The presence of Big Tech leaders in the front row at US President Donald Trump’s second inauguration signals their place in a ruling coalition with tremendous power. Through their control over the commanding heights of tech capital, this coalition has proven itself capable of subduing nearly every organization and nation reliant on AI and other digital technologies.[1] Given their role in global capitalism, this coalition undermines most of the world’s (digital) sovereignty.
The US government’s blatant support for US corporate giants — partly explained by China’s prowess — facilitates their control of the tech ecosystem beyond ownership, driving widespread adoption of American technologies. Here, “US Big Tech” is to be understood as denoting the world’s largest tech companies in terms of market capitalization, among which cloud giants (Amazon, Microsoft, and Google) loom large. These companies dictate the social production of information and knowledge, from narrative-building in social media to advanced AI technologies for warfare, surveillance, and control in the workplace, while evading regulatory efforts both in the US and elsewhere.
The main problem underlying this setting is that what contemporary technology societies need, and how technology can be used, is decided in the most undemocratic way behind a few corporate and political doors. Corporations’ authoritarian and expanding power plays right into the hands of the far right, who are eager to find further ways to undermine already insufficient capitalist democracies. In the recent past, tech could at least partly be associated with progressive thinking, for instance due to the free software movement. But we have since witnessed a political U-turn to conservative, far-right ideologies among tech company leaders and senior management.
Amid this context, figures like Elon Musk or Peter Thiel combine anti-state and antidemocratic discourses with techno-solutionism. Others in the tech world offer a less anti-state yet equally concerning tech-solutionist discourse: to govern effectively and efficiently — two adjectives imported from the private sector that neglect the purpose of public governance — states must adopt Big Tech’s AI and digital technologies. Satya Nadella (CEO of Microsoft) and Sundar Pichai (CEO of Google and of its parent company Alphabet), among others, embrace this more conciliatory narrative as they push governments around the world to migrate to their clouds.[2]
As tech leaders’ turn to the far right makes the headlines, it is worth recalling that the tightening and yet conflicting partnership between US Big Tech and the US government is not new. Exploring its recent history uncovers the latter’s regulatory capture and Big Tech’s capacity to effectively govern behind the scenes since before Trump’s second term. A historical account can equally shed light on what has changed in the relation between Big Tech and the US government when compared to previous administrations.[3] Taken together, this historical reconstruction points to a “frenemy” relation, with episodes of reinforcement of each of the parties’ respective power and episodes in which they cannibalize one another.
Regardless of how this alliance unfolds, there is an urgent need to create a democratic alternative to Big Tech’s controlled digital networks. To start building this, governments should levy specific taxes to partially compensate for the extractivist relation to intangibles (data and knowledge) and nature (electricity, water and mining extraction for datacenters and other infrastructure). While a tax will not overcome entrenched technological dependencies, it will partly make up for captured resources. It should be levied in coordination among likeminded governments that share the goal of prioritizing people and the planet.
More importantly, governments must decouple from these companies’ predatory ecosystems, partly through relying on the revenue from such a tax. Decoupling from Big Tech requires not only the use of open source or other alternative providers but, above all, the democratic definition of the priorities and types of technology to be adopted and developed. This requires three complementary policies. First, governments should work together to develop an alternative to the cloud as the economic space in which all the pieces of the digital stack come together. Second, they should use public procurement to steer the development of digital technologies by prioritizing public value over private gains or market efficiencies. Third, governments should establish a new institution for the democratic setting of the digital research agenda. This institution should be based on international collaboration, solidarity, and interdisciplinarity for mapping and advancing key research and development priorities.
The remainder of this report is structured as follows. In Section 2, I briefly recap the recent history of the relations between the US government and US Big Tech. This is followed by a section explaining how the promotion and regulation of AI have been co-created with Big Tech companies. Section 4 examines the close and expanding ties between US Big Tech and the US Department of Defense (recently renamed Department of War). Section 5 explores what has changed since Trump’s second presidency. I conclude by stressing the need to build an alternative digital ecosystem that is democratic, public-led, people-centered, and remains within planetary boundaries.
US Big Tech would not exist without the active, albeit hidden, industrial policies of the US government, which were channeled through the military during the Cold War.[4] By 1960, 80 percent of all the federal research and development (R&D) investment came from the Department of Defense and most of this money was funneled to elite universities and a selected group of leading corporations. By the mid-1960s, the US government paid for two thirds of the R&D conducted by electric equipment companies, which back then included giants like IBM, AT&T, and General Electric.[5]
Disguised by dominant pro-free-market ideology, the US state’s capacity to direct technological change was essential in achieving major breakthroughs, particularly in biotechnology and ICT.[6] Military initiatives promoting ICT are a central precedent of the US’s current leading position in AI. The Defense Advanced Research Projects Agency (DARPA) launched the first versions of the internet, called Arpanet, while the Pentagon successfully engaged university research to develop related technologies. The public sector in the US was also the main consumer of cutting-edge products for decades, from semiconductors to biotechnology.[7]
Several other legal and institutional transformations further contributed to a more stringent and extensive Intellectual Property Rights (IPRs) regime in the US including the authorization to patent publicly funded research, living beings, and software. The latter had been unbundled from the hardware, since then sold as separate products, a decade earlier. In January 1970, IBM separated hardware from software in order to avoid an antitrust lawsuit from the US Department of Justice and to reduce competitors’ claims for damages.[8] While that episode signaled that at least fractions of the public sector were opposed to the emergence of large tech players, the transformations of the intellectual property regime that followed from the 1980s onwards clearly favored tech and other corporate winners. This resulted in a national innovation system that revolves around corporate interests and that — sometimes even unintentionally — feeds science and technology to US leading corporations to reinforce their intellectual monopolies.[9]
The US IP regime was internationalized from the mid-1990s onwards with the signing into law of The Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS). TRIPS was originally drafted by corporate behemoths IBM, Pfizer, and Microsoft. Follow-up treaties further transformed knowledge into independent entities that could be owned, possessed, and controlled.[10]
Other state-led changes that have further favored leading corporations since the late 1970s include the weakening of antitrust controls,[11] and the development of complex forms of corporate arbitrage.[12] Large platform companies have also benefited from a digital policy vacuum. Delays in the establishment of clear rules on who could collect data and what types of data could be harvested have favored their practices of data accumulation. They have further benefited from the absence of trade policies for governing digital services exports and imports, and the difficulty of accurately measuring these transactions.[13]
Big Tech corporations undemocratically govern the internet thanks to the abdication of the US government, which has in turn reinforced its global dominance through these giants. Among others, the US government profited from their data to advance state surveillance. A neglected phenomenon as compared to the link between the Chinese state’s surveillance and its tech giants; this relationship became apparent with Edward Snowden’s disclosures. These leaked files included evidence of Google, Apple, Facebook, and other US internet companies providing data access to the US National Security Agency (NSA) for global surveillance.
