Prospects for new forms of crisis competition in the looming authoritarian era of widespread late-capitalist scarcity.
Competition is for losers – Peter Thiel[1]
More RAM! There simply cannot be enough memory produced after OpenAI secured a large share of the world’s production of this essential computer component in a veritable coup. Prices for DRAM[2] (especially DDR5, and to a lesser extent DDR4) are literally exploding,[3] hardware manufacturers are panic buying,[4] while memory producers are discontinuing their consumer products in order to supply only the corporate market caught up in the speculative frenzy.[5] It is an insatiable hunger for RAM, fueled by the current AI bubble, which was heightened to hysterical levels by OpenAI in October 2025 – and which has now spread to other components such as graphics cards (video memory) and SSDs.
What happened? In October 2025, OpenAI was able to simultaneously conclude two supply contracts for computer memory with two of the world’s largest manufacturers – Samsung and SK Hynix – securing around 40% of global production of this component in one fell swoop.[6] Altman managed to keep the content of the negotiations secret – neither Samsung nor Hynix were aware that OpenAI was concluding similarly gigantic deals with their competitor, which is likely to have had a positive effect on the AI company’s respective contract terms.[7] DRAM manufacturers could have at least pushed through higher prices if they had been aware that OpenAI was going to buy up almost half of the industry’s memory output. The deals might never have come about.
After this coup – the contracts were signed within a few hours of each other – became known, panic set in, bringing back memories of the shortage economy of Soviet-style state capitalism: All relevant IT market players, competitors from the AI industry, scalpers, and ordinary consumers willing to upgrade rushed to snap up production capacities, wholesale stocks, and memory kits. This panic-driven surge in demand was only indirectly related to actual demand: No one knows what other secret deals are being hatched by AI companies swimming in investor money. As a result, everyone is trying to secure their memory supply by hoarding purchases – which is leading to a general memory shortage. This high-tech hoarding is thus a consequence of the gigantic AI bubble, in which the US in particular finds itself.[8]
OpenAI is the first company that comes to mind here, as the AI corporation does more than just buy up 40% of finished DRAM production in order to use this memory in its data centers for capital valorization in the context of AI services. Altmann not only buys finished memory modules, but also the preliminary products, the wafers, which are now stored in warehouses. Around 900,000 DRAM wafers are purchased by OpenAI every month and simply stored without being “cut” and processed into RAM.[9] It is unclear when, and if at all, this memory will be used in the AI industry, which is struggling with infrastructure bottlenecks and energy and water shortages[10],[11] – not to mention the persistent teething problems and practical application hurdles that arise during the actual implementation of AI techniques in the rationalization of real workflows in many economic sectors.[12]
Monopolistic Crisis Competition?
What Altmann is practicing with his memory deal could be described as monopolistic crisis competition. It is a crisis form of market competition that directly and immediately aims to achieve a monopoly or a dominant market position. OpenAI wants to go from startup to monopolist in one fell swoop. Control over a large part of RAM production is motivated not only by the expansion of its own AI capacities, but also by the sabotage and obstruction of competition. The memory, which is gathering dust as a wafer precursor in OpenAI’s warehouses, cannot be used by competing companies to expand their own AI models. It is not the development of the most efficient and reliable automation systems that is decisive here, but control over the necessary resources, precursors, and/or production capacities.
“Competition is for losers” – Sam Altman seems to have taken to heart the lecture given by right-wing billionaire and Trump supporter Peter Thiel, which he introduced at Stanford a few years ago under this title.[13] In his remarks on successful corporate strategies, Thiel argued in a somewhat involuntary Marxist manner, openly advocating monopoly as the ultimate goal of market competition. According to Thiel, capital-rich corporations/startups must quickly copy the innovations of the technological avant-garde and expand rapidly. Pumped full of investor money, they ensure rapid growth by offering favorable entry conditions for their products and services, so that once they have achieved a dominant market position, they can slowly tighten the screws. This is the blueprint for the often lamented “enshitification” of the internet, for the gradual deterioration of the terms of use of many online services.
Google, Netflix, Microsoft, Amazon – there is no way around these tech giants in their respective market segments. This now leaves these companies, who often achieved their market dominance through temporary periods of loss, with all options open for profit maximization. On the one hand, OpenAI operates according to the same pattern, with the startup expanding its AI services as quickly as possible at great loss, even foregoing advertising revenue in order to push for monetization once it has established a dominant market position.
