A promised job lives in the future. A qualifying job belongs in a particular year's records. They may eventually describe the same work, but a community should not have to treat them as interchangeable while it waits.

Foxconn's project in Wisconsin is unusually useful for thinking about that distance. There was an enormous original ambition. There were revised agreements. There is also verified investment and employment. None of these facts obliges us to call the entire undertaking either a triumph or an empty lot. The harder question is what remains open to examination as one version of the project becomes another.

The story below follows public documents through a November 2025 announcement. It is not a calculation of every public dollar associated with the site, and it does not pretend that an announcement can establish what happened after its reporting date.

What the headline becomes in a contract

WEDC, Wisconsin's economic development agency, signed a fifteen-year agreement with Foxconn in November 2017. Its January 2019 board materials described investment of up to $10 billion, up to 13,000 jobs and as much as $2.85 billion in performance-based tax credits. That last figure was a conditional ceiling. It was not a cheque handed over at the signing ceremony.^[1]

There is a serious argument for helping a large manufacturing investment get started. Equipment, training and a network of suppliers can require commitments that are difficult to coordinate one by one. Public support may make that coordination possible. Nor is a business unreasonable for wanting to know the conditions before committing its own resources. These possibilities deserve attention before we declare every incentive either necessary or corrupt.

But a headline does not tell us what the agreement actually purchases. Wisconsin's Legislative Audit Bureau examined the arrangement in report 18-18. On printed page 8, it explained that 13,000 jobs were not necessary to obtain all the available job-creation credits in the contract's final year. Its examples included 10,400 jobs at a higher average wage, alongside 13,000 at a lower one. Payroll and headcount both mattered.^[1]

Those examples do not establish that anyone intended to deceive the public. They establish something simpler: the number people remember from a promise is not necessarily the threshold an administrator must enforce. A responsible assessment has to follow the promise into its operative provisions. What is mandatory, what is an upper ambition, and what can be exchanged for what?

The audit found another consequential difference. WEDC's policies said that certain unawarded job credits could not carry forward when an annual employment minimum was missed. Its contract allowed them to carry forward. The agency said it planned to alter its policies to match the contract. It would be wrong to describe that response as an audit having secured a stricter contract from Foxconn.^[1]

This is not a colourful detail. It is a reminder that oversight must reach the organisation doing the checking. An agency can be sincerely trying to bring investment to a state while still needing another institution to examine how it translates an agreement into eligibility. Trust in the recipient and trust in the reviewer are different questions.

The public should be able to trace the route from a promise to a condition, from a condition to a verified result, and from that result to an award. Otherwise the most important part of an incentive agreement happens in a language accessible only to its negotiators. A larger commitment makes that route more important, not less.

Changing a plan without changing the memory

In April 2021 Wisconsin announced a renegotiated agreement. The performance-credit ceiling became $80 million, the commitment period fell from fifteen years to six, and employment and investment targets were reset. The state's announcement also said incentives would no longer depend on manufacturing a specified product, provided the hiring and investment requirements were met.^[2]

Foxconn's stated explanation was changing economic conditions and the need for flexibility in responding to market demand. Wisconsin stressed a smaller taxpayer exposure and protection of existing public investments. These arguments need not be dismissed as excuses. Orders and technologies do change. Requiring a business to preserve an obsolete production plan for the sake of an old announcement could make useful employment harder to sustain.^[3]

Still, a sensible decision about what to do next cannot retrospectively establish that the original decision was sensible. There are two questions: how to use what now exists, and whether the earlier commitment was justified when it was made. A new agreement can improve the first without settling the second. Keeping both questions alive is not an insistence that the project must never change.

The governor's 2021 release described the difference between the old and new state credit ceilings as $2.77 billion in taxpayer savings. That is a comparison of potential commitments, not evidence that an equivalent sum of already-disbursed money was recovered. The same release distinguished the state credit agreement from local incentives and infrastructure investment. Reducing one ceiling did not erase every other public obligation.^[2]

The opposite mistake would be to count the original ceiling as money already paid, add other costs and announce a definitive cost per job. A serious total would need separate records on land, infrastructure, finance and actual credits. This essay has not assembled that total. Declining to offer a spectacular ratio is not declining to ask what the project cost; it is refusing to disguise an incomplete account as a finished one.

An existing investment does, however, change the next decision. Something worthwhile may still be possible with facilities and skills already in place. The danger arises when that reasonable consideration turns into a permanent answer: because so much has been committed, more must always follow. Then the original decision acquires authority over every later one without being examined again.

A new round of assistance should therefore face a new question. What does this additional commitment make possible now, and what would happen without it? Asking is not wishing for a factory to close or for workers to lose their livelihood. Real livelihoods are precisely why the answer should be more careful than another repetition of the original ambition.

The future needs its own column

WEDC's announcement of 25 November 2025 contains both a fresh expansion and a dated account of earlier activity. Start with the latter. The agency said that, as of 31 December 2024, it had verified nearly $717 million in investment and 1,242 jobs at the Mount Pleasant operation, qualifying Foxconn for $62.9 million in tax credits.^[4]

That is meaningful activity under the agency's stated measure. It should not disappear merely because the original ambition was much larger. Equally, it is not another way of saying that 13,000 jobs arrived. Recognising what exists and examining the distance from an earlier promise are compatible tasks. Neither requires us to erase the other.

Dates matter as much as magnitudes. The same 2025 release described the earlier amended target as 1,454 jobs by 31 December 2025. Its verified count of 1,242 was dated a year earlier. Subtracting the latter from the former does not establish failure at the later deadline. The announcement cannot, by itself, tell us the verified year-end 2025 result.^[4]

Nor does qualifying for credits establish that every corresponding dollar has already been paid. And verified employment at one employer is not proof that the incentive caused every job, or that the statewide net gain equals that count. To answer those questions, we would need evidence about what would otherwise have happened, including elsewhere in the economy. Compliance with a contract is one kind of evidence, not a complete evaluation of a policy.

Now turn to the future. The November announcement described an additional $569 million investment and 1,374 additional jobs over four years. The total performance-credit ceiling would rise to $96 million through the end of 2029, with a stated total job target of 2,616. Those are commitments and proposed activity for the new stage. They should not be folded into the preceding verified figures as completed results.^[4]

An announcement understandably points forward. An accountable public record must also remain readable backwards. If each new target displaces the previous one, the public is repeatedly invited to a departure and never given a clear account of the journey just completed. A useful update keeps both columns: what was expected, and what has been checked, each with its own date and definition.

Headcount is not the whole local story either. How long do the jobs last? What do people earn? Can acquired skills be used elsewhere? Can the infrastructure support a wider range of activity? The documents reviewed here do not answer all of that. Leaving these questions visible prevents a qualifying count from standing in for everything a community might reasonably care about.

Public support should not entitle a government to define every worker's future. Accepting it should not make a firm incapable of changing a product. The point of conditions is to make assistance answerable, not to freeze everyone inside one prediction. A workable arrangement leaves room for adaptation while preserving a record against which adaptation can be judged.

There is no need to choose between acknowledging today's work and retaining yesterday's questions. We can ask what an investment made possible, what it failed to deliver, and what a revised commitment should require. Those questions do not become disloyal when people have a livelihood at stake. They become more important.

When do promised jobs count? When there is work to count, certainly. But also when the people who helped finance the promise can understand the measure, question the assessment, and meet the next offer without being told that the previous commitment has already decided their answer.