Data access is only one of the multiple channels through which the US government and Big Tech became close allies. Known examples include the US Immigration and Customs Enforcement agency’s contracts with Palantir, which at least date from 2020, and several US police contracts with Big Tech.[14]
Yet, their intertwinement goes beyond procurement contracts for surveillance. Big Tech’s CEOs, founders, and executives have been regularly invited to advise and sit in on the policymaking process since long before Elon Musk became the head of the US Department of Government Efficiency (DOGE). Among other examples, Eric Schmidt and Bill Gates were invited by the New York Governor to “re-imagine education” after the Covid-19 pandemic, by introducing new technologies in the classroom.[15] Schmidt was then Google’s CEO and Executive Chairman. He also served as an Apple board member for three years during his — much longer — tenure at Google. Gates meanwhile likely needs no introduction. From all Big Tech’s spaces of influence, a particularly relevant one has been their role in shaping the US AI policy.
It has been a decade since the 2016 US National Artificial Intelligence Research and Development Strategic Plan defined AI as “a transformative technology that holds promise for tremendous societal and economic benefit.” A parallel report by the President of the Council of Economic Advisors similarly emphasized AI’s potential impact on productivity and growth. The US government’s insistence on the promises and centrality of AI has only grown since. AI was featured prominently in the FY 2020 Administration Research and Development Budget Priorities memo, and the criticality of AI for military applications was also stated in the US National Defense Strategy in 2018.
As AI won space in policy debates, so did the influence of Big Tech in defining the US AI policy. In 2018, the National Security Commission on Artificial Intelligence (NSCAI) was formed. Chaired by Eric Schmidt, this commission also featured top executives from Google, Microsoft, and Amazon among its members. The NSCAI wrote two reports for the US President and Congress with recommendations on ways to advance AI development for national security and defense.
The NSCAI’s main message was that the US was unprepared to defend its lead in the AI age and that this would come at the expense of its economy and democracy. Its overall assessment is synthesized in the following quote:
[f]or the first time since World War II, America’s technological predominance — the backbone of its economic and military power — is under threat. China possesses the might, talent, and ambition to surpass the United States as the world’s leader in AI in the next decade if current trends do not change. Simultaneously, AI is deepening the threat posed by cyber attacks and disinformation campaigns that Russia, China, and others are using to infiltrate our society, steal our data, and interfere in our democracy.[16]
The NSCAI argued that the US must overcome its vulnerabilities regarding the three main ingredients of every AI model: data, computing power, and talent. Their final document gave recommendations to tackle each of these issues. Their final report also argued that the US government must invest in AI R&D, particularly infrastructure, and suggested a reorganization of the US DoD and Intelligence Community to ensure that by 2025 they were AI-ready.[17]
In the 756 pages of its final document, the NSCAI provides detailed instructions on the role of each actor in the AI ecosystem. Concretely, most AI R&D, while financed by the state, should be done by firms and universities. The report also calls for more stringent IPRs for AI, data, and biotechnology. Explicitly referring to Big Tech’s capabilities in the tech war with China, the report claims that “even large tech firms cannot be expected to compete with the resources of China or make the big investments the U.S. will need to stay ahead. We will need a hybrid approach meshing government and private-sector efforts to win the technology competition.”[18]
This hybrid approach is quite beneficial for Big Tech: while the US government takes the economic risk and makes colossal R&D investments, they retain and strengthen the means to collect associated gains. Overall, the NSCAI report makes a strong case for techno-nationalism. In relation to chips, the NSCAI report concludes that “the United States should commit to a strategy to stay at least two generations ahead of China in state-of-the-art microelectronics and commit the funding and incentives to maintain multiple sources of cutting-edge microelectronics fabrication in the United States.”[19]
In line with these claims and also with the broad diagnosis of the US weakness and the policy recommendations suggested by the NSCAI, in 2021 Biden launched a 2 trillion USD infrastructure plan called “The American Jobs Plan” (AJP). To overcome what the plan described as the Chinese threat, it committed 180 billion USD for R&D in AI and biotechnology. Another 300 billion USD was committed to manufacturing subsidies, including major subsidies for chipmakers. While the AJP was not approved as such by the Congress, the focus on chips and tech-related R&D became the primary goal of the US Chips and Science Act.
