during them. A practitioner reading quickly will file that as reassuring.
Read the distribution instead. The figures that follow ask a different question against the same eight-year base: not how much was canceled across all of it, but how much was canceled in 2025 alone.
23 billion of the cumulative 29 billion, 79 percent of it, occurred during 2025 alone, across 24 projects. Against 6 billion in every prior year combined. On the broader measure, 42 projects tied to 30 billion, 65 percent. [3]
In two quarters of 2025, cancellations exceeded new announcements outright. In the second quarter, 4.5 billion announced against 5.1 billion canceled. In the fourth, three billion against eight billion. [4]
Roughly 27,000 operational jobs have been affected by cancellations since 2018. Roughly 18,000 of them, 68 percent, came from projects canceled in 2025. [5]
Now the part that matters most for a community choosing what to pursue. The figures below measure 2025 cancellations as a share of cumulative announced investment since 2018. That denominator applies to every percentage in the table and to every use of these figures anywhere. [6]
| Category | 2025 cancellations as a share of cumulative announced since 2018 |
|---|---|
| All clean-technology manufacturing | 8 percent |
| Electric vehicle supply chain, all | 9 percent |
| Vehicle assembly | 17 percent |
| Battery manufacturing | 7 percent |
| Critical minerals | 5 percent |
Within a single supply chain, vehicle assembly ran more than double battery manufacturing and more than triple critical minerals.
Outside the EV supply chain, CIM reports the 2025 cancellations in dollars rather than as shares. Solar manufacturing cancellations reached 300 million dollars against three billion announced during the year. Electrolyzer manufacturing cancellations totaled 400 million, with no new announcements at all. [6]
A community that had assumed a single clean-manufacturing realization rate in 2024 would have been wrong in both directions at once, depending entirely on which technology walked through the door.
For context on the aggregate: actual clean-manufacturing investment fell 17 percent in 2025 to 42 billion. New announcements fell 26 percent to 24 billion. The fourth quarter of 2025 produced the lowest quarterly announcement level since the fourth quarter of 2020. [7]
Four limits, stated plainly, because a hostile reader will find them anyway.
A cancellation rate is not a realization rate. Cancellation is one exit from the funnel, and it is the only one published cleanly. Delay, downsizing, rescoping and quiet abandonment are separate exits and none of them appear in any figure above. A project that shrinks from 2,000 jobs to 400 has not canceled. A project still waiting on an offtake agreement five years after its announcement has not canceled. From a community's perspective both are often closer to dead than live, and both are invisible in every dataset discussed here. The measurement gap has a measurement gap inside it.
Clean technology is the least generalizable manufacturing cohort available. The post-2022 surge was driven substantially by federal tax credits, and the sector subsequently met policy uncertainty, tariff changes and trade tension. [19] The policy dependence is precisely what produced the cancellations. That cuts hard against using this cohort as a proxy for ordinary manufacturing.
The 2025 concentration is unresolved. Four-fifths of the cancellation signal comes from a single year. Whether that is a cyclical shock or the year a long tail of options finally expired cannot be determined from this data, and this article does not claim to know. The semiconductor cohort will not finish reporting until 2033. [16] Anyone offering a confident answer now is guessing.
The honest claim is narrow. This is what one well-instrumented sector looks like. It is also the only sector visible at all.
The distinction that matters: the claim "the realization rate is low" can be attacked with data. The claim "the realization rate is unmeasured" cannot, because the absence of the measurement is the thing being asserted. This article makes the second claim only. Nothing above is a rate for food processing, plastics, aerospace or machinery. Nothing above should be carried into a conversation about any of them.
A note on how this article checked itself. During preparation, a figure of 266 billion dollars was carried from one CIM report to another. The number was correct in both. The denominator attached to it was not. In the global CIM report, 266 billion is global clean-technology manufacturing and industry investment in 2023, which fell to 155 billion by 2025. In the US quarterly reporting, 266 billion is total US clean investment across all tracked categories in 2024. Same figure, two universes, one wrong inference available to anyone reading fast. [9]
That is a case study in how a correct number produces a false claim, and it is included because it is the failure mode this article is about.
