This article is part of a series, "Organizing Against the Machines."
In December 2024, the University of Michigan announced plans to construct a data center in partnership with Los Alamos National Laboratory (LANL). The data center would be located in Ypsilanti, a town neighboring UM’s Ann Arbor campus. The project would cost $1.2 billion, with UM directly contributing $200 million and financing $630 million. UM claimed that the data center, billed as an “AI research facility,” would “support Los Alamos scientists and engineers in conducting research focused on critical national security AI challenges.” The project’s website touts numerous benefits: It will “advance science and education” while “creating good jobs and opportunities for Michigan residents.”
Soon after this announcement, a community movement based in Ypsilanti began assembling to stop the data center. Over the next few months, activists canvassed their neighbors, voiced their objections in packed city council meetings, and showed up to university information sessions, apparently taking UM spokespeople by surprise with their strident opposition. The movement drew connections across several troubling aspects of the project, including UM’s extractive relationship with surrounding communities; the environmental harms to water, air, and land; and the disruption to Ypsilanti's already-fragile electrical grid.
And then there is the other partner in the arrangement. LANL, of course, is renowned for developing the atomic bomb, and the maintenance and modernization of the US nuclear stockpile continues to be one of its top priorities. For the community movement, this is a red line: The intensified investment in nuclear war and militarism could not be swept under the rug, greenwashed, or legitimated through any amount of community engagement. UM officials, seeking to sell the project as an unmitigated good, have tried very hard to talk around this particular aspect of the project, repeatedly asserting that the facility will not manufacture nuclear weapons. These assertions clumsily elide the fact that there are lots of other ways for a data center to support the development and production of nuclear weapons, like weapons simulations; indeed, after months of denial from UM, a LANL representative confirmed that nuclear weapons research would be a priority for the project.
The UM-LANL partnership arose amid an intense AI push by the university. In August 2023, UM released UM-GPT, a UM-specific version of OpenAI’s ChatGPT. Since then, the university has released a flurry of other UM-branded generative AI models; for instance, UM Maizey, a platform that augments generative AI functionality with bespoke university datasets, has been integrated into the Canvas learning management system, so that instructors can create their own “classroom AI tutors.” Meanwhile, several departments have rolled out new degree programs focused on AI, while lucrative new funding programs solicit proposals for research and teaching projects that use AI. The concurrent financial investments are substantial: In 2024, the university committed nearly $200 million to three venture capital funds personally overseen by OpenAI founder Sam Altman.
Many commentators have noted how these initiatives, which seek to affix the value of intellectual work to an economy governed by financialized speculation, erode the careful and relational work of actually teaching, learning, and researching. Such effects are intensely tangible. During the four-month teaching strike by the Graduate Employees’ Organization in 2023, UM pushed faculty to shoddily fill in for striking workers by providing sham grades; it is not a leap, then, to interpret university efforts to develop a “GraderGPT” as AI-enabled scabbing. Amid a blossoming of AI-related funding opportunities, UM eliminated block funding, an arrangement that provided arts faculty with much-needed money to get creative projects off the ground. Just a few months after the beginning of the current Trump administration, an announcement from then-president Santa Ono on the closure of diversity, equity, and inclusion offices absurdly promised “24/7 AI tutors and a personal AI assistant,” presumably to patch what was torn out in student-facing programs.
In other words, the University of Michigan has cut positions and funding sources while pouring a combined $850 million into the UM-LANL data center, demonstrating that administrative budgetary decisions have had less to do with revenue shortfalls than with a reallocation of priorities toward defense and supposed cutting-edge technology. At the university level and nationally, domestic programs that could address a crisis in cost of living and a decades-long slowdown in productivity are overshadowed by ever-larger promises of innovation and of national dominance that offer grand narratives about a return to some imagined level of prosperity. The congruence of these issues is made obvious not only in the UM-LANL data center but also in the US military’s investment in AI, the promises of Silicon Valley firms to “reindustrialize,” renewed investments to close the border through new surveillance technologies, and cutting-edge forms of repression amid mass resistance to these changes.
The connections between resurgent militarism and deteriorating labor conditions are clear to community and university activists. In a recent Ypsilanti city council meeting, a resident testified, “My grandfather was a rocket scientist who worked on Trinity [the first successfully detonated nuclear bomb]. He died a violent, lonely alcoholic. So, when I think about the jobs the data center will bring to our area, I think about the impact of introducing nuclear technology to the world and deploying it on civilians. And the impact that that had on my family, the impact on the health and well-being of my family from living next to a nuclear test site, and the spiritual impact that it had on my family for generations.” This testimony poignantly encapsulates our present orientation as labor organizers at UM. As UM student Dylan Cohen astutely notes, the university is deepening its “entrenchment into finance capital and the business of war,” laundering this process as investments in research and “innovation.” This entrenchment doesn’t just erode our working conditions; it also marks us workers as complicit in the violence that will be enacted on Ypsilanti—and on other communities, should UM further cement its role as an accelerator for tech-enhanced militarism. Our task, then, is to uproot the many sites where this entrenchment is taking hold, including by stopping the unfettered proliferation of generative AI and by joining the community movement to stop the data center.