The narrative that China could surpass the US in digital technologies, posing a major threat to democracy, and that US Big Tech companies are the best defense against this threat, has been a consistent argument used by these corporations to discourage Western governments from regulating them. This rhetoric was vividly showcased during the 2020 congressional hearing as the primary reason why US Congress should avoid acting against Big Tech.
By early 2023, a new US report on AI came out entitled “Strengthening and Democratizing the U.S. Artificial Intelligence Innovation Ecosystem. An Implementation Plan for a National Artificial Intelligence Research Resource” (NAIRR). The NAIIRR task force was composed of current and former Google employees and made consultations with Microsoft and Amazon. Among its recommendations on how to produce and use AI for research, the report suggested expanding the use of public clouds in connection to the extended use of AI:
To facilitate a federation of existing on-premise and commercial cloud resources, established Federal agency programs could be leveraged. For example, the NIH Science and Technology Research Infrastructure for Discovery, Experimentation, and Sustainability (STRIDES) Initiative program provides access to Amazon Web Services, Microsoft Azure, and Google Cloud resources.[20]
The NAIRR recommendations came out in January 2023, less than two months after Chat GPT’s first generative AI interface was introduced (in late 2022). By the end of that year, the US government introduced its executive AI order on the “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.” Big Tech’s influence can be read between the lines of this regulation. The order does not mandate AI companies disclose the data that was used for training their models or indeed the model parameters. Instead, it mainly regulates harmful AI uses. This was precisely what Big Tech had lobbied for given that AI regulation was perceived as inevitable under Biden’s administration.
Months ahead of this regulation, during an interview with CNBC on May 16, 2023, Microsoft’s CEO Satya Nadella had made an open case for regulating AI applications, remaining silent about the regulation of AI production where Microsoft looms large:
We do not have to wait for regulation to have standards or adopt the misstandard, I call it the start of any self-regulation. Then, on top of that, if we talk about regulation, maybe we can unpack it from the application domain, because after all context in which something is being applied, in education, healthcare, retail and we can have the regulatory frameworks that already exist.[21]
Furthermore, when the order was introduced in late 2023, then US Vice President Kamala Harris declared, “Let us be clear: when it comes to AI, America is a global leader. It is American companies that lead the world in AI innovation. It is America that can catalyze global action and build global consensus in a way that no other country can.”[22] She did not mention the sustained public investment in science and technology or the US national innovation system, in particular its universities. For the former vice president, the US was an AI leader thanks to its tech companies. Her statement can be read as a sign of consolidation of Big Tech’s sociotechnical imaginary.
The Science and Technology Studies (STS) professor Sheila Jasanoff has defined sociotechnical imaginaries as “collectively held, institutionally stabilized, and publicly performed visions of desirable futures, animated by shared understandings of forms of social life and social order attainable through, and supportive of, advances in science and technology.”[23] While they become an “imaginary” once they are disseminated and adopted widely — even if there can be multiple imaginaries in tension at the same time — would-be imaginaries can be first introduced, articulated, and propagated by specific actors, ranging from states to corporations.
It could be therefore argued that US Big Tech companies like Microsoft, Google, Amazon, Meta and even other players who depend on their funding and/or clouds like OpenAI or Palantir are decisively trying to institutionalize a new chapter within the history of socio-technical imaginaries that link science and technology to growth and progress. This chapter is one in which their technologies are thus deemed essential not only for fulfilling that desire, but also for preventing the obverse, which they explicitly and systematically represent as China’s technological development, distracting the conversation from the fact that their own AI is developed and used for the same antidemocratic purposes. As this imaginary becomes the dominant one, they further entrench their power and expand their income and wealth concentration.[24]
Trump’s administration revoked Biden’s AI regulation on January 23, 2025, in an order entitled “Removing Barriers to American Leadership in AI.” While it ultimately proved harmless to Big Tech power, Biden’s executive order at least placed some limits on their business by restricting certain AI uses and represented a compliance cost.
Revoking Biden’s AI regulation was presented as a prerequisite for the US “AI Action Plan” launched by mid-2025. According to the Plan, the “AI race” is supposed to usher in “a new golden age of human flourishing, economic competitiveness, and national security for the American people.” It is hard to see a new golden age unfolding as the plan’s “Build, baby, build!” policy, aimed at expanding and maintaining AI infrastructure, especially the US power grid, will only deepen the planetary crisis.
Equally hard to believe is the promise of a golden age in economic competitiveness. Those advocating for AI as a growth engine rely on investigations that only look at microeconomic studies, such as the effect of introducing generative AI in call centers, disregarding their macroeconomic effect.[25] Meanwhile, economist Daron Acemoglu has projected that macroeconomic productivity gains from AI adoption will be less than 0.53 percent over the next decade.[26]
It is far more likely that the US will instead grow through the global appropriation of intellectual rents, capturing value from the rest of the world as they capture data and knowledge that is co-created globally.[27] The other source of US growth will also be reliant on rent, but one comparable to land or urban rent paid by those storing and processing their digital solutions, including AI models, on Big Tech clouds. Needless to say, it is highly unlikely that this value capture will result in better working conditions and salaries in the US, since this hasn’t been the case since the 1970s.[28] Most likely, this scenario will further benefit the minority of executives, CEOs, and a few others that manage Big Tech stocks.
The US AI Action Plan reaffirms the government’s alliance with US tech giants — a relationship some thought was weakening when Musk distanced himself from the administration. But behind the scenes of the souring in Trump-Musk’s relationship, the ties with Big Tech have only grown stronger. On top of offering them an expanded power grid, streamlined permits to install datacenters, and an international push for adoption of their models and solutions, the Plan includes deregulation of AI and the revision of all Federal Trade Commission investigations launched during Biden’s term to prioritize AI innovation over any form of regulation, once again buying into Big Tech-crafted narratives as to why they should not be regulated.