The question to ask at this point is why clean technology has a facility-level ledger and steel, aerospace, food processing, plastics, machinery and every other manufacturing sector do not.
The answer is not that anyone suppressed it. Measurement gets built where a constituency wants the answer and pays for the counting. It goes unbuilt where no such constituency exists.
CIM states its own origin plainly. It was created because no comprehensive tracking of actual clean-technology investment existed, which made progress impossible to assess. [18]
Its published acknowledgments name its funders. The 2025 quarterly reporting credits Invest in our Future and the William and Flora Hewlett Foundation. The 2024 reporting credited the Breakthrough Energy Foundation and Hewlett. [17]
Private philanthropy funded the only announced-to-operational ledger in American manufacturing.
Note what else that acknowledgment line shows. The funder set changed between editions. One foundation carried through, one did not, and a new one appeared. That is entirely ordinary in philanthropy and no criticism of anyone in it. It is also the reason a community should not plan around the ledger continuing to exist. Statistical programs authorized by statute persist through institutional indifference. Grant-funded measurement persists as long as somebody renews the grant.
What the partial ledgers actually measure. A reader looking for a substitute will find several near-misses. It is worth being precise about what each one does and where it stops.
The Census Bureau operates three separate programs measuring construction. Value of construction put in place collects expenditures by project. The construction sector of the Economic Census collects establishment-level statistics. The Annual Capital Expenditures Survey collects company-level capital expenditure. Census states that the methodologies and universes differ significantly and that the three are not directly comparable as published. [12] Each is well built for its purpose. None was designed to connect a specific announcement to a specific operating plant, and value of construction put in place measures work performed in a period regardless of when the project began, so a year of construction spending cannot be matched to a year of announcements. [13]
Federal project-level data exists only where a federal check was written. The Government Accountability Office (GAO) reports that as of July 2025 the Department of Commerce had awarded 30.9 billion dollars to 19 companies for 40 semiconductor projects, selected from 103 applications. [15] Real ledger, real attrition, subsidized semiconductor projects only.
State incentive-compliance files track clawback triggers, which is a narrower question than realization: a project can miss its announced employment badly and still satisfy the agreement it signed. Industry association surveys are voluntary and self-reported. And aggregate construction series are sometimes pressed into service as a proxy for realization, which they cannot be, for the reason above.
Every one of these measures something real. None of them closes the loop, and no combination of them does either.
The counter-reading, tested. If opacity were being maintained deliberately, CIM would not publish cancellation rates by technology on a public dashboard, and GAO would not publish the application-to-award attrition. Both do. The suppression reading fails on its own evidence.
The structural reading survives, and it is less comfortable. A deliberate choice can be reversed by a decision. An absence can persist indefinitely with nobody responsible for it.
Since no measured funnel exists, here is the structural one. No percentages appear against any stage, because no published national model measures how many projects clear them. [21]
And then the realization tail, which is where a community's actual interest lies: operational, employment realized, fiscal impact realized.
Every stage is a real gate. Every stage has killed real projects. A community told a project is at stage five and treating it as stage seven is not being deceived. It is measuring the only thing it has been given.
A municipality does not have unlimited staff, incentive dollars, industrial land or political capital. Every hour spent pursuing a project carries an opportunity cost, and that cost is paid whether or not the project arrives.
So the relevant question is not whether the community could land a two-billion-dollar project. Of course it could.
The relevant question is what happens to the rest of the pipeline while it tries.
Consider two postures.
The trophy posture. Pursue the transformational project. If it lands, the community's trajectory changes. If it does not, the year produces very little, and the existing manufacturer who wanted to add a line spent that year without a returned call.
The portfolio posture. Pursue the large project and the mid-sized greenfield and the ten expansions and the twenty-five existing-business conversations at once, resourced deliberately rather than as leftovers.
The portfolio posture will not produce the spectacular press release. It produces something a community can compound: repeatability.
Existing-industry retention and expansion is the least glamorous line in this work and among the most valuable. The company already knows the community. The workforce exists. The building may exist. The supply chain is connected. The number of gates in Section VII that such a project has already cleared before anyone announces anything is not small.
