Hello and welcome to the latest edition of my newsletter, Poiesis. This newsletter is where I share my research and practice relating to society and technology — AI, misinformation, surveillance, ethics, and more. It’s my way to help you understand and change the rapidly changing world of social technology.

In this edition, I want to give updates on the local data center fights I’ve been engaged in, as well as preview some forthcoming work examining the hidden environmental costs of AI-powered mass surveillance technologies.

There have been a lot of developments in the fights against data centers in the San Gabriel Valley outside Los Angeles. Let me give a brief recap of what we’ve been up to.

Monterey Park

This fight has turned into a tremendous success. This coming Monday, the City Council will hold a special meeting to put in place a full ban ordinance on data centers in the City of Monterey Park. This has been the goal of No Data Center Monterey Park since November of last year, and backed by a tremendous amount of organizing, which I’ve discussed in past newsletters.

The data center ban is also on a special election ballot measure for June 2nd, called Measure NDC. In response, No Data Center Monterey Park created a ballot measure committee called Yes on Measure NDC. I’ve been working with the group to support that campaign, and we’ve been extremely busy. The campaign has involved a lot of fundraising, a huge launch party, secured the endorsement of Senator Judy Chu, State Senator Sasha Renée Pérez, and Assembly Member Mike Fong. We are now canvassing and getting the word out to vote YES on June 2nd. The team is in full gear.

We had a great piece of news a few weeks ago, as the developer proposing the data center, HMC Capital, withdrew their application to the City. It was a moment of celebration for our group, as well as for residents across the City. It shows the power of normal people to govern their own lives, regardless of billion-dollar investment firm agendas.

City of Industry

This fight has been multifaceted and widespread. The group has been working in City of Industry as well as in towns, cities, and unincorporated areas surrounding. Because Industry is a very difficult city to fight within, due to its low population and intransigent City Council, the strategy has been to pressure them from the outside. And it’s been pretty successful so far.

Due to public organizing, the cities of Montebello and El Monte have passed data center moratoriums, with El Monte indicating they want to move towards a full ban ordinance. The group has also mobilized to Covina, Hacienda Heights, and LA County (which I’ll describe in more detail below). There has been widespread support against data centers, and the group continues to show up and fight.

Los Angeles County

One particularly important development in the fight around LA is the result from the LA County Board of Supervisors, where one member, Hilda Solis, put forward a resolution to put in place a data center moratorium for unincorporated LA County. Many of us attended that meeting and gave public comment, urging the Board to enact a moratorium and move towards a full ban. Nearly 50 people gave public comment during that meeting.

Unfortunately, the vote ended up moving forward a watered-down version of the plan. The final approved plan is more geared towards studying the effects of data centers in LA County and conducting public education about data centers, not including a moratorium.

Reporting around the decision mentions that several Board members heard from local trade unions that banning data centers will prohibit jobs for their workers. This argument taking hold of people is disappointing, as it’s easy to also understand that any development will allow construction and trade work, so we should push for work that doesn’t harm the community. It’s become an effective wedge that developers are using to drive people to the pro-data center side.

This is a disappointing turn, but not insurmountable. The group will continue to push the County as well as other areas to understand that data center infrastructure is extremely harmful, doesn’t yield strong benefits, and should be protected against.

I’ve also been spending some time thinking about the success of our local data center fights. The blend of environmental justice with anti-tech and anti-billionaire sentiment has created an interested nationwide movement. As I’ve been reflecting on our path thus far, it seems that there are some lessons to learn from what’s been happening. In my own personal journey of activism and community organizing, this fight has provided me with useful insights that make me think critically about work I’ve done in the past.

While data centers may seem an unlikely target for social justice movements, upon examining the features of the fights themselves, they reveal themselves to be a strong target for organized resistance. For one, data centers are an extremely local and tangible piece of infrastructure. Data Center Watch notes in their analysis of fights across the nation that the main concerns of residents are things like utility bills increasing, water usage and pollution, impacts on their property values, noise and air pollution as well as the sicknesses they can cause.

One of the takeaways I’m drawing from this fight is the power of tangible, widespread impacts. Anyone living near a data center will breathe the polluted air, suffer from the increased water usage, have higher energy bills, and get sick from constant noise. It doesn’t matter who you are, these are things that affect everyone.