A commitment to invest public funds in AI science and technology in close partnership with the private sector, especially cloud giants, also stands out. The AI Action Plan suggests forming agreements with US cloud giants to provide the research community with greater access to digital infrastructure as well as models, data, and software as a service. This initiative falls under the NAIRR pilot, following recommendations from the 2023 report mentioned above.
The US Action Plan has essentially adopted NAIRR recommendations while going even further, as it encourages researchers to make more datasets publicly available, while remaining silent on the fact that Big Tech companies secretly hoard the largest and most diverse datasets. By disclosing more datasets, researchers will be inadvertently favoring those with the most advanced algorithms and capabilities, who will in turn be able to make use of these to further improve their models and develop new ones. These are, of course, cloud giants and their satellite start-ups.[29]
From a geopolitical standpoint, it is particularly telling that the AI Action Plan is not only aimed at fostering AI development and adoption within the USA; it explicitly states that to, win the AI race, the US “must also drive adoption of American AI systems, computing hardware, and standards throughout the world.” It must thus be read as part of the US strategy for rebuilding its hegemony and governance — in alliance with and relying on its Big Tech — over other sovereign territories.
The Plan assigns roles in what would further cement a US-led AI international division of labor: major US tech firms and allied start-ups handle AI production, while the rest of the “allied world” adopts their technologies. It has never been made more explicit that the US government does not intend each nation to democratically decide its (technological) future, which should instead be determined by a few US corporations.
Overall, the turn towards corporate diplomacy signaled by the US National Security Strategy is already anticipated in the AI Action Plan since it emphasizes not only the need to reinforce the innovation prowess of American businesses, but also to expand the US commercial ties with its allies and even non-allied countries such as China on that basis. This is seen as a way to “remain the world’s strongest, richest, most powerful, and most successful country for decades to come”.[30]
If the AI Action Plan is successful, we will be interacting with ever more US-based generative AI models when it comes to customer service, translations, writing software, and marketing campaigns, and even with tasks previously performed by artists, journalists, software developers, and scholars. AI has also been replacing decision-making in warfare. The centrality of the cloud in the AI Action Plan is particularly stressed in its call to “aggressively adopt AI” within the US armed forces. The Plan openly acknowledges the US Department of War’s dependence on cloud giants, stating that the DoD should prioritize “agreements with cloud service providers, operators of computing infrastructure, and other relevant private sector entities.” The close ties between the US DoD, now renamed the US Department of War, and Big Tech, are among the oldest and strongest within the US administration.
Between 2008 and 2021, Microsoft was paid USD 3.2 billion from contracts with the US DoD. In comparison, Amazon received USD 50 million in the same period.[31] Nonetheless, the centrality of Amazon for the DoD, in particular for the provision of data processing services, has expanded recently. Among other services, the US Air Force’s satellite control system runs on AWS. Even if it runs on an isolated service called AWS GovCloud (US) that complies with all the US security requirements, the technological dependency remains.
On the cloud, services are sold as black boxes so that organizations do not get real access to the technology.[32] Hence, as the US DoD privileged the cloud to use AI and other digital tools, it significantly deprived itself from learning by doing, using, and interacting with the technology. If the military does not understand how the technology on which it depends works, its chances to autonomously define a strategy shrink and the dangers associated with using AI for military purposes can only expand. The result has been an asymmetric relation that already favored corporations during Biden Administration.[33]
In 2023, under the auspices of the accelerated adoption of generative AI, both the head of the CIA and the director of the FBI emphasized, once again, the threat of being outpaced by China.[34] This is why, according to the FBI director, the FBI was “proactively aligning resources to engage with the intelligence community and our private sector partners to better understand the technology and its downstream impacts.” This is exactly what Trump has done across the board for defense technology since the outset of his second term in office.
According to USAspending.gov data for the 2025 fiscal year (ending in September 2025), the current total value of Microsoft’s 153 active contracts was over USD 1.5 billion.[35] AWS had 80 active contracts with the US DoD for a current total of over USD 573 million. In comparison, the current total value of the US DoD contracts with its most known Defense Tech suppliers, Palantir and Anduril, were USD 1 billion (53 contracts) and USD 515 million (83 contracts), respectively. Google meanwhile only had 11 active contracts, totaling USD 18 million. This is why the USD 200 million contracts signed in July 2025 were seen as an extraordinary development for the company, which in 2025 changed its policies to openly selling defense technology.
Big Tech companies are not like other military contractors, or even defense tech companies, insofar as they develop military and commercial solutions side by side. Indeed, Amazon, Microsoft, and Google offer a diversity of tailored military services on their clouds, but the military industry is only one of their many cloud and AI application verticals. Business-wise, military applications operate just like other industries. If anything, they are costlier because cloud providers need to comply with all the security requirements.
This is why, as the US military becomes locked into the cloud, compared to traditional military suppliers and defense tech companies, contracts with the DoD are loose change for Big Tech. In 2024 alone their respective total revenues were USD 281.7 billion for Microsoft, USD 638 billion for Amazon, and USD 350 billion for Alphabet (Google’s parent company). Unlike companies such as Palantir or Anduril, making money is certainly not the main driver behind cloud giants’ contracts with the DoD and more broadly with the US government. As an example, Google has recently agreed on a 71 percent temporary discount on technology sold as a service to the US government.[36]
Amazon, Microsoft, and Google do not depend on US government contracts to churn a profit, though they can and will use the government’s increasing dependence on their technologies to negotiate more favorable regulation, lobby for US support against other countries’ regulations, and encourage the promotion of their AI ecosystem. These (geo)political motives are fundamental for their sustained dominance in the long run and can be seen as prime motives driving cloud giants to openly align with the US DoD and with the US government more broadly.
