In just a few years, generative AI has become ubiquitous in the media,
particularly on the internet. This technology raises ecological andsocial issues that have been widely discussed before, including in this
journal[6]. But the question we raise here is: what are the economic
impacts of the emergence of this technology? Who is funding it, for what
purposes and with what results so far, and with what consequences for
the tech industry and the capitalist economy in general?
This question is worth asking, because generative AI has been the
subject of colossal investments in recent years by major tech players,
even relative to the sector's enormous revenues. The return on
investment is therefore closely scrutinized by the financial world. As
we will see, it has been slow to match the risks taken.
Since OpenAI's announcement of ChatGPT in November 2022, the media hype
surrounding these technologies has been considerable, and their supposed
impact on the economy has been widely discussed. We often hear dramatic
predictions about the impact on employment: ChatGPT could put almost
everyone out of work, and no manufacturing sector is expected to be spared.
However, the predicted revolution seems to be taking longer than
expected. A recently released MIT study[1]examined the adoption of this
technology in a wide range of companies. The results highlight two
things. First, the interest among companies is overwhelming: more than
80% of them report having launched at least one internal generative AI
project. This finding is contrasted by the second point: in the
overwhelming majority of cases (95% of companies), these projects have
remained at the pilot stage and have not seen any real adoption in
production. More generally, the study estimates that the arrival of
generative AI has only brought about limited structural changes in most
of the sectors studied.
To explain this phenomenon, the study points to a technical limitation:
the inability of these tools to learn from their mistakes and use user
feedback to improve their relevance and adapt to the context. Unless a
new technological milestone is reached, the impacts of AI on production
methods are therefore likely to be confined to a more limited scope than
previously announced. This is not good news for the tech industry, which
has bet heavily on the economic benefits of generative AI and
desperately needs to see new markets open up to generate revenues
commensurate with its bet.
Systems that are too resource-intensive
Let's take a closer look at the structure of the economic sector that is
striving to sell us this so-called new industrial revolution. Today,
only a handful of companies are truly developing a business centered
around generative AI. They can be roughly classified into three types of
activity.
First, there are developers of prediction models like GPT or Claude.
These models are software programs capable of completing a text (or an
audiovisual object like an image) in a "realistic" way, that is, similar
to the data provided for training. These companies' business initially
involves extracting data (usually from the internet, legally or
otherwise), then using it during a very costly training phase that
allows them to fine-tune their models. This phase requires a huge amount
of numerical computation, necessitating huge server farms equipped with
cutting-edge processors. The company then monetizes the use of its models.
The second type of business involves leveraging the models provided by
the previous companies to offer a service to individuals or businesses.
The best-known product in this category is, of course, ChatGPT, a
chatbot that relies on GPT models to interact with its users. Other
services exist, for example, to complete or generate computer code. This
category of players is therefore downstream in the production chain.
The last category is upstream: these are companies that sell the
computer hardware (mainly processors) needed to train and use generative
AI models. In reality, the plural is superfluous here. One company has
managed to carve out a monopoly: Nvidia, the largest graphics card
designer, supplies almost all of the processors used to train and use
generative AI models.
Often presented as "dematerialized," digital technologies, and AI in
particular, rely on immense data centers consuming immense amounts of
energy and resources.
Florian Hirzinger - www.fh-ap.com
These different players therefore have different business models. The
question is whether these models are viable from a capitalist
perspective. And therein lies the problem: of all these companies, only
Nvidia manages to make a profit from its business. All the others are
swallowing up astronomical amounts of money, without managing to find a
real market for their products.
Let's first return to the emblematic case of OpenAI. The company falls
into the first two categories we mentioned: it produces a model, GPT5,
which can be accessed through various services, which places it in the
first category, as well as a chatbot, the famous ChatGPT, which places
it in the second.
Anatomy of a Bubble
ChatGPT is by far the most popular of all existing AI services, with a
reputation comparable to major social networks like Facebook or
Instagram. It boasts 400 million active users. However, ChatGPT has two
disadvantages compared to social networks: first, the advertising that
generates revenue for these platforms is less well integrated. Second,
usage costs relative to the number of users are massively higher. As a
result, even ChatGPT's paid subscriptions are far from covering usage
costs, to the point that each new user increases OpenAI's deficit.
The company remains vague about its financial results. It can
nevertheless be estimated that it will have earned $4 billion in revenue
by 2024[2]. But the cost of training and using its models alone is
estimated to reach $5 billion. Adding in other costs such as salaries,
this would result in an expenditure of $9 billion, representing a net
loss of $5 billion. To offset these losses, OpenAI is raising funds at a
frenetic pace, arguably unprecedented in capitalist history. It raised
$10 billion in June 2025, before raising $8 billion in August of the
same year.
