Foxconn: Perpetual Motion of the Extraction Engine

The project changed form, but the extraction did not end. It evolved and continued—with ever-increasing imbalance between benefits and costs.

The Foxconn–Wisconsin project began as a presidential spectacle. In 2017, state and local officials joined Foxconn and President Donald Trump to announce a $10 billion Gen 10.5 LCD complex with 13,000 promised jobs, framed as the “Eighth Wonder of the World.” Wisconsin offered a multibillion-dollar subsidy package—including up to roughly $3 billion in state tax credits—along with tailored legal treatment, expedited regulatory processes, and special judicial protections to secure the deal.

Building the Deal and Who Wins, Who Pays traced how that spectacle became structure: a large industrial platform built through aggressive land assembly, local borrowing, state infrastructure spending, and regulatory alignment. They also examined who benefits and who carries the cost. Foxconn’s obligations shrank through renegotiation. Microsoft entered with a large AI data center campus. OpenAI joined as a hardware partner. Public entities and local communities remained tied to long-term debt, altered land use, and displacement.

This final installment examines what happens over time.

The original LCD project did not succeed as promised. Such outcomes are common in public-private partnerships heavily weighted toward corporate benefit. The more relevant question is how the system behaves after failure: Who continues to profit, and who continues to pay?

The answer is clear. The system persists—and in key ways becomes even more valuable to private actors—while public obligations remain fixed.

Contents
The Platform in 2025
Persistence Through Layering
Increasing Value to Private Actors
Fixed Public Obligations
Structural Drivers of Continuity
From Project to System
Final Observation
References

The Platform in 2025

The original promise was straightforward: a large-scale LCD manufacturing facility that would employ up to 13,000 people. That facility was never built at the announced scale. A 2021 renegotiation reduced Foxconn’s commitments to roughly $672 million in investment and about 1,454 jobs, with no requirement to produce LCD panels.

Yet the physical platform created for the original vision remains firmly in place. Approximately 2,500 acres were assembled into a single industrial site through negotiated purchases, “blight” designations, and eminent domain. Local governments issued large bond packages to finance land acquisition, grading, and infrastructure. The state expanded highways and connector roads, including major upgrades to Interstate 94. Utilities were extended to support heavy industrial use.

These interventions permanently transformed the region into a pre-built industrial platform. Land use has been altered for decades. Infrastructure capacity is now installed. Debt obligations stretch far into the future.

The uses of that platform, however, have shifted dramatically. Foxconn maintains a reduced presence, operating under a flexible “advanced manufacturing” label and qualifying for performance-based tax credits at lowered thresholds. In late 2025, the company announced an additional $569 million investment tied to expanded operations.

Microsoft has built and continues to expand a large AI data center campus on land originally assembled and prepared for Foxconn. In September 2025, Microsoft described the Mount Pleasant facility as part of the world’s most powerful AI datacenter network and committed an additional $4 billion, bringing its total investment in the site to more than $7 billion. In January 2026, local officials approved plans for 15 more data centers on the site.

OpenAI has partnered with Foxconn to design and manufacture AI data center hardware in the United States, drawing on Foxconn’s facilities—including those in Wisconsin. The collaboration, announced in November 2025, focuses on co-designing racks, components, and supporting systems for AI infrastructure.

The platform now supports AI infrastructure rather than LCD manufacturing. That shift required no new public buildout—it simply repurposed what Wisconsin had already provided. Public wealth extraction continues through all these corporate activities.

Persistence Through Layering

The system endures because new corporate activity layers onto existing public commitments rather than replacing them.

The original public investments remain unchanged:

Local debt tied to land assembly and infrastructure
State-funded transportation and utility expansions
Permanent land-use changes and community displacement

These obligations do not shrink or reset when corporate plans evolve.

