The Vectorocene

David M. Berry

"In technology, a physis is being organized through which mankind's contact with the cosmos takes a new and different form" (Benjamin 1996: 487)

"Behind each leap in A.I. are corresponding jumps in computing power... A.I.’s growing abilities are linked to increases in the size of data centers. To create a cutting-edge model, huge amounts of computing power are needed to analyze and find patterns in data... Massive amounts of electricity will also be needed to support new data centers.... By the end of 2030, that is expected to quadruple, eclipsing the power used by all countries in South America and Africa combined." (Satariano et al. 2026).

Screenshot from Nostos, a game vibe-coded as an experimental practice of what I call "continuous partial world-building" set in a world after an AI takeover, see https://nostos-ai.vercel.app/

Data centres drew 64 gigawatts globally last year, comparable to Germany's entire electricity consumption. Roughly 20 million H100-equivalent chips are in use now, up from about 2.4 million in March 2024, with over 200 million projected to be in use by the end of 2028. Is this the start of a new epoch where vector space leaves traces in the environmental record, in cooling water, electricity use, silicon, lithium, the datacentre and the chip modules? 

If so, this new vector computation might signpost the arrival of what I propose calling the Vectorocene, an age defined by data vectorisation, AI embeddings, and vector infrastructure (cf. Parikka 2014; Stiegler 2018). 

This is an epoch that is currently dominated by the US, according to The New York Times, America has over 5,500 datacentres, roughly ten times as many as the next country, and American companies hold about 80% of global compute (Satariano et al. 2026). The compute belongs to OpenAI, Anthropic, Amazon, Google, Microsoft and Meta, amongst other companies. The H100 chips and others to come lay down a silicon and rare-earth stratum on which a new form of computation is being constructed (see Berry 2025, 2026). Built on the continued relevance of the so-called scaling laws, which are prompting a forecast investment of $1 trillion by 2029, vector computation implies a new accumulation regime I call vector capitalism. This generates a new set of externalities which will further pressure the environment, the economy and societies more generally. 

This article was inspired by discussions at the Retro AI conference held at USC organised by Mark C. Marino 31 July-1 August 2026. 

Bibliography

Benjamin, W. (1996) "One-Way Street", in Bullock, M. and Jennings, M. W. (eds.) Selected Writings, Volume 1: 1913-1926. Cambridge, MA: Belknap Press, pp. 444-488.

Berry, D. M. (2025) "Synthetic media and computational capitalism: towards a critical theory of artificial intelligence", AI & Society. Available at: https://doi.org/10.1007/s00146-025-02265-2

Berry, D.M. (2026) ‘Vector Theory: Epistemology, Political Economy, and Probabilistic Computation’, Philosophy & Technology, 39(3), p. 149. Available at: https://doi.org/10.1007/s13347-026-01162-w.

Parikka, J. (2014) The Anthrobscene. University of Minnesota Press.

Satariano, A., Mozur, P., Gu, J. and Metz, C. (2026) "The Impending, Inescapable Deluge of A.I.", The New York Times, 29 July. Available at: https://www.nytimes.com/interactive/2026/07/29/technology/ai-chips-data-center-boom.html

Stiegler, B. (2018) The Neganthropocene, Open Humanities Press.