Evidently, the task of confronting the capitalist, imperialist university requires a more capacious understanding of campus labor power. A common tack for labor organizers in higher education is to contest detrimental political developments in the realm of “bread-and-butter” academic issues—research funding, classroom conditions, shared governance, academic freedom. Especially in the last year, as political decisions by lawmakers and university administrators alike have had disruptive effects across exceptionally wide swaths of the academic workforce, this strategy has produced high levels of mobilization of academic workers. But such large-scale agitation, often detached from political analysis, does not automatically build organized power.
We faced precisely this challenge in our initial efforts to organize against UM’s AI push. Over several months, we convened meetings about AI that drew students, staff, and faculty from across UM. The meetings became a regular site for people to vent about a pervasive atmosphere of “AI hype” and to share stories about the technology’s detrimental effects on their workplaces. But converting shared frustration to political action was difficult. We struggled to systematically expand our ranks beyond already-concerned activists. And if AI was already everywhere at the university—and was already taken, however reluctantly, by many of our coworkers to be inevitable—then it was hard to identify particular targets around which to build organizing capacity. Meanwhile, the structural conditions undergirding this AI miasma—consolidating university-industry-military relationships—often remained submerged under our more immediate frustrations.
The UM-LANL data center surfaces these political and planetary stakes: It starkly illuminates the present terrain we wish to transform, where fascistic states and tech companies—not workers or community members—shape what the university invests in and how it operates. The growing community movement against the data center has been instrumental in helping us crystallize our analysis and concretize our efforts. Stopping the data center serves as an impetus, and an orienting mechanism, for building worker and community power. In coalescing our shared campus-community project around this focal point, we are collectively articulating an emboldened vision of labor for the common good.
In October 2025, we organized a mass meeting about the data center, where over two hundred UM students and workers discussed the data center, AI in their workplace, and how we could build student and worker power against them. In December 2025, we released a petition demanding the end of the UM-LANL partnership, which garnered over eight hundred signatures. As we wrote in the petition, “We—the workers, students, faculty, and staff—make this university the ‘world-class’ institution it boasts to be.” Accordingly, in our ongoing outreach efforts—flyering around campus, having coffee chats with union organizers, discussing the topic with coworkers, publishing op-eds in the university newspaper, and attending joint meetings with community members—we have sought to build the power necessary to challenge the university.
As a core part of these efforts, we’ve been articulating ways of connecting our more immediate working conditions to the university’s broader investments. Echoing the Ypsilanti resident’s testimony about their grandfather’s involvement in the Trinity project, we’ve asked our colleagues: Are you able to tell your adviser “no”? Do you have the power to say no to overtime and working overnight, no to AI, no to working on the project with Los Alamos? These questions embed an invitation for people to imagine—and collectively work toward—a university where resources are committed to our working, learning, and living conditions rather than militarism and finance capital. Foregrounding these political possibilities has animated our labor organizing: For instance, in some engineering departments, which have had a scant union presence historically, the data center has proved to be a potent issue for moving graduate student workers to sign union authorization cards and begin talking to their coworkers.
Following months of sustained campus-community activism, the data center project has been delayed by two years. Local government officials in Ypsilanti have passed resolutions opposing the data center, while state legislation has been introduced to rescind a $100 million grant awarded to UM for the project. As an implicit acknowledgment of the growing opposition, UM has ramped up its public relations efforts. Conspicuously, the university’s marketing materials and op-eds seem to gingerly sidestep the project’s overtly militaristic aims, underlining that UM cannot justify the violent horizon of this supposed research future.
Flimsy messaging aside, the fact that UM continues to press ahead with the project despite these setbacks serves as yet another reminder that its entrenched investments in capitalism and imperialism will be hard to dislodge. In the last two years, student-led efforts to sever the university’s investments in Israel’s genocidal occupation of Palestine have been met with brutal repression, including police violence, retaliatory firings, and numerous disciplinary charges. This repressive securitization will continue as UM pursues the LANL partnership. Protesters at an event about the data center received campus bans from university police—reprisals highly reminiscent of those levied at pro-Palestine activists.
Beyond repression from above, our movement must also contend with existing divisions that fracture our organizing. Campus activists can often be paternalistic toward community organizers, overlooking the community’s incisive analysis of the university and crucial leadership in the fight. Campus labor unions do not share an antimilitaristic orientation, making it difficult to coordinate efforts. And faculty at the University of Michigan continue to believe that there is no choice other than to pursue apparent opportunities related to AI and national security in a time marked by cuts to research funding.
For all these challenges, the data center must be stopped, new solidarities must be formed, and a truly public university must be wrested from UM’s current entanglements with the market and the state. We must, then, take seriously a vision of campus labor power that is capaciously collective.
Justine Zhang is an assistant professor in the School of Information at the University of Michigan, Ann Arbor. Her email address is [email protected]. Shreya Chowdhary is a PhD student in UM’s School of Information and a department organizer in the Graduate Employees’ Organization (GEO), AFT-MI Local 3550. Her email address is [email protected]. Nathan Kim is also a PhD student in the School of Information and a department organizer in the GEO. His email address is [email protected].