None of this argues for abandoning transformational projects. When one appears, drop everything and swing. Build the infrastructure and the shovel-ready assets that make the community competitive when it does.
Just do not build the annual plan assuming one is coming.
The conventional question is how big the project is. The better question is how likely it is to become real. The most useful question is what the portfolio is worth once both are accounted for. [24]
That changes what a community tracks. Instead of announcements alone:
Investment realization. Actual investment against announced investment, by cohort year.
Job realization. Actual employment against announced employment, measured at a stated interval after opening rather than at ribbon cutting.
Project conversion. Operational projects against announced projects.
Pipeline velocity. Median time from announcement to construction start, and from construction start to operation.
Attrition by stage. Where in the funnel the community's own lost projects died.
The last one is the only item on this list entirely within a municipality's power to build, and no consultant will hand it over. A community that has run economic development for twenty years already owns the raw material. The projects announced are on the record.
Determining what actually happened to them takes more than a windshield survey. A constructed building can sit at a fraction of planned capacity, or quietly convert from high-value manufacturing to low-density warehousing, without any change a passing car would register.
Municipalities already hold the instruments that answer this properly and do not typically point them at the question: utility load and interconnection records, personal property and equipment tax filings, business license and occupancy renewals, and local employer counts through state workforce data. Each is routine administrative data the community already collects. Read together against the original announcement, they distinguish an operating plant from a completed shell. [22]
Nobody is going to build the national ledger. A community can build its own.
One caution about the evidence in this article, stated mechanically.
The cohort examined here was announced under a specific set of federal incentives and encountered a changed policy environment before most of it was built. That sequence is what makes it measurable and what makes it unrepresentative at the same time.
This is not a verdict on whether those incentives were good policy. Readers will arrive at that question with settled views and this article has nothing to add to them.
The mechanical point is narrower and applies regardless of where anyone stands. A municipality controls none of the regime risk and all of the opportunity cost. The company can redeploy capital to another jurisdiction or another product line. The community cannot redeploy the year its staff spent, the site it held, the infrastructure it committed or the political capital it burned securing a package. Regime risk is transferred to the party least able to hedge it, and the transfer is invisible on the day of the announcement.
That risk did not begin in 2022 and does not end with clean technology. Tariff schedules move. Procurement rules move. Tax treatment moves. Currency and input costs move. A project announced today is being underwritten against a policy environment that will not be the environment when the concrete is poured.
Which is an argument for portfolios rather than trophies on grounds that have nothing to do with any particular administration. A portfolio holds projects with different exposures. A trophy holds one.
There is nothing wrong with chasing the big project.
There is something wrong with confusing possibility with probability.
Two projects with the same announced value are not the same economic opportunity if one has cleared five gates and the other has cleared one. The press release does not tell a community that. It cannot. It is written on the day the project has cleared the fewest.
Communities should build economic-development strategy around portfolios rather than trophies. The trophy is visible. The portfolio compounds. And the portfolio is what keeps working while the community waits for a grand slam that may not be coming this year, or the next one.
The measurement gap described here will not be closed by a federal statistical program in any timeframe useful to a community making decisions this budget cycle. The one sector that can see itself can see itself because someone bought the instrument, and that someone can stop.
Two things remain within a municipality's control.
The first is internal. Build the cohort record: every project announced in the community over the past ten years, its stated investment and employment, its status today read against administrative data the community already holds, and where in the funnel the lost ones died. This is unglamorous and entirely feasible, and a community that has it negotiates differently than one that does not.
The second is dispositional. Resource the singles and doubles as strategy rather than as what staff do between megaproject pursuits. The existing manufacturer adding a line has already cleared most of the funnel. That is not a consolation prize. It is the highest-probability investment in the pipeline, and it is usually the one nobody staffed.
Do not stop swinging for the fences.
Just do not stop hitting singles.
The funnel in Section VII is the project's path. This is the measurement ladder. They do different jobs and should not be merged.
