By George Banks
The world has entered a period of deep geopolitical and economic turmoil. The increasing inter-imperialist tensions, seen in Trumps trade wars, in Russia’s invasion of Ukraine, and the Israeli genocide in Palestine, combined with the general sluggishness of the major economies, have created an increasingly unstable economic situation.
The US stock market, and by extension global exchanges are being propped up by the so called ‘Magnificent Seven’ tech companies – Nvidia, Microsoft, Alphabet (parent company of Google), Meta, Apple, Amazon, and Tesla. Of these, Nvidia, Microsoft, Alphabet, and Meta, are heavily reliant on the potential of AI for their future growth. Lately Apple and Amazon have also rushed to join the AI feeding frenzy.
The Magnificent Seven account for 37% of the S&P 500 as of 21 October 2025. That’s close to three times as much as in 2015, when they made up 12% of the S&P 500 — an unprecedented rate of growth among a handful of stocks.
Despite the torrent of investment, the most credible warnings of a bubble come from the architects of the AI boom themselves. OpenAI CEO Sam Altman concedes the speculative mania, telling reporters, ‘When bubbles happen, smart people get overexcited about a kernel of truth… Are we in a phase where investors as a whole are overexcited about AI? My opinion is yes.’
Similarly, Meta’s Mark Zuckerberg has acknowledged that the surge in investment ‘stands to form a bubble,’ and empirically, based on past infrastructure buildouts, he believes ‘that something like that would happen here.’
Yet, their strategy is to charge ahead regardless. Zuckerberg insists the risk of a crash is preferable to being late to an ‘era-defining technological transformation’, hoping AI will be an ‘outlier’ where endless growth prevents collapse. Some hope!
Many economists calculate that the AI boom accounted for more than half of America’s 1.6% growth rate over the first six months of 2025. They see the trend continuing into the third quarter at least. Data for that period would have come out this week, if the agencies that publish it weren’t shuttered amid a budget deadlock in Washington.
AI investment is not confined to the US, and companies have big plans to expand into foreign markets. In Britain, Microsoft, NVIDIA, Google, and OpenAI have committed a combined £31 billion to boost AI infrastructure. OpenAI signed a deal with the British government to explore investing in AI infrastructure. The company will expand its London office and partner with the Ministry of Justice to use AI tools, such as ‘justice transcribe’ for probation officers.
Nvidia, Microsoft, OpenAI and Nscale, the British builder of data centres and chips, are deploying 120,000 advanced GPUs (processors, essential for AI computing capabilities) across new data centres in the UK. This has now increased to 200,000 following a new deal struck with Nscale, Nvidia and OpenAI.
The hope of the major capitalists is that all this investment will lead to a technological leap forward to an AI which can outstrip human intelligence, a concept referred to in the industry as ‘artificial general intelligence’. In theory such technology would render vast swathes of human labour obsolete. Whichever capitalist develops this technology first would be at a huge advantage compared to their rivals. However, there is no evidence that such a leap forward is even possible.
The bubble
At the heart of this frenzy is a closed loop of speculative investment, reminiscent of the worst excesses of the dot-com bubble. The AI ecosystem is dominated by a few major players, chief among which is Nvidia, the company that manufactures the GPUs.
Nvidia’s astronomical rise, from $10.9 billion in revenue in 2020 to a projected $130.5 billion in 2025, demonstrates the sheer scale of the capital being sunk into hardware. For reference, a single Nvidia H200 GPU, the top of the range technology used in data centres, costs about $32,000. A single data centre requires thousands of these.
However, a clear warning sign flashes in its Price-to-Earnings ratio (P/E, a measurement of how inflated the stock market price is compared to actual profits) of around 58. This is more than triple the traditionally healthy ratio of 15-20, indicating that its stock price is driven by speculative hype, not underlying profitability.
Nvidia holds a de facto monopoly on the advanced processors required for AI data centres, making it the only viable supplier for this global build-out. The ecosystem supporting this infrastructure includes major players like Oracle, the 15th largest company globally, and AMD. This monopoly position is facilitating an unprecedented capital expenditure, with Google, Meta, Microsoft, and Amazon projected to spend nearly $400 billion on data centres this year alone.