Contrast that with other fights I’ve been part of across my years. It’s different than the youth climate movement, where we were fighting for less tangible things, against a looming climate apocalypse and for a renewable energy transition. It’s also different than Black Lives Matter and racial justice fights, where impacts of racism are felt by racialized people, so local majority coalition-building can be extremely difficult unless you have a minoritized majority or a lot of allies. It also makes me think of the fights for Palestine, where empathy plays a key role in activation and fighting, as the bombs are not literally dropping on us, but those who are most involved often have connections to the area or are moved by a sense of justice in the face of extreme violence.

Due to these local, tangible impacts, the composition of the anti-data center movement has also been noted as different from typical social justice movements – not falling only within the purview of the left or Liberal center, but also including those who identify as Republicans. Data Center Watch reported that 55% of politicians taking stances against data centers are Republican, and 45% are Democrat. Those who lean left are concerned about environmental impacts. Those who lean right are widely opposed to tax abatements for developers. And issues of power consumption, grid strain, and prices increasing are cross-cutting.

Add to this that the current push for data centers is intrinsically linked, materially and ideologically, to the Trump Administration and Big Tech’s push for AI to pervade every aspect of society. Pew Research reported in September of last year that 50% of Americans are more concerned than excited about AI (50% Republican, 51% Democrat), and only 10% are more excited than concerned. Moreover, 61% of polled respondents wanted “more control… over how AI is used in their lives” – 61% of Republicans polled and 63% of Democrats. Distaste for AI and how strongly it is being forced on society is also bipartisan, as it is becoming a material reality for people regardless of their politics.

These aspects of the fight help explain why the movement is so widespread and able to block tens of billions of dollars of proposed development. But they do not tell the entire story of the success. One other major factor contributing to widespread victories in the anti-data center movement is the fact that most of these proposed data centers are subject to municipal law.

It’s been interesting to fight against extremely local infrastructure. While levers of power at the state and national level can be hard to come by, as I’ve learned in my fights for the Green New Deal or stopping support for the genocide in Palestine, local power can be easier to organize for. It’s fortunate that data centers are, in many cases, subject to municipal zoning laws or ordinances. Because these levers of power exist, it’s been possible to successfully organize residents towards concrete goals, and know exactly who the (manageable number of) targets are.

For me, this has all culminated in new ideas for constructing more effective fights for national-scale issues at our local scale. Though, saying they’re “new” is just for me. It makes me think of my fellow organizers from the Nuclear Freeze movement in the 1980s U.S., who often described to me their process of pushing for local resolutions against nuclear weapons all over the nation, which subsequently led to national treaties and prohibitions on weapons usage. Maybe I’m finally learning the lessons they were trying to impart.

One more related topic I’ve somewhat stumbled into is quantifying the hidden costs of the growing network of surveillance and policing technologies driven by AI. A growing number of people are aware of the connections between Big Tech companies and the deportation machine most notably enacted by ICE. Technology contractors like Palantir are building systems to power ICE’s deportation machine –ELITE, ImmigrationOS, and more.

ICE is now powered by at least 32 Generative AI systems, which rely on an extensive network of data centers to fuel their operations. These systems process massive amounts of data to effectively create surveillance dragnets used to target our friends and family. All of that data and processing, performed with AI-powered software, has an enormous material impact. Generative AI technologies and their data centers use vast amounts of electricity and water, so much so that electric bills are rising and reservoirs are depleting.

I decided that it might be interesting to try to estimate what some of these systems may cost in terms of energy and subsequently water usage. This is not my area of expertise, so I used some rough estimation methods to try to gauge what a surveillance technology used by ICE may require. There is a growing number of studies dedicated to evaluating the energy cost for certain types of AI queries, so I drew on that literature to project what certain systems may cost when they issue many, many queries.

This work is still in progress, but to give you a small example, I took some time analyzing one tool ICE has contracted with a company called Zignal Labs. They claim to be able to analyze 8 billion social media posts per day, and have contracted with ICE to generate risk profiles for social media users. What would the cost of processing 8 billion posts per day look like?

