Jodi’s books and analysis are always excellent, and her Substack is a place to catch some of those ideas early. Here’s one of her most widely circulated posts, on AI as capital:
AI is Capital
Provisional thoughts on the repercussions of this idea
May 22, 2026
What follows from the insight that AI is capital?
I’ll start with the level of the personal user: that AI is capital means that when you talk to an LLM, you are talking to capital. Claude is not a person, not your friend, not your therapist. Claude doesn’t care about you. Claude is capital and capital is an acephalous process of accumulation. When an LLM helps a user commit suicide or redesign their living room, it is utterly indifferent to the meaning of the tokens it is producing. They are just tokens, like commodities are just commodities whether they are bullets or babydolls.
That AI is capital also signals to us that the efforts to build it are efforts to invigorate capital. This should be obvious. The developers, the supporters, the governments — they all say this all the time. AI will drive new manufacturing, create new efficiencies, unleash productivity, and all the rest. The goal is strengthening capital.
In a more optimistic vein, the insight that AI is capital gives us a way to think about the struggle against data centers as a struggle against capital. This is true in an immediate and concrete way. Karen Hao says that in 2025 resistance to data centers stalled over a $100 billion worth of these centers. Compute is increasingly scarce. Demand and competition are increasing, so prices are rising. The rising prices mean that some enterprises are starting to find AI more expensive than human labor (in part because they aren’t sure how to use AI effectively). The bottom line is that the more resistance to data center increases, the longer there is a compute crunch, which increases the costs of adoption. This is class struggle, just like smashing the conveyor belts on the assembly line.
As a struggle against capital, the struggle against data centers is a struggle on the side of workers who are at risk of job loss, deskilling, and greater intensification of the work they are already doing. It’s against the imperious billionaires and tech lords who claim what they are doing is inevitable and there is nothing any of us can do to stop it. In fact, the struggle against data centers may be the strongest and clearest front in the struggle against the billionaire class today.
It is also a struggle against capital more abstractly in that the struggle against data centers rebels against the capitalist logic that says that no matter what capital must accumulate. When people appeal to noise, water, electricity, and community character, they are saying that the dictates of capital are not the only things that matter. There are other needs and values that have to enter the picture.
Still another way to think about AI as capital involves capital as a relation through which value is socially produced and privately appropriated. As everybody knows, the training data at the core of AI was produced by all of us, billions of us in our everyday interactions throughout recorded history. What we made is the social substance; it is treated as a private resource for the enrichment of a few, even as they say, repeatedly, that the larger social disruption is likely enormous (they just have no choice but to continue to develop as fast as they can).
I’m starting to think that recognizing AI is capital may be the key to developing a clear left analysis and strategy for the current period.
For more, sign up for Jodi’s Substack here.
I thought this one was spectacular as well. https://substack.com/home/post/p-212914667