The construction of these facilities is a global phenomenon with severe material and ecological consequences. They consume vast tracts of land, immense human labour, and finite resources. Furthermore, they are a significant source of electronic waste, consume colossal amounts of increasingly scarce water for cooling, and rely on critical minerals and rare elements extracted through often destructive mining practices.
Most alarmingly, their enormous electricity demand is directly driving up the emission of planet-warming greenhouse gases. The fact that the chips in these data centres depreciate extremely quickly, due to the speed of innovation in this area, adds to the pace of production needs.
The AI industry is propped up by a circular and precarious system of vendor financing. A prime example is Nvidia, the world’s most powerful company, investing $100 billion into OpenAI, which then funnels money to partners like AMD and Oracle to build data centres that ultimately purchase their essential chips from Nvidia, creating a dangerously self-referential loop of investment.
The most vulnerable element of the entire AI ecosystem is the application layer itself, typified by its largest player OpenAI. Despite creating household-name products like ChatGPT and Sora, the company is fundamentally unprofitable. OpenAI lost $5 billion last year and is projected to lose a further $14-15 billion this year, exemplifying the core contradiction – no company has yet discovered a viable model to generate profit from AI technology.
The central justification for such immense investment is that AI will unleash a wave of productivity, revolutionizing every industry, but this has yet to be seen. In these projects, the practical outcome in increasing productivity in the sphere of production has not been measurable yet. If this results in substituting living labour, whilst increasing the proportion of constant capital, in the end this will lead to a fall in profit rates, the underlying cause of crises.
The reality is that these ventures are currently financial failures. Crippling operational costs mean even paid services like ChatGPT lose money. Microsoft, Google, and Meta have spent a combined $227 billion to generate just $13-14 billion in return. To justify this spending, the industry must generate an implausible $1.5 trillion in revenue by 2030, revealing a vast gap between speculative investment and tangible economic return.
This frenzy mirrors the dotcom and cryptocurrency bubbles, where stock prices were driven not by fundamentals but by speculation, excitement, and a pervasive fear of missing out. Far from being a warning sign, each new vendor financing deal, like Nvidia investing in OpenAI, triggers a massive sector-wide surge in asset values, fuelled by investors who understand little about the technology’s limitations but are captivated by the illusion of a graph that only goes up.
For a Socialist Plan of Production
The current AI boom has less to do with creating real value and everything to do with speculation. Karl Marx’s analysis of ‘fictitious capital’ is more relevant than ever. He described how financial markets create a world of value detached from the surplus value produced by human labour power. Speculation embodies the capitalist desire to ‘make money from money,’ the buying and selling of assets based on inflated expectations.
In a speculative market, such as the AI industry, asset prices are driven by expectations of future price increases rather than the actual value derived from labour and production, leading to a disconnect between price and underlying value.
This is amplified by the operation of credit, which allows individuals to make enormous bets far exceeding their own capital. This enables a few individuals to create enormous transactions with no relation to the productive economy. This is one of the key drivers of capitalist crises.
The AI bubble is not an anomaly; it is the logical endpoint of a senile capitalism. It demonstrates a system capable of marshalling trillions of dollars, the world’s brightest minds, and vast material resources not to solve human need, but to pursue a phantom of profit. When the bubble bursts, it will be the working class that pays the price through job losses, slashed pensions, and another round of austerity to bail out the system.
The technology itself holds potential, but it cannot be fully realised under this chaotic and exploitative system. The workers of the world must demand an end to this speculative madness. We need to seize these powerful technologies from the hands of the billionaires and integrate them into a rationally planned socialist economy, where AI is developed and deployed democratically to shorten the working day, abolish dangerous, and physically and mentally exhausting toil, solve pressing social problems, and liberate humanity for creative endeavour, rather to inflate the stock portfolios of the capitalist class.
