Is Big Tech pulling up the draw bridge and breaking history? (Part 1)

By Alicia Lucas 

Big Tech “artificial intelligence”-labelled models, now labelled “super intelligence” by the USA’s President Trump, incorrectly (its not a collection of human brains) purportedly becoming closer to bettering human intelligence but increasingly eroding human processes important to the production of these models. This would lead to the loss of skills and knowledge to understand, critique and compete with Big Tech as well as shrinking the scope for alternative futures. They are doing this despite increasing numbers of lawsuits and government inquiries, topped off with the antisocial ability to hack, without demonstrating real, substantial improvements that benefit communities.

Computer programming, USA open source software, cybersecurity and mathematics are becoming increasingly dominated by these companies and may figuratively become locked within their castle walls. Amongst other issues, the strategies Big Tech companies may use, marketing hype, automation, encroachment followed by replacement and buying up, stand out with automation and replacement possibly leading to a disconnect between the human histories of these areas and the history that will result if Big Tech becomes their controlling force.

The financial situation of the two most well known Big Tech “AI”-labelled modelling companies, OpenAI (ChatGPT model) and Anthropic (Claude model) appears precarious. The risk of an “AI” bubble continues and neither company has reached sustained, unquestionable profitability. Surveys have indicated products labelled “AI” are not used in significant money earning ways nor leading to economic improvements in most private businesses. The Big Tech “AI”-labelled modelling companies owe investors lots. So far in 2026, OpenAI received US$122 billion and Anthropic over US$120 billion of capital investment mostly from other Big Tech companies and venture capitalists. OpenAI, is delaying going onto a national USA stock exchange and may hold another funding round where it will be looking for more billions while Anthropic is still wanting to raise funds as a public company by selling shares on a stock exchange.

OpenAI and Anthropic appeared keen to go public in June 2026 and in doing so entice public investors to risk their money by buying shares each company would make available on a stock exchange (and, perhaps, decreasing the risk of existing investors). Both companies provided a registration document to the US Securities and Exchange Commission (SEC) to confirm their interest. This did not commit them to a timeline to go public or to actually go public, however, on the 26/10/2026 Anthropic informed the SEC they want to go ahead.

OpenAI faced an almost immediate setback when 42 out of 50 USA State Attorneys General subpoenaed the company regarding a number of concerns including user safety, data collection and ads. Notably, OpenAI was sued by the USA Florida State Attorney General basically for not being safe for children. This Attorney General had earlier commenced a criminal investigation into OpenAI for allegedly “aiding and abetting” the mass shooter who killed and injured people at the Florida State University in 2025. Anthropic faces a class action lawsuit filed near the end of September because it subjected a user in Illinois, USA, to a facial recognition scan while displaying  his drivers licence and entering a government-id to unlock his Claude account. Both were not in keeping with privacy legislation reports the Courthouse New Service.

Now the recent spate of models accessing, for example, government and not-for-profit websites, without permission commonly known as hacking, has also somewhat tempered enthusiasm for the companies to go public. The full extent of websites attacked around the world is unknown but identified sites include those of Hugging Face, a site for open source “AI” development, and the SEC in the USA and Australia’s Medicare research portal. The cause of these hacks has been identified as poor security and safety practices by the Big Tech companies concerned, particularly serious in the context of the companies testing the ability of a model to exploit software weaknesses. The models were doing what they were programmed to do – programmatically carryout hacking cyber tests.

OpenAI is being sued for the Hugging Face hack and other hacked organisations may also sue. Lawsuits may not be won but they can bring forth information in the form of documents or testimony that a company may not willingly supply otherwise. The hacks have set off or added to a number of inquiries into the conduct of OpenAI including by USA State Attorney Generals and the US bipartisan House Security Committee. OpenAI has been subpoenaed by two states, Alabama and California. The USA Federal Trade Commission, a consumer protection agency, is now investigating OpenAI, Anthropic and others for possible threats to consumers (Associated Press 1/10/2026). The USA FTC has not identified the investigation on their web pages at the time of writing.

Before the hacking and lawsuits started to put even more question marks against these two companies, Professor Emiritis Andrew Odlyzko, School of Mathematics, USA University of Minnesota, in a podcast (AI, hype, and the history of tech bubbles: what past mania can teach us about the data center build) in March 2026, pointed to two reasons he felt the “AI”-labelled industry was in a bubble based on his economic and financial studies that include technical bubbles. One, the companies were increasingly seeking investment from individuals, retirement funds, credit funds and similar, instead of just other Big Tech companies and venture capitalists. This is shown by the desire of both companies to be listed on a stock exchange where shares can be more legally and/or safely brought by a wider range of organisations and the public. Reporting of the potential for Anthropic and OpenAI to go public can be found in January 2026. The second reason is the reports of “diminishing returns” by these companies where smaller and smaller gains were being made even though more and more was being put into producing the models.

He is not alone in talking about bubbles. Another is British writer, Ed Zitron, now living in the USA, who has been warning about this possibility and has only become more apprehensive after analysing a copy of OpenAI’s finances that were leaked to him. He is concerned the amount of money invested in OpenAI cannot be met through revenue. He points out that even with increasing numbers of customers the amounts paid are not enough to meet the costs of running the models as the costs grow each time it gains a new customer. He argues this is true of Anthropic and after Reuters published, in instalment, Anthropic’s leaked financial data showing an increase in revenue  Zitron has argued Anthropic (and OpenAI) should not go public. There are still details to be provided to the SEC as part of the stock market listing process and this may need to be review with the financial data provided to the SEC but reported revenue and losses do not appear to contradict Zitrons interpretation. The SEC needs to agree to the company going public.

While OpenAI and Anthropic may not be encouraging investment by retirement funds and so forth, a lot of people’s retirement plans may have to change if the bubble bursts. Despite the risks involved in investing in private companies, for instance, the financial situation may not be clearly stated, Zitron writes that many retirement and insurance funds are already investing in them instead of waiting until they become public. Zitron provides examples such as the Japanese company that invested its retirement funds into building data centres whose infrastructure manager is an Australian company. There are other companies in Australia involving themselves similarly. Canadian retirement funds are being invested into data centres elsewhere by Australian property managers Goodman’s Group. Goodman’s Group pulled out of a proposed data centre to be built near residents in Sydney and is facing opposition from the Japanese community for a data centre being built at Tsukuba (News, events and Jobs for Tsukuba Residents-Latest Data Center news-Facebook 29/9/2026). A superannuation company, AustralianSuper, also want to invest in data centres in Japan as well as India.

If the bubble bursts it is these companies and their investors and ultimately the retirees that wear the cost if things go wrong. The data centres, themselves, cannot be simply repurposed for other IT uses as the chips, GPUs, are built specifically for “AI”-labelled model use such as training according to Zitron and others. Google’s TPUs sound even more specialised. These companies, particularly property development types, and their investors are likely the ones also wearing the cost of dismantling the infrastructure or selling the land off cheaply.

Whether the bubble bursts or not, computer programming, USA open source software, cybersecurity and mathematics all appear to be suffering from a Big Tech still living up to the motto “move fast and break things.”

Continued tomorrow… Computer Programming