New activity simply attaches to the same base. Microsoft’s data center development sits inside the tax-increment financing (TIF) district originally created for Foxconn. Its investments generate tax increment that flows through the same structure, while local agreements and state-level data center tax exemptions (including breaks on equipment and construction inputs) return value to the companies. Foxconn, meanwhile, continues to collect performance-based credits under the renegotiated, lower-threshold agreement.

Each new layer builds on the last. Underlying public commitments stay intact. New corporate uses gain the benefit of that infrastructure and financing without resetting the terms. The result is cumulative: fresh value streams are added while old obligations persist.

Increasing Value to Private Actors

The system does more than persist—it grows more valuable to the private actors using it.

The original project focused on LCD panel manufacturing, a lower-margin and highly competitive sector. Today’s activities center on AI infrastructure: hyperscale data centers, servers, and supporting hardware. These segments sit at the heart of current corporate strategy and investor expectations, representing some of the most capital-intensive and profitable areas of the modern economy.

The same publicly financed platform now underpins far more lucrative operations. Foxconn gains flexibility to pivot toward higher-margin AI-related production while maintaining a U.S. footprint with reduced commitments. Microsoft benefits from a site that was already cleared, graded, and connected to major power and transportation infrastructure—avoiding the political and logistical costs of starting from scratch. OpenAI gains access to domestic hardware manufacturing capacity without bearing the full capital burden of building it.

As the economic and strategic value of the supported activities rises, so does the platform’s extraction potential. Gains flow to corporate balance sheets and investors. The enabling structure remains publicly financed.

Fixed Public Obligations

On the public side, commitments do not scale down with changing corporate plans. Local governments remain responsible for decades-long debt incurred to assemble and prepare the site—obligations structured around projected tax increments. State infrastructure spending represents sunk costs; highways, interchanges, and utilities require ongoing maintenance regardless of occupancy or activity type.

Land-use changes are irreversible. Homes and farms were cleared to create a contiguous industrial campus. Those communities do not simply re-form when priorities shift.

These costs are tied to the platform’s existence, not to the success of any specific corporate tenant. Public actors therefore face constant pressure to keep the site active under almost any viable use—to service debt and justify past decisions. That pressure reinforces the system, narrowing choices and tilting new agreements toward terms favorable to incoming corporations.

Structural Drivers of Continuity

Several deeper features make the system especially difficult to unwind:

Sunk costs and path dependence
Large, irreversible investments create strong incentives to continue. Abandoning or restructuring the site would force explicit recognition of losses. Maintaining activity allows those costs to remain diffused over time.

Mismatched timelines
Political and symbolic benefits—announcements, media coverage, and campaign optics—were captured quickly in 2017. Fiscal and spatial consequences, however, unfold over decades. Decision-makers who secured early gains are rarely the same ones managing long-term obligations.

Fragmented accounting
The full public cost is scattered across local bonds, TIF reports, state budgets, tax expenditures, and infrastructure programs. No single ledger captures the total investment or ongoing burden. This fragmentation makes holistic evaluation difficult and allows piecemeal adjustments to appear as isolated fixes rather than part of a larger pattern.

These features do not create the extraction engine, but they sustain and protect it.

From Project to System

The Foxconn–Wisconsin case is often described as a failed project. It is more accurately understood as a functioning system. The original narrative—an LCD factory delivering 13,000 jobs—collapsed. The platform built to support it did not. Instead, it was repurposed. New corporate actors entered. New revenue streams emerged. Public commitments, both old and new, remained firmly in place.

Over time, the system has pivoted toward activities of greater value to private actors while the public cost base has stayed largely unchanged. This adaptability, combined with structural resilience, defines the extraction engine. Foxconn in Wisconsin stands as a clear example of perpetual motion in action.

Final Observation

The platform now supports some of the most capital-intensive and strategically significant sectors of the current economy—AI infrastructure and hardware—using land, infrastructure, and financing assembled by one region of Wisconsin under an entirely different set of promises.

The platform continues to generate value. The direction of that value has not changed:

– Public entities remain locked into long-duration costs and constraints.
– Private actors retain flexibility and capture the upside created on that public base.