The unprecedented agreement sealed between Big Tech and the current Trump administration is not only a change in degree but in content, form, and context.
At the level of content, the current Trump administration is seeking to plan the development of AI and other digital technologies side by side with — while aiming to establish supremacy within — Big Tech.
While previous administrations promoted industrial policies that directly responded to Big Tech requests, the current US administration is going one step further and strategically investing, acquiring equity in companies that occupy critical segments of digital technologies’ value chains in which the US depends on foreign actors, while reinforcing the US global dominance in other critical segments of these chains.
The government’s decision to authorize Nvidia to sell H200 AI chips in China — even if these are not the company’s cutting-edge chips — should be read through the lens of the US government more actively planning the global economy in alliance with US Big Tech. China was already developing frontier models using Nvidia GPUs smuggled from other countries, while local companies were making progress in chip development.
The US government thus acknowledged that China is a huge AI market and that leaving it underserved in AI chips risked giving more leeway to Chinese players; a leverage that one day may eat out part of the US business in third countries. It thus authorized Nvidia to sell AI chips, yet not its best technology. In exchange Nvidia will have to pay 25 percent in fees to the US government on Chinese sales[37] — a sign of how the US government is not willing to simply leave Big Tech running the show.
China also dominates the mining and, especially, separation and refining of rare earths elements. In April 2025, the Chinese government introduced export controls on seven heavy rare earth elements. The list was expanded to include five additional elements in November 2025. AI and broadly digital technologies are among the affected industries, since rare earths are used in semiconductors’ manufacturing and as chip components that process every digital service on the cloud and on premise. This affects every digital technology, because they are all interconnected and they all require infrastructure to run which requires semiconductors, and because most use the cloud, which is pay as you go, cloud giants adjust the price of compute. To reduce such dependence, the US government has made equity investments in companies extracting critical materials. For instance, it committed to invest USD 35.6 million into Upper Kobuk Mineral Projects (UKMP), a joint venture owned by Trilogy Metals and the Australian mining firm South.[38] The investment is aimed at the development of critical mineral resources in Alaska’s Ambler mining district.
On top of these smaller commitments, the US government acquired a ten percent equity investment in Intel, to assure that CPUs, another core segment of digital value chains, remained under the control of US companies. Sam Altman’s proposal to give the US government a five percent stake in OpenAI further exemplify corporate-state entanglements.[39] It also seeks to respond, albeit through an extremely small stake, to growing societal discomfort with AI-driven wealth concentration.
These decisions better fit China’s state planning toolbox than to the US’ allegedly liberal ruling.
In terms of form, corporate and political powers in the US no longer hide their close connections. On the contrary, they celebrate and share them publicly, while openly defending one another. By supporting Trump like never before, Big Tech in exchange received the overt support of the US government precisely when they needed it the most, as lawsuits against Big Tech opened during Biden’s administration were brought to an end and governments around the world were advancing targeted regulations.
In August 2025, Trump posted on his social media platform Truth Social that he stood up to defend American tech companies against any other government’s attempt to levy taxes or regulate digital markets and services. The post stated that he would increase tariffs on those countries’ exports and institute export restrictions. Trump reaffirmed this warning in late June 2026, when he threatened to impose 100 percent tariffs if the European Union agreed on a common digital tax.[40]
Meanwhile, a federal judge ruled that Meta’s acquisitions of Instagram and WhatsApp did not violate US antitrust law. Google was favored too after initially losing a landmark case against the United States District Court for the district of Columbia, initiated during the Biden Administration. Google had been found to be a monopolist that had unlawfully acted to maintain such a position in general search engines. The requested remedies decided by the end of 2025, which did not include any divestment or wide opening of its private datasets, are relatively lenient for a company the size and power of Google.[41]
The systematic evidence of the government’s support allows US tech giants to soften their previous rhetoric portraying China as a threat. Even if Chinese companies remain their main rivals, what prevails for Big Tech — yet not so much for still unprofitable AI start-ups that offer a single product that China also develops — is that China is a vast market and a source of knowledge and data that they are eager to tap into further.
Multiple factors have also favored the strengthening of the US’s corporate and political ruling coalition during Trump’s second term. Contextual motivations for US Big Tech leaders to strengthen ties with Trump can be summed up as the structural change in the political tide and geopolitics.
In its different expressions, the far right has moved from the fringe to the center of the political arena globally. They are being socially normalized as a political actor that is here to stay. By controlling social media and internet searches, companies like Meta, X, and Google gauge and simultaneously influence the political mood. They are firsthand witnesses as well as responsible for the propagation of far-right ideas.
Concerning geopolitics, the culture surrounding the use of digital technologies for warfare shifted with Russia’s invasion of Ukraine and Israel’s genocide in Gaza. In line with a time of open conflict, the legitimacy of the US ruling coalition goes as far as Big Tech’s warfare involvement.
By 2017, Google had sizeable contracts with the Pentagon, secretly kept under Project Maven. Made public by its employees, Google was forced to withdraw from the project in 2018 to avoid further employee criticism and protect the reputation it had built around a narrative that included the motto “don’t be evil”. Yet by 2025 Google was publicly extolling its contracts with the US DoD.