But the new factor here is the “scarcity strategy” that the AI company is apparently pursuing. It is also a kind of crisis hedging, a safeguard for the coming crisis. Sam Altman seems to be aware of the precariousness of his situation. The players in the current boom are well aware that many startups will not survive the inevitable bursting of this AI bubble. OpenAI is the “early bird,” the startup that—unlike Google, Meta, or Microsoft—has no profitable business lines that could feed the loss-making AI business. When the bubble bursts, when the gap between imminent profit expectations and bleak market reality becomes unbridgeable, OpenAI can at least hope to survive thanks to its control of 40% of memory production.
The End of Plenty
Altman’s attempt to hinder competition by buying up memory and thus dominate the AI market also highlights the state of industrial production conditions in the global high-tech industry. The market economy ideology of rising demand being immediately satisfied by rapidly growing market supply is currently colliding with oligopolistic reality in an industrial sector characterized by gigantic investment hurdles. Three corporations (Samsung, Micron, SK Hynix) are responsible for more than 90% of global DRAM production, whereby the construction of new manufacturing facilities would require billions in investments in highly complex machinery, workflows, and scarce skilled personnel as a presupposition. The aforementioned memory manufacturers are now in capitalist heaven: there appears to be a tacit agreement to fully utilize existing production capacities without pumping billions into new fabs, while DDR5 memory kits, which were available for around €80 six months ago, are now trading for just under €400.
As already mentioned, the IT industry is well aware that it has been caught up in a bubble. And that is precisely why there will be hardly any newcomers entering memory production, as no one can predict when the AI bonanza will come to a miserable end. The risk is simply too great to invest billions in factories during the boom only to find yourself in a market flooded with cheap memory once the boom is over. The tendency of late capitalist commodity production to constantly increase investment expenditure is particularly evident in the “inflexible” supply in the current memory crisis. This is precisely why OpenAI considered the strategy of “artificial scarcity” to be promising.
This artificial shortage created in the storage market is merely a reflection of the increasing actual scarcity of resources, raw materials, intermediate products, and energy sources that the late capitalist world system faces in its boundless drive for valorization. Microsoft, for example, is sitting on a mountain of unused, extremely expensive AI graphics cards that the company bought up during the current boom without ever using them.[14] There is simply not enough electricity or the necessary energy infrastructure to use all this computing power to train new AI models. The AI bubble, which consumes dystopian amounts of energy, could thus run out of hot air not only because of the discrepancy between gigantic investments and meager returns, but also because of bottlenecks in energy sources or resources. This fundamentally distinguishes the current AI boom from the US real estate bubble, which also fueled enormous resources in a speculative construction boom – but which collapsed due to its internal contradictions, namely the accumulation of bad mortgages by the exhausted US middle class.
The external, ecological barrier to capital thus also appears to set certain limits on the speculative bubble formation that has so far prolonged the systemic crisis in the 21st century.[15] However, shortages and undersupply are emerging in many other economic sectors, in raw materials such as rare earths or lithium, or in foods such as cocoa or coffee, which are already suffering from the climate crisis. Persistent inflation,[16] especially in food prices, is fueled not least by this “external barrier” to capital – while the capitalist drive for profit has only one answer to all these problems: more growth. Capitalism is thus degenerating in its old age into an economy of scarcity à la the GDR, minus the social characteristics and egalitarian population structure of state socialism.[17]
Efforts to control scarce or artificially scarce resources and/or intermediate products in order to achieve a monopolistic position are therefore likely to become a common competitive strategy in the future. What OpenAI is doing is only the beginning of a new era of monopolistic competition in late capitalism, whose compulsion to valorize is increasingly coming up against the external, ecological barrier of capital – the finiteness of resources and the full onset of climate catastrophe.[18] The differences between market competition and the usual geopolitical and crisis-imperialist strategies of resource plundering will thus become increasingly blurred.[19]
The Longing of IT Capital for an Active State
Until now, whenever a bubble formed in the 21st century, the state’s big moment came only after it burst, when it was time to cushion the devastating economic consequences with loose monetary policy and trillion-dollar crisis and investment programs. But this time, the AI gurus are calling on the state for help in the middle of the bubble. It is precisely the absurd, dystopian energy hunger of the AI industry, its inability to quickly modernize the infrastructure ruined during the neoliberal decades, that already necessitates an economically “active state.” The high-tech oligarchs are mutating into a real-life satire of Keynesian ideologues, as they wreak havoc in the milieu of old-left parties.