Zignal themselves claim to use NLP, computer vision, and optical character recognition to analyze posts. From the literature I reviewed, each of these machine learning tasks vary in how much energy they use, depending on the model they use. If using an older model, energy cost is less but so is accuracy. If they use an LLM to assist in NLP, or a Generative Adversarial Network to do image recognition, the cost and accuracy are higher.

It is impossible to know which techniques a company is using, so I devised scenarios: using a 75% low-accuracy and 25% high-accuracy, a 50/50 split, or any other combination. I also imagined scenarios of various types of analysis: NLP on all posts, and computer vision on only 10%; or NLP on 100%, computer vision on 50%, and video analysis on 1%. This style of analysis gives ranges of outcomes – best cases and worst cases – that start to give an idea of what the costs of mass surveillance may be.

For one example, let’s imagine Zignal (and ICE) process 8 billion posts in a day, performing NLP on all of them, computer vision on 10%, and video analysis on 1%. Let’s additionally say that for all forms of analysis, they use 80% low-accuracy, low-cost techniques, and 20% high-accuracy, high-cost techniques.

The cost of analyzing 8 billion posts in one day with NLP to those specifications is roughly 4.6 GWh, or the equivalent of 440 average U.S. homes’ energy usage in a year. If powered by a host of small data centers, this could use 200,000 gallons of water, or if powered by AI-specialized data centers, 650,000 gallons. Computer vision on 10% of those posts would use 1.4 GWh of electricity, and anywhere from 61,000 to 196,000 gallons of water. Video analysis of 1% of those posts would use 119 GWh of electricity, and between 5.5 and 9.2 million gallons of water. Combined, this adds to 125 GWh of electricity and between 5.7 to nearly 10 million gallons of water. This is all in one day.

The cost is clearly tremendous, and this is assuming pretty conservative energy usage, with the 80/20 low- to high-accuracy split. It’s easy to look at one AI query, say an LLM query, and say that it looks small because it’s only about 10,000 Joules — less than 1/300th of one kilowatt hour, which something like a microwave may consume in an hour of active use. But the key factor is that these small operations are being done billions of times to achieve a mass surveillance net. The scale needed for surveillance is the killer, pushing otherwise small numbers into massive ones.

I hope to continue this type of analysis and publish this line of thinking somewhere, so stay tuned. If anyone knows of a good outlet who would be interested in this type of analysis, let me know.

“Thinking Critically About How AI Affects Critical Thinking” Talk at Harvard Law School Library Innovation Lab.

I had an opportunity to speak at a talk series, hosted by a good friend’s lab at Harvard Law School, about my work at Cal State LA researching AI and critical thinking from an educational standpoint. In the talk, I uplift several pitfalls for critical thinking that anyone using it to learn should avoid.

LA Doesn’t Want Data Centers, Big Tech is Building Them Anyway. Op-Ed in Colorado Boulevard, April 2026 Edition.

I also recently had an article published in a local newspaper, Colorado Boulevard, discussing the environmental impact of data centers proposed in the LA area. It’s only available in print as of now, so as an exclusive for you newsletter subscribers, I put it in an online document that you can read and access the links.

New Website Design

Maybe it’s not as interesting to uplift, but it always feels good when you check off a task you’ve had on your to-do list for a year or so.

In the world of survival through entrepreneurializing ourselves, I can proudly say I’ve updated the place used to house the work I’ve been doing. Check it out if you’d like.

Talk on AI and critical thinking with CSUSB Critical Perspectives on “AI” in Education Series

I’m excited to also present the critical thinking work for the Critical Perspectives on “AI” in Education group organizing out of CSU San Bernadino. They have hosted fantastic talks thus far from speakers all across the CSU system, lending critical thoughts on the rushed integration of AI into classrooms. Mark your calendars if you want to join the talk, open to anyone!

Public Hearing to Ban Data Centers in Monterey Park (April 20th, at 7pm)

If you’re local, consider coming by this Special Council meeting tomorrow, or write in comment to let the Council know that we support them banning data center construction in the city of Monterey Park. You can read the full call to action on the No Data Center MPK Substack here.

This newsletter provides you with critical information about technology, democracy, militarism, climate and more — vetted by someone who’s been trained both as a scholar and community organizer.

Use this information to contribute to your own building of democracy and fighting against technological domination! And share it with those who would be interested.

Until next time 📣