The pact between Trump and Big Tech and its legitimacy have been favored by the explosive adoption of AI following the release of ChatGPT. The AI boom, a boom that included DeepSeek’s release of its competitive models as open systems in January 2025, gave another convincing reason for acting together. The costs borne by Big Tech — in terms of global reputation and concessions of power to successive US governments — seemed less significant than what they stood to win: US government backing at a moment when concerns about the role of Big Tech were growing while parliaments across the globe discussed AI regulation and the European Union was moving to implement its AI Act.
The coalition scored a victory with the European Commission’s decision to postpone the enforcement of the AI Act. AI had become globally adopted, and its promises widely accepted, putting pressure on the European Commission on top of direct trade negotiations with the US. In summary, the coalition reinforced its ideological victory in a context that favors the far right as well as AI-solutionist narratives. The US government’s alliance with Big Tech, ruling through their technologies, and wielding the threat of weaponizing AI-generated interdependence, could only be materialized in this scenario.
The concept of “weaponized interdependence,” introduced by the political scientists Henry Farrell and Abraham Newman, describes how asymmetric global networks controlled by multinational corporations can be exploited by their home states to advance political objectives.[42] Because AI is being applied across industries and value chains, it offers the most extensive potential for weaponizing interdependence.
Farrell and Newman identified two mechanisms of state-driven weaponized interdependence: “chokepoint” and “panopticon” effects. Chokepoints refer to bottlenecks that occur when companies from a single country dominate a critical segment of a value chain. A clear example is the US’s dominance in the frontier design of semiconductors or China’s control over the rare earth elements value chain. Chokepoints often arise from the intellectual monopolization of essential knowledge required for production or from the control over a scarce natural resource.
The panopticon effect, on the other hand, refers to having exclusive access to the structure, flow, and content of a network. Essentially, it refers to the ability to monitor all information created and exchanged within a network, understanding who does what and how. The cloud enables Amazon, Microsoft, and Google to enjoy such a panoptic effect.[43] US companies also hold a collective chokepoint in frontier AI models with only a few exceptions like Mistral, which anyway depends on US capital, or DeepSeek, which anyway offers its models on Amazon, Microsoft, and Google clouds.[44]
Against this backdrop, there are plenty of opportunities for the US government to weaponize AI-related interdependence. Microsoft complied with the US executive order against the International Criminal Court (ICC) based in The Netherlands, when it suspended the email account of the ICC prosecutor that was investigating Israel’s war crimes.[45] The US Department of Commerce’s temporary ban on foreign use of selected Anthropic models can equally be read through this lens. It can also be interpreted as another example of the US government flexing its muscle within its ruling coalition with US Big Tech.
More broadly, there is also a major opportunity for an even more extended use of AI for surveillance and warfare. This corporate and political coalition provides a privileged space to military agreements. In exchange, Big Tech gets more than simply economic compensation. Their loyalty is rewarded through a fierce state defense of their interests against foreign governments in their attempts not only to regulate US tech companies but also expand digital sovereignty.
For this alliance among frenemies to stand, there is a cost for both the US government and Big Tech. For Big Tech, this is the acceptance of adverse resolutions in antitrust cases, albeit when such resolutions do not significantly hurt their power. Amazon will pay a USD 1 billion civic penalty and refund USD 1.5 billion to Prime subscribers in a settlement over claims from the US FTC that showed that it had deceived consumers into signing up into the service.[46] Meanwhile, Google has accepted its loss in what was probably the US landmark antitrust case since the lawsuit against Microsoft over 20 years ago.[47] Google also agreed to pay USD 24.5 million to settle a 2021 lawsuit against YouTube that claimed that it unfairly blocked Trump’s account after the US Capitol attacks on January 6, 2021.[48]
In turn, as I have shown throughout this report, the US government accepts this state of affairs in order to expand its dependence on Big Tech, most notably on cloud giants — as these companies’ technologies remain black boxes. This weakness opens space for a future scenario in which the US government becomes simply too dependent on Big Tech and subordinates itself to their ruling power. Its growing reliance on a few corporations for providing technologies that become central to every corner of the public administration is, slowly but steadily, weakening state political power against corporate power.
When speaking of power relations, Giovanni Arrighi draws on Talcott Parsons’ distinction between distributive power as a zero-sum game, and collective power as a positive-sum-game, to argue that the latter is exercised when “cooperation among distinct agencies increases their power over third parties or over nature.” [49] In the case of AI, however, this mutual dependency is unbalanced as it can be said that the US is governed by a few private sector giants who do not depend on the former to make their business, especially since the rest of the world’s governments are equally becoming dependent on their technologies, which hinders their chances of effectively regulating Big Tech and developing decoupled alternatives. Compared to the Cold War period, governed interdependence in the development of frontier technology remains in the hands of US actors, but the roles have been inverted.
It is hard to see how future US administrations could replace their dependence on Amazon, Microsoft, and Google or even regulate these corporations, once the whole public sector runs on their clouds. It is even harder once we compute how previous administrations, while promoting and aligning with Big Tech, simultaneously attempted but ultimately failed to regulate them. When referring to Big Tech in the 2020 US Congress hearing, Congressman David Cicilline concluded: “Simply put, they have too much power.” This was followed by a stronger antitrust foothold against Big Tech led by the US FTC during Biden’s administration. However, despite these attempts, these giants have only become more powerful since then.
