Foxconn Pledged $10B and an LCD Plant to Wisconsin. It Spent $1B and Never Built the Factory

On July 26, 2017, in the East Room of the White House, Terry Gou stood beside a president and promised America the largest greenfield foreign investment in its history: a $10 billion factory in Wisconsin, a Gen 10.5 LCD display plant so vast it would need a small workforce of a mid-sized city.¹ The state answered with a subsidy package north of $4 billion — the biggest ever handed to a foreign firm by U.S. state and local governments, and the third-largest in the country's history.³ Bulldozers came. More than a hundred homes were cleared to make room.⁷ Then, quietly, the one thing the whole ceremony was built around — the display factory — was dropped from the plan, well before anyone renegotiated a word of it.⁷

The official story is that Foxconn came to Wisconsin to build LCD panels and ran into reality. That reading treats the deal as a manufacturing decision that failed. It wasn't. The factory was the packaging. What was actually being exchanged in that East Room was political goodwill — at a moment when Trump's trade threats hung directly over the business Foxconn most needed to protect.

The 13,000 jobs were never a promise — they were a ceiling: read the fine print and the famous figure was hedged from the first day

Start with the number everyone remembers as a lie. Foxconn's own release in July 2017 said the plant would create 3,000 jobs 'with the potential to grow to 13,000.' Trump used the word 'potential.' Walker framed 13,000 as what the deal 'could eventually grow to,' alongside 22,000 indirect jobs and 10,000 construction jobs.¹ Every principal hedged in real time. PolitiFact looked at the governor's claim that Foxconn was 'bringing' 13,000 jobs and rated it Half True on the spot — because the figure was conditional on hitting incentive thresholds, and Walker admitted as much.² The 13,000 was the top of a sliding scale, not a floor. It became a broken promise only in memory, sanded smooth into a round number nobody actually guaranteed.

This matters because it tells you what the announcement was for. A firm commitment carries penalties; a 'potential' carries headlines. The design of the pledge — big, round, aspirational, legally soft — is the design of a signal, not a contract. And the incentives were built the same way: performance-based, refundable credits that paid out only against verified capital spend and hiring.³ Foxconn could stand at the podium, collect the political value of the moment, and owe almost nothing if the factory never rose. The state kept the risk that the jobs wouldn't come; Foxconn kept the option to walk. Both sides knew this. The public didn't.

Why a company that assembles iPhones would build a factory it didn't want: the trade war made a symbolic American plant worth more than an actual one

Here is the mechanism the factory story misses. Foxconn's core business is assembling Apple's iPhones in China for the American market. In mid-2017 the single largest threat to that business was the man Terry Gou was standing next to — a president campaigning on tariffs against exactly the China-to-U.S. supply chain Foxconn lived on.⁸ Against that threat, what was a $10 billion pledge worth? Not as a factory — as insurance. A visible, flag-draped commitment to build 'in America' bought goodwill with an administration that could otherwise tax the company's real profit engine. Judged as trade-war insurance, the announcement was cheap and it worked. The factory was the receipt, and the receipt didn't have to clear.

So the LCD plant quietly vanished. By the time the golden shovels came out, the display factory was already being scaled back; it was gone from the plan before the 2021 renegotiation made the retreat official.⁷ That renegotiation is where the gap became visible in ink. On April 20, 2021, Wisconsin and Foxconn tore up the original terms and wrote new ones: planned investment cut from $10 billion to $672 million, promised new jobs from 13,000 to 1,454, and the ceiling on tax credits slashed from $2.85 billion to $80 million.⁴ The state got roughly a fifteenth of the money and a ninth of the jobs it had been sold — and, crucially, it stopped paying for a factory that was never going to exist.

What Wisconsin was left holding: the credits stayed small because the delivery did — but the land and the roads were already spent

Track the jobs year by year and you see a project that never became what it was named for. Under the renegotiated deal, WEDC verified 579 eligible jobs for 2020, 768 for 2021, roughly 1,029 in 2022, and 1,114 in 2023 — with cumulative state tax credits climbing past $52 million.⁵ By the end of 2024, Foxconn had invested nearly $717 million and created 1,242 jobs, earning $62.9 million in credits.⁶ These are real jobs and real dollars — but they are a rounding error against the original headline, and the shape of the incentive contract is why the public purse wasn't drained further: the credits are performance-based, so small delivery meant small payouts. The clever design that let Foxconn walk also protected the taxpayer from paying for a phantom.

The part that didn't self-correct is what the state spent before any verification kicked in. Local and state governments poured hundreds of millions into land acquisition, roads, and infrastructure for a factory footprint that was never filled, clearing more than a hundred homes in Mount Pleasant to do it.⁷ Those costs are sunk regardless of how few LCD panels ever shipped — which was zero. The performance clauses protected the credit line. They did not protect the neighborhood.

Isn't a failed factory still just a failed factory?: the honest counter is that intent is unknowable and the outcome speaks for itself

The fair objection is that this reading is too clever by half. Maybe nobody schemed a signal; maybe Foxconn genuinely meant to build LCD panels, misjudged the display market, and retreated like any company would — and 'it was always a political play' is just tidy hindsight dressed as insight. That's a real challenge, and the honest answer is that intent can't be proven from the outside. But the point doesn't require a conspiracy. It requires only that you notice the deal was built like a signal: a number everyone hedged in public, incentives that paid nothing for non-delivery, and a founder standing beside the one person whose trade policy could gut his real business.²³⁸ Whether or not anyone intended the theater, the structure rewarded the announcement over the factory — and got exactly that. And the coda cuts both ways: in late 2025 Foxconn pledged another $569 million and roughly 1,374 jobs, this time for AI data servers.⁶ Read charitably, the campus finally found a product. Read plainly, it took eight years and a different technology to build a fraction of what was promised for the first one.

Foxconn was never in the LCD business in Wisconsin. It was in the goodwill business, and Wisconsin was the storefront. The $10 billion factory was a promise sized to fill a headline, not a plant — and the incentive contract, read closely, was built so the promise never had to become a building. The state got a cleared field, a hundred fewer homes, and about a tenth of the jobs it was sold; Foxconn got the one thing it actually needed in 2017, delivered the moment the cameras rolled. The eighth wonder of the world was never the factory. It was that everyone applauded a receipt that would never have to clear.