The industry – whose leaders normally have a penchant for right-wing libertarian market ideology – is seeking government subsidies, investments, or guarantees on several levels.[20] OpenAI has been pressuring the Trump administration, which is politically closely intertwined with the IT oligarchy, for months to extend tax breaks to the AI industry, specifically to infrastructure investments in data centers. The Bloomberg news service referred to this as Silicon Valley socialism.[21] Furthermore, taxpayers are to bear the risks of these investments in the form of government guarantees for corresponding loans, in order to reduce borrowing costs and expand investment activity.[22] The political interdependence between Silicon Valley and the Trump administration is to be followed by the economic interdependence of big business and big politics, as is characteristic of fascist forms of crisis in capitalist rule.
The insane expansion of the industry, especially the industry leader OpenAI, also seems to be aimed at simply exceeding a critical mass above which, in the event of a crisis, corporate bankruptcy must be prevented for economic reasons. The dizzying borrowing, the $1.4 trillion investment projects that are having very real economic effects, the pursuit of the closest possible ties to government financial flows—all of this suggests that Altman simply wants to make his AI corporation too big to fail. This is similar to what happens with systemic banks in financial market crises. This strategy of growing beyond the possibility of bankruptcy is so obvious that Altman felt compelled to publicly contradict it.[23]
Another argument used by AI capitalists to legitimize government support consists of the usual geopolitical competition considerations. If the US does not pump billions of taxpayer dollars into the AI industry, China will win the race for new military-grade technology, according to the usual line of argument. The whole thing is garnished with the usual bootlicking and ass-kissing necessary to secure the goodwill of the Mad King in the White House. Nvidia CEO Jensen Huang,[24] is not only keen to maintain the best possible relations with the military-industrial complex and the Pentagon, with whom the company cooperates in the development of AI-supported weapon systems.[25] Huang also had no qualms about paying homage to Trump in bizarre statements, by praising the US president for courageously standing in the way of the “demonization of energy.”[26] The CEO of the world’s largest corporation, who sounded like a troll lurking on Twitter or Reddit, was apparently cheering the US’s departure from any form of climate protection.
This seems to be the political operating cost of the AI bubble. All the major players in the AI bubble are aware that they are in a speculative bubble, the industry knows that a crash is inevitable – and a large part of their activity during this boom consists precisely of preparing for the coming crash, maintaining good contacts with a state apparatus, securing themselves as comprehensively as possible in order to survive the crash and then rise to dominance in the “cleansed” market. The industry is simply dominated by the hope of hitting the big jackpot as a survivor of this cleansing market storm.
Farewell to the Illusion of Consumer Capitalism
Why all this crazy effort, such as burning vast quantities of energy sources in the midst of the rapidly unfolding climate crisis? Critics of the AI industry contrast this massive burning of resources with the digital rubbish spewed out by generative AI to flood the internet. A new word has already been coined for this: AI slop. But this is only a by-product that is only significant for the culture industry.[27]
From an economic perspective, the holy grail of the industry is the production of AI systems that can take over as many fields of work as possible, either completely or at least partially – the AI gurus simply want to sell automation.[28] This is where the dizzying potential for growth and profit lies. This is the real jackpot. Those who survive the coming crash can hope to lead a total transformation of the late capitalist mode of production that promises fantastic growth prospects and profits.[29] But this is also where the insurmountable central contradiction of the capitalist mode of production lies, its internal barrier, which is becoming fully apparent and generally visible in the AI boom.
