The combined financial valuation of the three most important Big Tech companies in the US — OpenAI, Anthropic and SpaceX — is approaching $4-trillion, about the size of Japan’s GDP. (Image: ChatGPT) There is a curious conversation that has started in the halls of technology and government power. One that would have been unthinkable even a year ago. It is this: big technology companies, particularly the frontier AI giants, are becoming so powerful and so important to the future economic life of the average citizen, that they are in danger of bursting out of their traditional capitalist corporate skins.
Perhaps the Big Tech companies need to cede some of their wealth and power to the state, and onwards to the citizens who will use their products.
That’s the theory, anyway. And the most surprising people are warming to the prospect, even among the billionaire class.
Let’s start with this. The combined financial valuation of the three most important tech companies in the US (OpenAI, Anthropic and SpaceX) is approaching $4-trillion (R65.25-trillion).
How big is that?
It is about the size of India’s entire GDP.
About the size of Japan’s GDP.
About 15 times the size of this year’s estimate of global IPO proceeds.
Big Tech’s impact on humanity
It’s not just the money though. One does not have to read very deeply to understand that the impact of these technologies on the polity will be like nothing else experienced in human history on a timescale that seems to compress daily.
These are not ordinary companies; their growth and power have no precedent. Perhaps they should be treated differently from other companies. Everyone, including the founders, is beginning to come to this realisation.
I have written previously about the podcast Moonshots, which I listen to regularly because it is home to the most aggressive of the “accelerationists”, those tech bros for whom the imminent arrival of artificial general intelligence and the certain manifestation of a utopian science-fiction future are indistinguishable. I listen to it because it gives me a view of what is possible if the stars align and it allows me to (briefly) tamp down my natural scepticism.
The panellists on this podcast have historically been blisteringly critical of government meddling anywhere in tech development. They are rethinking this position.
Which is why this particular podcast episode was so startling – a sober debate about the relinquishing of equity stakes in these frontier companies to the government (deftly skipping over the slippery conflict of having an equity partner who is also the policymaker and regulator).
Sanders floats idea of one-time tax on Big Tech
At one point, the conversation turned to a proposed bill (authored by Senator Bernie Sanders) in which a one-time 50% tax (payable in stock) would be levied on specific companies, to be lodged in a public fund holding voting rights and board seats and providing citizens’ dividends. The panellists agreed that 50% was a non-starter, but the principle was not contested – they proposed a more realistic version.
Not Sanders’s half, but a 5–10% stake, imposed not as expropriation but as a condition of doing business – the price of operating on American soil, using American data, drawing on American energy. That is no longer agitprop from the campus left. It is a sober prediction from people who spend their days raising and deploying capital. (I should note here that in China, there is no debate about this because the CCP is expected to be a shareholder in similar companies).
Here is the wider rationale – modern states derive their legitimacy, and indeed their fiscal stability, from broadly shared economic prosperity. If AI really does become as transformative as its proponents predict, and if the gains from that transformation accrue disproportionately to the shareholders of a handful of frontier AI companies, governments will eventually come under pressure to do more than merely collect taxes. They will be asked why the public, which funded much of the underlying scientific research and will bear many of the social costs of disruption, owns none of the asset itself.
The intellectual foundations of AI were built over decades through publicly funded university research, government grants, defence spending and taxpayer-supported infrastructure. Why should the upside be privatised if so much of the initial investment was socialised?
The question of state equity cannot be simply waved away as “socialism”. When a company’s product is simultaneously a consumer chatbot and a strategic weapon, the tidy separation of private capital as allocator and government as regulator stops describing reality.
Trump knows this. Aboard Air Force One in early June, he floated taking stakes in OpenAI, Anthropic and xAI, suggesting the taxpayer should be “essentially a partner”.
The groundwork was already laid – a 10% position in Intel taken in 2025, smaller holdings across semiconductors, quantum and critical minerals, and a February 2025 executive order calling for a national sovereign wealth fund.
The Moonshots panel prediction of a modest, condition-of-operation stake is simply an extension of this trajectory.
That this argument is beginning to emerge from both ends of the political spectrum (Trump and Sanders) is a sign that something interesting is happening. It may be tempting to dismiss all of this as political theatre. But there is another way of looking at it.
AI as national infrastructure
Perhaps governments have quickly recognised that AI companies are becoming a form of national infrastructure. Not infrastructure in the old sense of roads, bridges and dams, but cognitive infrastructure – the systems that will increasingly mediate education, healthcare, defence, public administration and scientific research.
The value of OpenAI or Anthropic may ultimately have less to do with selling software subscriptions than with becoming the operating layer upon which society runs.
There is an obvious counterargument. State ownership has a habit of producing complacency, bureaucracy and general dispiriting inefficiency. Governments that own AI companies may become reluctant to regulate them properly. Politicians may start treating frontier models as instruments of industrial policy rather than engines of innovation.
Investors, meanwhile, will understandably recoil from the idea that successful private enterprises can be compelled to surrender equity simply because they have become too successful. And there is this – why only Anthropic and OpenAI and xAI? Surely there are others that qualify? Where is the line drawn?
There are real concerns about government intervention, but they may be less important than they first appear. The history of capitalism is not, in fact, a history of rigid separation between state and market. It is a history of continual renegotiation. Whenever a technology becomes sufficiently consequential – whether railroads, banking (saved in the US in 2008 by state-funded equity intervention), nuclear energy or telecommunications – the state drifts from regulator towards partner, guarantor and, sometimes, owner.
The strange conversation on Moonshots may prove to be one of those moments that only makes sense in retrospect. A handful of giddy and very smart technologists were talking about AI valuations and foundation structures and future partners.
But what they were really talking about was the birth of a new kind of institution – private companies that have become too important to remain private. DM
Steven Boykey Sidley is a professor of practice at JBS, University of Johannesburg, a partner at Bridge Capital and a columnist-at-large at Daily Maverick. His new book, It’s Mine: How the Crypto Industry is Redefining Ownership, is published by Maverick451 in South Africa and Legend Times Group in the UK/EU, available now.