Despite these lackluster results, OpenAI is arguably the company that is
performing best-excluding Nvidia. Other models are much less used and
generate much less revenue. Startups attempting to build services based
on models are facing increasing difficulties: they struggle to provide
real added value to other sectors. The few services it provides are
ultimately limited in variety and often resemble some form of chatbot.
The exception to this rule is Cursor, an AI-based computer code editor
that is seeing real adoption-without generating any profits yet. Even
then, the productivity gains for the IT industry are far below the
spectacular announcements made by their suppliers[3].
The reliability of models also remains a problem: AI-generated codes
continue to contain inaccuracies and security flaws, and text generation
continues to suffer from "hallucinations"-for example, it is common for
it to generate false scientific references. These problems are amplified
as the task becomes complex.
Another major problem for these startups: they are massively dependent
on access to AI models (GPT, Claude, etc.). However, since model
production is currently a financial drain, the companies that supply
them could be forced to massively increase their prices, which would, in
turn, make the already fragile economic model of the startups that rely
on them even more unsustainable.
Cédric Durand, Techno-feudalism, La Découverte, 2020, 256 pages, EUR18.
In this book, the author develops the idea that the GAFAM monopolies and
the digital economy are producing social regression.
Degeneration
To overcome these contradictions, the industry is counting on a new
technological leap. But this path seems doomed to failure. The quality
of the models depends above all on the quality and quantity of the input
data.
However, the industry is beginning to run out of new data: it has
already used almost everything available on the internet. AI is
beginning to face a paradoxical problem: an increasingly large part of
its training data consists of data itself synthesized by AI, which leads
to a degeneration of the models[4]. It is clear that progress in AI is
reaching a plateau and that improvements are becoming increasingly
marginal. The recent release of GPT5 has only heightened these concerns,
as the new model has failed to live up to its promises[5].
Faced with this impasse, OpenAI and its ilk will ultimately be forced to
restrict free access to models, or even degrade the quality of the
service offered for the same price, or even a higher price - a
phenomenon already underway. But denial is currently leading them to
ramp up a massive investment policy - by purchasing ever more equipment
- without, for the time being, managing to expand their revenues.
Fabien Lebrun, Barbarie numérique, L'Échapée, 2024, 432 pages, 22 euros.
In this book, the author examines the very concrete impacts of the
extractivism that fuels the digital economy.
The entire sector is in a very fragile position. Tech companies have
embarked on a desperate race that resembles a financial bubble. The
investments made don't even constitute capital that can be used in the
long term-the intense use of processors shortens their lifespan, and the
fleet will have to be renewed within a few years at this rate. If the
bubble were to burst, the sector would find itself with an absurd number
of servers it wouldn't know what to do with.
The reasons that led to this downward spiral are the underlying reasons
that make capitalism a system perpetually in crisis. Since at least the
early 2000s, the tech sector has been built on the assumption of
continuous hypergrowth. Recent years have seen the slowdown of this
hypergrowth, and in response, a succession of desperate attempts to
artificially reactivate it; with the "metaverse," blockchains and NFTs,
and then generative AI. It is becoming clear that this model is reaching
the end of its contradictions.
The shockwaves that a collapse of generative AI could produce would have
consequences for the economy as a whole, the first victims of which, as
always, will be the most vulnerable. Time will tell whether capitalism
will be able to bounce back from this crisis as it did from that of
2008, or whether, on the contrary, these contradictions will lead to
more profound upheavals - for better or for worse.
Nicolas (UCL Caen)
Confirm
[1]"The GenAI Divide: State of AI in Business 2025," MIT, July 2025.
[2]See Edward Zitron, "There is no AI revolution," Wheresyoured.at,
February 24, 2025.
[3]Mike Judge, "Where's the Shovelware? Why AI coding claims don't add
up," Mikelovesrobots.substack.com, September 3, 2025.
[4]"Can artificial intelligence collapse on itself?", Le Monde,
September 10, 2023.
[5]Christophe @Politicoboytx, "ChatGPT-5 threatens to burst the
generative AI bubble," faketech.fr, August 21, 2025.
[6]"Artificial intelligence: AI at the service of the bourgeoisie,"
Alternative libertaire No. 358, March 2025.
https://www.unioncommunistelibertaire.org/?Economie-L-IA-generative-fera-t-elle-tomber-la-Big-Tech
_________________________________________
A - I N F O S N E W S S E R V I C E
By, For, and About Anarchists
Send news reports to A-infos-en mailing list
A-infos-en@ainfos.ca
Geen opmerkingen:
Een reactie posten