Big Tech has looted humanity’s creative output to create artificial intelligence. We don’t just deserve a share of AI’s profits — we deserve control over it.
Elon Musk is officially the richest man on Earth. The source of his wealth is not his own cunning. Nor can it be fully explained by capitalism’s financial architecture, tax loopholes, and the value created by the workers employed in his various ventures.
Tech billionaires’ companies are robbing humanity of its distinctive capacity to create. From social media data to academic outputs, literary pieces, artwork, movies, music, podcasts, software code, our health care records, you name it: generative AI models are trained on creative products from around the world. Once operational, data retrieved from user prompts is used to fine-tune them, improving AI’s predictive capacities. AI is, unmistakably, a product of humanity.
Bernie Sanders’ call for levying a one-time 50 percent tax on AI companies like OpenAI, Anthropic, and xAI — to be paid in stock — is, seen from this perspective, a reasonable compensatory measure.
More Than American Data: A Global Looting #
There are even more compelling reasons to justify shared ownership. AI models are cocreated by the many and then monetized by the few. The very conception of AI models and their underlying technologies rely on a global network of scientists and engineers based at universities, public research organizations, and thousands of companies at least partially funded by taxpayers’ money. These actors profit very little, if they profit at all, and have virtually no say in what models are developed or how.
Elsewhere, I have shown how Microsoft and Google dictate the agenda of the frontier AI research global community. These and other giants like Meta do open-source-washing. They open-source strategic software pieces — sometimes using a form of pseudo-open-source — to turn them into de facto standards. Massive adoption leads to a captured user base, one that will more easily consume other bits of their technology — these kept as proprietary software. Big Tech companies’ additional strategy to control technology and other organizations beyond ownership has made the headlines. They are the largest corporate venture capitalists, investing in thousands of start-ups to get access to and steer their developments. This is how companies like OpenAI and Anthropic came to be.
Bernie’s diagnosis is correct: AI is indeed a product of humanity ripped off by a few corporate giants. But he falls into a nationalist trap that has even led the Financial Times to associate him with Donald Trump when suggesting that Americans should own half of AI companies’ stocks. The creative products used for training, as well as researchers, developers, and start-ups integrating Big Tech’s controlled networks are not only American. They come from around the world. The AI sovereign wealth fund should thus belong to the true public, to the world’s population.
Geography is only one of the reasons why the scope of Bernie’s proposal matters. If the aim is to grant humanity both ownership of benefits and control over the development of AI models, getting a cut of the AI companies going for initial public offerings this year would be far from enough. The ultimate AI steersmen and main beneficiaries are Nvidia and the cloud giants. Every single AI model can be used as a service on at least one of the world’s largest tech supermarkets, aka Amazon, Microsoft, and Google clouds. Using AI means using their clouds. AI adoption takes place through their clouds. The more models are released, the more cloud giants’ predatory ecosystems expand. And while cloud giants also develop their own models, start-ups depend on Amazon, Microsoft, and Google infrastructures and cloud marketplaces. From known dependencies, like OpenAI’s reliance on Microsoft Azure, to DeepSeek and xAI’s Grok.
And it is precisely their clouds’ data centers that complete the spoilage. This represents an extraction of nature that compounds the appropriation of data and knowledge. Their expansion should be regulated, and society should also be compensated for their use of land, energy, and drinking water. An international AI sovereign wealth fund could compensate for this twin extractivism of nature and human creations. Still, it would not grant humanity the decision-making power.
Distributed Ownership, Concentrated Control #
At its debut, individual investors — meaning laypeople instead of institutional investors such as hedge funds or asset managers — were allocated between 20 and 25 percent of SpaceX stock. The illusion of distributed ownership begins to fade when compared with Musk’s 42 percent, resulting in 80 to 85 percent voting rights. Ownership does not equal control. Like Meta, Alphabet, and Palantir, SpaceX went public offering different classes of stocks. Each class has different voting rights, preserving control in the hands of founders and CEOs. Musk’s shares have ten times more voting power.
Concretely, even if Bernie’s proposal succeeds, Musk would still retain the control over what SpaceX does and how it operates. For one thing, it would be unlikely for the AI sovereign wealth fund to block SpaceX from building extraterritorial data centers, regardless of their environmental impact. Likewise, the public will have zero bargaining power to prevent a company like OpenAI from selling its advanced models to the military; deter Palantir from working side by side with Immigration and Customs Enforcement and, again, the military; or convincing Anthropic to stop developing models that could induce a cybersecurity catastrophe.
Society cannot trust the willingness of a few CEOs and founders to halt the development of AI that can be used as an autonomous weapon, as a tool to expand control in the workplace, or as a school assistant that spoon-feeds students and does their homework. By themselves, these people will never slow the pace at which their own companies develop technology and expand scale. Scaling up — concentrating more data to train ever larger models in massive data centers with greater numbers of processors — is strategic for cloud giants and AI companies. It is an exclusionary system by design that ensures their lead.
Scale enables control without ownership across the networks of organizations described above. Thousands of actors integrate innovation systems dictated by a few tech giants. Following the latter’s research and development and business priorities, these ecosystems also contribute to developing technologies that reinforce military and workplace control, exacerbate disinformation, and undermine people’s capacity to think. These are not inevitable or accidental outcomes. Big Tech companies plan science and technology aiming to produce AI that renders users dependent, and that advances the corporate and political interests of those with the money to afford it.
Another AI agenda cannot wait. To prioritize communities and the planet, the public must be in control. International democratic control over the AI ecosystem is at least as urgent as redistributing its benefits. It would require creating public institutions that are both independent from corporate power and autonomous from the whims of individual governments.
An AI sovereign wealth fund could thus only be celebrated as part of a broader international strategy. As a stand-alone initiative, it risks backfiring. Just as sovereign oil funds can dampen popular enthusiasm for a renewable energy transition, a sovereign AI fund could bolster the adoption of today’s AI, the world’s largest control technology. A just society is not only one in which wealth is redistributed. Control, thus decision-making power, and knowledge should be equally shared.