The crisis process that began with the stagflation period of the 1970s and the IT revolution of the 1980s, which was prolonged in the 21st century by means of the globalized financial bubble economy on credit now finds its crisis-ridden conclusion in the AI revolution.[30] The class struggle fetishized by the old left is only a surface phenomenon; it is an intra-capitalist conflict over the distribution of surplus value, fought between variable capital (“the working class”) and the capitalist functional elites. What is decisive, however, is the internal contradiction of the valorization process itself: the substance of capital is wage labor, but at the same time, due to competition-driven rationalization, capital strives to minimize wage labor in the production process. The decline of the industrial workforce in most industrialized countries, which was the result of the first IT revolution in the 1980s, is now spreading to large areas of the service sector and the IT sector.[31]
Since the implementation of Fordism after World War II, mass demand from a broad middle class was considered a central economic presupposition for the valorization process of capital; mass production had to find mass demand—and this illusion of consumer capitalism was maintained even during the neoliberal era within the framework of the financial bubble economy on credit. Even as the industrial workforce dwindled, the financialization of capitalism continued to generate demand and jobs, albeit at the price of increasing financial instability and periodic crashes. Capital needs solvent mass demand in order to complete the cycle of valorization in commodity production. Otherwise, the valorization process collapses in on itself.
And it is precisely this ideology of consumer capitalism, based on economically necessary mass demand, that is already becoming insubstantial and hollow in the rise of the AI bubble. Speculative fervor is leading to inflation, not to an expansion of consumption, as in previous bubbles. Consumers are already feeling this, especially in the high-tech sector and within the gaming scene. On the one hand, the current artificial shortage is making consumers realize that mass consumption is effectively being capped in a substantial part of the consumer electronics sector in order to fuel the AI boom. The market is simply supplying the most affluent customers – and those are corporate customers. And it is precisely this clientele that the AI industry is primarily targeting with its automation products.
However, this is only the beginning of the coming AI crisis, assuming the teething problems and start-up difficulties in the automation of wage labor actually be overcome, as the industry hopes. But there can hardly be enough consumers if wage workers are replaced by AI systems or paid less. The old Fordist equation, according to which workers constitute the sales market at the same time as their demand, will no longer work—precisely because the globalized deficit economy of the neoliberal era has exhausted itself. Programmers, for example, are already successfully using AI as a tool, resulting in substantial productivity gains that are reducing working days to working hours.[32] Strictly speaking, AI does not replace programmers in individual work processes; it only makes them more productive and reduces the demands of the profession. The labor market then takes care of the rest.
The AI Cult and the Automatic Subject
Capitalist consumers then simply become superfluous human material, while the corporations and companies that massively increase their productivity with the help of the AI industry can no longer find buyers for their goods and services. The subjectless rule of capital inevitably threatens to shatter on this internal contradiction, on its internal barrier, as soon as the AI bubble runs out of hot speculative air. It is clearly evident—even the old left, highly trained in crisis ignorance, can hardly overlook this.[33]
The doubly free wage laborer, as produced by capitalism, is thus acutely threatened with extinction in the current crisis. The system will consequently enter fully into the post-capitalist transformation that is already looming. And it is fascism that seeks to steer this inevitable transformation process in a barbaric direction: On the one hand, through the introduction of forced labor, as indicated in German pre-fascism or in the prison system of the US. On the other hand, through the marginalization, exclusion, deportation or – as a last resort – simply the extermination of the “superfluous humanity” that capital produces in its agony.
The growth mania, the breakneck expansion of the AI industry, also implies all too clearly that the IT princes of Silicon Valley – who effectively want to replace humanity – have already expanded the fascist death cult of the 21st century, as it bubbles up on both sides of the Atlantic, with their own facets. The AI industry’s growth mania, which exceeds anything seen before, is also driven by an ideological factor. The transhumanism rampant in Silicon Valley forms the perfect ideology for the openly misanthropic final phase of the capitalist systemic crisis, in which only blind delusion can obscure the evident destruction of the ecological and social foundations of human civilization under the most absurd ideological contortions.
Transhumanism does not need to do this; it sees humanity as nothing more than a starting aid, a bootloader for artificial intelligence, which is supposed to inherit humanity, so to speak. That is why transhumanists do not care whether the hunger for resources and energy of AI capital further drives the climate crisis, or whether data centers are draining groundwater from entire regions.[34] They see themselves in a race against time—the self-optimizing superintelligence known as singularity, the artificial AI god that transhumanism wants to create, is to become reality before humanity’s capital depletes the foundations of life.
In fact, transhumanism wants to transform the real-abstract automatic subject of capital into reality, to concretize it, to breathe artificial life into the fetishism of capital throughout society. If necessary, the world will be sacrificed to the desired techno god on the altar of the AI industry. And no one knows exactly what the IT titans are checking out in their AI labs, as the Trump administration, which is allied with the industry, gives them a free hand in this regard. As mentioned at the beginning, OpenAI has secured 40 percent of global DRAM production – and we can only hope that this is really just a monopolistic competition strategy.