1. Announced investment. What the company says it intends to invest.
2. Committed investment. Investment supported by financing, incentives, site control or formal corporate commitment.
3. Construction investment. Capital actually deployed into the project.
4. Completed investment. Physical facility substantially complete.
5. Operational investment. Facility producing commercial output.
6. Realized economic impact. Actual employment, payroll, tax base and supply-chain activity.
The analytical mistake is treating level one as though it were level six. The objective is not to stop reporting level one. It is to measure the conversion between them.
Announcements are pipeline.
Construction is evidence.
Operations are outcomes.
Realized economic impact is performance.
Until the denominator is measured, the numerator is marketing.
[1] Clean Investment Monitor, Rhodium Group and MIT Center for Energy and Environmental Policy Research. "The State of US Clean Energy Supply Chains in 2025." April 24, 2025. Source for 380 clean-technology manufacturing facilities announced since enactment of the Inflation Reduction Act on August 16, 2022, of which 161 were operational as of March 31, 2025, which CIM characterizes as nearly half. Reported here as raw counts. This is a facility-count measure and is not a dollar realization rate. https://www.cleaninvestmentmonitor.org/reports/us-clean-energy-supply-chains-2025
[2] Clean Investment Monitor, Rhodium Group and MIT CEEPR. "Clean Investment Monitor: US Q4 2025 Update." February 19, 2026. Source for 279 billion dollars of announced US clean-technology manufacturing investment across 657 distinct projects since the first quarter of 2018; 29 billion in formal cancellations across 52 projects; and 46 billion across 95 projects on the broader measure including facilities halted before site selection and facilities retired. https://rhg.com/research/clean-investment-monitor-us-q4-2025/
[3] Ibid. Source for the 2025 concentration: 23 billion dollars, 79 percent of the cumulative formal cancellation total, across 24 projects during 2025, against six billion in all prior years combined. On the broader measure, 42 projects tied to 30 billion dollars, 65 percent of that cumulative total.
[4] Ibid. Source for the two quarters of 2025 in which cancellations exceeded new announcements: the second quarter, 4.5 billion announced against 5.1 billion canceled; and the fourth quarter, three billion against eight billion.
[5] Ibid. Source for approximately 27,000 operational jobs affected by manufacturing cancellations since 2018, with approximately 18,000, or 68 percent, associated with projects canceled during 2025.
[6] Ibid. Source for the by-category cancellation figures. CIM reports 2025 cancellations as a share of cumulative announced manufacturing investment since the first quarter of 2018 as follows: all clean-technology manufacturing 8 percent; the electric vehicle supply chain 9 percent; vehicle assembly 17 percent; battery manufacturing 7 percent; critical minerals 5 percent. That denominator applies to each figure and must be stated wherever they are used. These figures measure 2025 cancellations only and should not be confused with the cumulative measure in the preceding paragraph, which expresses total cancellations since 2018 against the same eight-year announced base. Outside the EV supply chain, CIM reports cancellations in dollars rather than as shares: solar manufacturing at 300 million dollars in 2025 against three billion announced during the year, and electrolyzer manufacturing at 400 million with no new announcements during the year. No claim is made here about technologies CIM does not enumerate.
[7] Ibid. Source for full-year 2025 aggregates: actual clean-manufacturing investment fell 17 percent to approximately 42 billion dollars while newly announced manufacturing investment fell 26 percent to approximately 24 billion, and the fourth quarter of 2025 recorded the lowest quarterly announcement level since the fourth quarter of 2020. These two series describe different stages of the project pipeline. Actual investment in a given year includes spending on projects announced in prior years. The two figures must not be divided against one another to produce an annual conversion ratio.