In light of the above, the US government equity in key companies like Intel — and eventually OpenAI — and its share of Nvidia exports to China could also be interpreted as a way to position itself as an economic power that can sit and negotiate with other Big Tech with more leverage. Energy and its infrastructure are other assets in the hands of the US government. OpenAI has urged the US adminstration for more energy to power AI datacenters in a move that represents the interests of cloud giants as much as its own and that, once again, uses the threat of China as the main reason why its entreaty should be heard.[50] Tech giants depend on electricity and drinking water to keep the AI engine running. While the US government could use this weakness and negotiate the conditions for its energy investment with Big Tech, the AI Action Plan already included the government’s decision to expand and maintain the US power grid, signaling that official will comply with Big Tech’s demand. The US National Security Strategy equally stressed this point, which was further backed by the Department of Energy’s “Manhattan Project 2.0,” which supports expanding nuclear energy in the service of the AI race.[51]
Meanwhile, cloud giants and Nvidia continue to announce investments for datacenters around the world, thereby reducing their reliance on the US government’s political support. US corporate giants are also putting in additional efforts to present themselves as allies to other governments in their attempts to expand digital sovereignty. They even use phrases like “sovereign cloud” to describe how their expansive infrastructure investments around the globe make them even more indispensable.[52]
Overall, the coalition between the US government and Big Tech remains a long-lasting arm-wrestle between frenemies. Today’s more stable, though still weak, equilibrium in their distribution of power is advantageous to both corporate and political leaders amid increased sovereignty concerns in the rest of the world. Either the coalition stands, or it breaks because one of its parts manages to structurally best the other in such a way that it no longer needs to make the concessions negotiated within it. The expected scenario is a more unilateral, undemocratic control of US powers over critical technologies which will reinforce the US’s — corporate and/or political — world dominance, exercised not only over other economies but also impacting most US citizens and organizations.
Alternatives are urgently needed to prevent the further extraction of nature, data, and knowledge from around the world while there is an entrenchment of an AI divide in which the majority of people use, and through their prompts contribute to the further entrenchment of, technology as a black box controlled by a minority.[53] Given the centrality of digital technologies for the public sector, opinion forming, and decision-making, a future without alternatives would most likely bring less democracy, more inequality, and a tech-driven worsening of the ecological crisis.
Governments cannot rely on the expectation of competition from Chinese actors as a market solution to concentrated power. The interplay between the Chinese government and its Big Tech does not significantly differ from the US story. A world more evenly divided between technological (and thus economic and political) dependence on US or Chinese powers would still represent one of mass subordination.[54]
Yet this is not a limitation for building and expanding degrees of digital sovereignty in the rest of the world if we define it not as a strictly territorial or national question but as a matter of governance. We must turn attention to who decides and how what technologies are developed and used, whose technologies, what data is retrieved, for what purposes, and where and to what extent a region can cope with datacenters. All these are matters that should be defined democratically, by empowering workers and citizens to govern the technology they regularly use.
Concretely, what is needed is an independent, public-led ecosystem that could be built as a collective effort of those who stand against antidemocratic ruling powers. Such a solution must be tailormade to the size of the collaboration and could also include US states or cities, such as the city of New York, that want to build an alternative to Big Tech and its authoritarian and predatory systems. It should be integrated with publicly governed infrastructure, platforms, and models. In other words, to replace Big Tech’s predatory domination of tech ecosystems, the focus should not only be on digital infrastructure but also on replacing their clouds as the economic space for development, exchange, and deployment of digital technologies.
Foundational technologies including infrastructure for building this alternative should be shared openly and should be developed by prioritizing their potential contribution to solving key social and ecological problems and empowering the majority of people. This alternative should also be governed by new democratic institutions with representation from unions, independent civil society organizations, and experts not only on the actual development of digital solutions but also on their multiple implications.
To this end, a starting point could be to transform existing state-owned telecommunication and software companies into a (or a few, one per region) consortium of interconnected and democratically governed digital companies. This collaboration of existing state-owned companies — extending from Brazil and Uruguay to Vietnam and even cases in Europe — should be mandated to:
While open sourcing and building digital commons should be a primary aim of this public-led digital stack, the single promotion of these shared or open proprietary regimes will not suffice. The adoption and promotion of open-source solutions could have offered an alternative to intellectual monopolies in the 1990s, but even back then adoption was dwarfed when compared to Windows operating system and the Office package. Software was installed in our devices from the outset, enabling a system in which open source — and piracy — could effectively replace proprietary technologies. Since then, two major transformations have occurred.
One has been Big Tech’s subsumption of open source into its broader predatory ecosystem, which renders impossible to imagine a sovereign solution based only on open source. The cloud is the other major transformation. It has enabled such an entanglement of digital technologies that it seems impossible to build independent, standalone solutions that do not eventually depend on the cloud either to rent computing services or run and eventually sell resulting solutions. The advent of AI has rendered this option even less feasible. The degree of centralization of the whole digital network that we witness today inside the clouds of Amazon, Microsoft, and Google, coupled with the massively increased world dependence on digital technologies, cannot be solved by relying exclusively on the decentralized ideas of software and AI engineers.
It would in any case not be enough to focus on investment in and reliance upon open source as the major alternative, as the open-source community, even if more democratic than Big Tech, remains a small fraction of our society. If we truly want to democratize digital technologies, its development cannot be exclusively entrusted to an expert group as they are not prepared to identify their technologies’ social, ecological, political, ethical, and economic implications. Even after considering different types of expertise, the kind of public alternatives to Big Tech that should be built remains a political decision. As such, it cannot exclusively be commissioned to experts.
Governments should equally use public procurement to ensure the feasibility and steer the development of digital technology towards prioritizing public value. Public procurement contributes to rendering this solution feasible. It will be further accomplishable if such an alternative ecosystem is designed, at a preliminary stage, for serving a specific area within the public administration. This will make it not only technically feasible but also more affordable.