I finance my journalistic work largely through donations. If you like my writing, you are welcome to contribute – either via Patreon or Substack.
[1] https://www.youtube.com/watch?v=3Fx5Q8xGU8k
[2] https://en.wikipedia.org/wiki/Dynamic_random-access_memory
[3] https://geizhals.de/kingston-fury-beast-schwarz-dimm-kit-32gb-kf560c30bbek2-32-a3164911.html
[4] https://winfuture.de/news,154997.html
[5] https://arstechnica.com/gadgets/2025/12/after-nearly-30-years-crucial-will-stop-selling-ram-to-consumers/
[6] https://www.mooreslawisdead.com/post/sam-altman-s-dirty-dram-deal
[7] https://www.slashcam.com/news/single/OpenAI-s-Secret-DRAM-Deal–Is-Sam-Altman-to-Blame–19700.html
[8] https://www.konicz.info/2025/11/09/die-kuenstliche-intelligenzblase/
[9] https://www.tomshardware.com/pc-components/dram/openais-stargate-project-to-consume-up-to-40-percent-of-global-dram-output-inks-deal-with-samsung-and-sk-hynix-to-the-tune-of-up-to-900-000-wafers-per-month
[10] https://jungle.world/artikel/2024/16/kuenstliche-intelligenz-energieverbrauch-klimawandel-mehr-hunger-mehr-durst
[11] https://www.mooreslawisdead.com/post/sam-altman-s-dirty-dram-deal
[12] https://www.konicz.info/2025/11/09/die-kuenstliche-intelligenzblase/
[13] https://www.youtube.com/watch?v=3Fx5Q8xGU8k
[14] https://redmondmag.com/blogs/generationai/2025/12/microsoft-is-sitting-on-a-pile-of-unused-gpus.aspx
[15] https://www.konicz.info/2022/01/14/die-klimakrise-und-die-aeusseren-grenzen-des-kapitals/
[16] https://www.konicz.info/2021/08/08/dreierlei-inflation/
[17] https://www.konicz.info/2021/10/14/ddr-minus-sozialismus/
[18] https://www.konicz.info/2022/01/14/die-klimakrise-und-die-aeusseren-grenzen-des-kapitals/
[19] https://www.konicz.info/2022/06/23/was-ist-krisenimperialismus/
[20] https://www.banking.senate.gov/newsroom/minority/warren-presses-trump-administration-on-plans-to-prop-up-openai-and-big-tech-with-taxpayer-dollars-at-the-expense-of-working-class-americans
[21] https://news.bloombergtax.com/tax-insights-and-commentary/openais-tax-subsidy-efforts-amount-to-silicon-valley-socialism
[22] https://www.brookings.edu/articles/openai-floats-federal-support-for-ai-infrastructure-what-should-the-public-expect/
[23] https://www.ft.com/content/5835a5a3-36db-41d7-9944-d9823dbdffc5
[24] Nvidia graphics cards form almost the entire technical hardware basis of the AI boom. The graphics card manufacturer has now become the most valuable company in the world.
[25] https://www.youtube.com/watch?v=cUrJVdF2me0
[26] https://gizmodo.com/nvidia-supercomputers-for-trump-2000678264
[27] https://exitinenglish.com/2024/07/07/ai-and-the-culture-industry/
[28] https://exitinenglish.com/2024/08/03/ai-the-final-boost-to-automation/
[29] https://www.konicz.info/2025/11/09/die-kuenstliche-intelligenzblase/
[30] https://www.telepolis.de/article/Die-Krise-kurz-erklaert-3392493.html
[31] https://exitinenglish.com/2024/08/03/ai-the-final-boost-to-automation/
[32] https://arstechnica.com/information-technology/2025/12/how-do-ai-coding-agents-work-we-look-under-the-hood/?comments-page=1#comments
[33] https://www.msn.com/en-us/technology/artificial-intelligence/bernie-sanders-calls-for-robot-tax-to-protect-workers-from-the-impacts-of-ai/ar-AA1O5s7I
[34] https://www.konicz.info/2017/11/15/kuenstliche-intelligenz-und-kapital/
Originally published on konicz.info on 01/05/2026
EXIT! in English 