[8] Clean Investment Monitor, Rhodium Group and MIT CEEPR. "Clean Investment Monitor: US Q3 2025 Update." November 20, 2025. Source for approximately one of every ten dollars announced in industrial decarbonization since the first quarter of 2018 having been invested as of the third quarter of 2025, against roughly six of every ten actualized in clean electricity and manufacturing. The six-in-ten figure covers clean electricity and manufacturing together and is not a manufacturing-only figure. Also the source for CIM's stated explanation of the gap, namely that industrial decarbonization technologies are early stage, depend on a mix of public and private funding and policy support for commercialization, and are often contingent on securing offtake agreements; and for the finding that from the first quarter of 2018 through the third quarter of 2025 companies canceled 17 billion dollars of industrial decarbonization projects against 15 billion invested in the 57 facilities that began operating. https://rhg.com/research/clean-investment-monitor-q3-2025-update/
[9] Rhodium Group and MIT CEEPR, Clean Investment Monitor. "Tracking Global Clean Technology Investment." March 18, 2026. The inaugural global edition of the Clean Investment Monitor, extending to every country the facility-level tracking previously maintained for the United States alone. Source for global clean-technology manufacturing and industry investment peaking at 266 billion dollars in 2023 and declining to 155 billion in 2025, and for the separate estimate of 1.96 trillion dollars in total global clean investment in 2025. The 266 billion figure appears in both the global and the US reporting and refers to different universes: in the global report it is global clean-technology manufacturing and industry investment in 2023, while in the US Q4 2025 update it is total US clean investment across all tracked categories in 2024. Neither figure should be presented without its geographic and sectoral qualifiers. This report is cited for context only and does not support any claim about a distinct US policy cohort; the global manufacturing decline is substantially driven by Chinese post-subsidy overcapacity, a different mechanism from the US decline. https://rhg.com/research/clean-investment-monitor-global-2025/
[10] U.S. Census Bureau. "Value of Construction Put in Place, Definitions." Construction Spending program. Census defines value of construction put in place as the value of construction installed or erected at a site during a specified period, inclusive of materials, labor, equipment, architectural and engineering work, contractor profit and certain other costs. It is a measure of construction activity, not of announced investment or of completed manufacturing capacity. https://www.census.gov/construction/c30/definitions.html
[11] U.S. Census Bureau. "Value of Construction Put in Place," manufacturing series. Census classifies construction work by type and reports manufacturing separately within private and total construction. The methodology does not identify which economic-development announcement or corporate commitment generated any given dollar of construction spending. https://www.census.gov/construction/c30/pdf/totsa.pdf
[12] U.S. Census Bureau. "A Comparison of the Value of Construction Put in Place Series, the Construction Sector of the Economic Census, and the Annual Capital Expenditures Survey"; and U.S. Census Bureau, Annual Capital Expenditures Survey, "Comparisons With Other Surveys," technical documentation, last revised October 8, 2021. Census states that it operates three distinct surveys measuring the construction sector, that value of construction put in place collects expenditures by project while the Economic Census collects establishment-level statistics and ACES collects company-level capital expenditure, and that the methodologies and universes differ significantly such that the three are not directly comparable as published. ACES separately states that its estimates are not directly comparable with investment data from other sources. https://www.census.gov/construction/c30/vip_csec_9798.html and https://www.census.gov/programs-surveys/aces/technical-documentation/comparisons.html
[13] U.S. Census Bureau. "Construction Spending, About the Survey." Value of construction put in place represents work performed during the reporting period regardless of when a project began or when contractors were paid. Annual construction spending therefore cannot be matched to announcements made in the same calendar year. A project announced in 2022 generates construction spending in 2023, 2024 and 2025. https://www.census.gov/construction/c30/about_the_survey.html
[14] Reserved.
[15] U.S. Government Accountability Office. "Semiconductors: Information on Projects Funded to Strengthen U.S. Supply Chain." GAO-26-107882, December 11, 2025. Source for the finding that as of July 2025 the Department of Commerce had awarded 30.9 billion dollars to 19 companies for 40 semiconductor projects, selected from 103 applications. Cited for the attrition between the proposal universe and the funded universe, and for the fact that federal project-level tracking exists only where federal funds were disbursed. https://www.gao.gov/products/gao-26-107882
[16] Ibid. Source for the finding that Commerce-funded semiconductor projects were expected to reach completion as late as 2033. Any announcement-to-operation analysis conducted in 2025 or 2026 therefore contains a substantial cohort whose outcomes cannot yet be observed. This is an observation-window limit on all work of this kind, including this article.
[17] Clean Investment Monitor, Rhodium Group and MIT CEEPR, published acknowledgments. The US Q3 2025 Update, November 20, 2025, states that the nonpartisan, independent research was conducted with support from Invest in our Future and the William and Flora Hewlett Foundation. The earlier report "Tallying the Two-Year Impact of the Inflation Reduction Act," 2024, states support from the Breakthrough Energy Foundation and the William and Flora Hewlett Foundation. Both state that results reflect the views of the authors and not necessarily those of the supporting organizations. Cited for two claims: that the only facility-level ledger in American manufacturing is philanthropically funded, and that its funder set has changed across editions. No inference about any funder's intent is drawn or intended.
[18] Rhodium Group. "The Clean Investment Monitor: Tracking Decarbonization Technology in the United States." Rhodium states that CIM was created because no comprehensive tracking of actual clean-technology and infrastructure investment existed in the United States, making on-the-ground progress difficult to assess. Cited as the source's own account of the gap it was built to fill. https://rhg.com/energy-climate/data-and-tools/clean-investment-monitor/
[19] Clean Investment Monitor. "The State of US Clean Energy Supply Chains in 2025." CIM notes that the post-IRA manufacturing surge was driven substantially by federal tax incentives and that the sector subsequently faced policy uncertainty, tariff changes and trade tension. Cited in support of the statement that this cohort is the least generalizable manufacturing cohort available, not as a proxy for ordinary manufacturing.
[20] Methodological note on the baseball framework. Singles, doubles, triples, home runs and grand slams are an analytical analogy illustrating differences in project scale and in the number of gates a project has cleared. They are not a statistical classification, and the dollar and employment ranges given are illustrative bands drawn from ordinary practice, not measured category boundaries.
[21] Methodological note on the funnel. The sequence announced, survives diligence, financed, site control, permitted and incentivized, ground broken, completed, followed by operational, employment realized and fiscal impact realized, is a proposed analytical framework developed for this article. It is not an existing federal statistical classification. No conversion percentage is attached to any stage because no published national model measures stage-to-stage conversion for US manufacturing projects.
[22] Methodological note on the internal cohort record. The Section IX recommendation that a community build its own cohort record, and the administrative data sources named for verifying facility status, are practice recommendations rather than descriptions of any existing standard. Availability and access rules for utility, tax, licensing and workforce records vary by jurisdiction. No national methodology governs how such a record should be constructed, which is itself a consequence of the measurement absence this article describes.
[23] Methodological note on realization rates. No single realization rate should be inferred for US manufacturing generally from the CIM clean-technology data. CIM covers a defined clean-technology and industrial universe. Census measures aggregate construction activity. Federal programs such as CHIPS cover selected subsidized projects. These datasets are complementary and cannot be combined mechanically into an all-manufacturing announcement-to-operation conversion rate. No figure in this article should be used as such a rate.
[24] Methodological note on probability weighting. The Section IX framing, that project value should be considered alongside probability of realization, is a decision framework and not a calculation this article performs. No probability figures are offered, because none are known. A complete municipal model would additionally incorporate timing, municipal costs, incentives, infrastructure expenditure, employment quality, fiscal impact, opportunity cost and the distribution of outcomes within a project rather than a binary between delivery and cancellation.
[25] SelectGlobal LLC. "Atlas: The Denominator Wars, Part 1." SelectGlobal Atlas, July 2026. Establishes the analytical framework applied here: distinguishing the headline numerator from the underlying denominator and testing claims against the population, base rate and conversion rate that generated the reported result. https://www.selectglobal.net/select-global-llc-blog/atlas-the-denominator-wars-part-1
SelectGlobal LLC | www.selectglobal.net
Fork Framework(TM) is a trademark of SelectGlobal LLC.
Michael T. Edgar is the Founder and CEO of SelectGlobal LLC. SelectGlobal is a jurisdictional intelligence firm that maps how policy mechanics, procurement authorities, appropriations cycles, and geographic realities converge to create time-bounded windows of validated federal demand, and connects allied-nation manufacturers to those windows before capital is committed. Edgar is a licensed architect (NCARB certified), a former member of the U.S. Investment Advisory Council, and a board director of the International Trade Association of Greater Chicago. His analytical work on institutional transition, reindustrialization geography, and allied-nation market entry draws on 30 years of advisory and project delivery across architecture, real estate development, and international economic development. www.selectglobal.net