In regard to the necessary investment, governments should agree on levying specific taxes to partially compensate for these giants’ extraction of intangibles (data and knowledge) and nature. A tax on these twin extractivist practices, based on the complementarity between data (and knowledge) extraction and the use of electricity and drinking water in their datacenters will not in itself terminate Big Tech’s predation. However, it will contribute to redistributing associated gains. Such a tax should ideally be coordinated and levied together by likeminded governments sharing the goal of prioritizing people and the planet. It should also be complemented with moratoriums on the construction of new private hyperscale datacenters.
Finally, expanding digital sovereignty by building an alternative to Big Tech, is not only a matter of redistributing the economic gains associated with the collective coproduction of digital technologies. It is also a matter of redirecting digital technologies so that they are developed for the benefit of communities while remaining within planetary boundaries. Concretizing this aim requires new public institutions that are capable of democratically governing technological development. They must prevent a public-led digital ecosystem from replicating the priorities and ways of producing and appropriating knowledge that fit the needs of concentrated corporate and political power.
A new digital R&D agency for the democratic setting of the digital research agenda should be created. It should promote international collaboration, solidarity, and interdisciplinarity for mapping and advancing an R&D agenda that should emerge from combining creativity with a focus on addressing the needs and desires of the communities.
Creating such an agency could be an opportunity to design an institution that exemplifies the democratization of science and technology not only by broadening access to their outputs, but more importantly to their production and the setting of R&D agendas understood as inherently political. In doing so, the construction of a public-led digital ecosystem opens a door for reimagining public governance and accountability in ways that expand democracy and rebuild our communities within planetary boundaries.
[1] See Cecilia Rikap, The Rulers: Corporate Power in the Age of AI and the Cloud, Verso, 2026.
[2] Cecilia Rikap, Teoríade la Dependencia Digital. Soberanía y Desarrollo en el Capitalismo del Siglo XXI, Caja Negra Editores: 2026, chapter 5.
[3] Cecilia Rikap, “The US National Security State and Big Tech: Frenemy Relations and Innovation Planning in Turbulent Times.,” Review of Keynesian Economics, 2022, vol. 12, no. 3, pp. 348–64. Available here.
[4] Linda Weiss, America Inc.?: Innovation and Enterprise in the National Security State, Cornell University Press: 2014.
[5] David Noble, Forces of Production: A Social History of Industrial Automation, Transaction Publishers, 2011, chapter 1.
[6] Fred Block, “Swimming against the Current: The Rise of a Hidden Developmental State in the United States,” Politics & Society, 2008, vol. 36, no. 2, pp. 169–206; Mariana Mazzucato, The Entrepreneurial State: Debunking Public vs. Private Sector Myths, Anthem Press, 2013; Weiss, America Inc.?
[7] Graham Dutfield, That High Design of Purest Gold: A Critical History of the Pharmaceutical Industry 1880–2020, World Scientific, 2020; Chris Miller, Chip War: The Fight for the World’s Most Critical Technology, Simon & Schuster: 2022.
[8] Tim Wu, “Tech Dominance and the Policeman at the Elbow,” Columbia Public Law Research Paper, 2019, nos. 14–623. Available here.
[9] Cecilia Rikap, Capitalism, Power and Innovation. Intellectual Monopoly Capitalism Uncovered, Routledge, 2021.
[10] Benjamin Coriat and Fabienne Orsi, “Establishing a New Intellectual Property Rights Regime in the United States: Origins, Content and Problems,” Research Policy, 2002, vol. 31, nos. 8–9, pp. 1491–507; Rochelle Dreyfuss and Susy Frankel, “From Incentive to Commodity to Asset: How International Law Is Reconceptualizing Intellectual Property,” Michigan Journal of International Law, 2014, vol. 36, no. 4, pp. 557–602.
[11] Mark Glick, “Antitrust and Economic History: The Historic Failure of the Chicago School of Antitrust,” The Antitrust Bulletin, 2019, vol. 64, no. 3, pp. 295–340.
[12] Ronen Palan and Richard Phillips, “Arbitrage Power and the Disappearing Financialized Firm,” Finance and Society, 2022, vol. 8, no. 1, pp. 22–41; Richard Phillips et al., “Group Subsidiaries, Tax Minimization and Offshore Financial Centres: Mapping Organizational Structures to Establish the ‘in-Betweener’ Advantage,” Journal of International Business Policy, 2021, vol. 4, no. 2, pp. 286–307.
[13] Xiaolan Fu and Pervez Ghauri, “Trade in Intangibles and the Global Trade Imbalance,” The World Economy, pre-print, Wiley Online Library, 2020. Available here.
[14] Johana Bhuiyan, “Documents offer rare insight on Ice’s close relationship with Palantir”, The Guardian, September 22, 2025. Available here; Michael Kwet, “The Microsoft Police State: Mass Surveillance, Facial Recognition, and the Azure Cloud,” The Intercept, July 14, 2020. Available here.
[15] “Cuomo Questions Why School Buildings Still Exist”, Answer Sheet, The Washington Post, May 6, 2020. Available here.
[16] National Security Commission on Artificial Intelligence, Final Report, Washington DC, 2021, p. 7.
[17] National Security Commission on Artificial Intelligence, Final Report.
[18] National Security Commission on Artificial Intelligence, Final Report, p. 25.
[19] National Security Commission on Artificial Intelligence, Final Report, p. 12.
[20] “Strengthening and Democratizing the U.S. Artificial Intelligence Innovation Ecosystem: An Implementation Plan for a National Artificial Intelligence Research Resource,” National Artificial Intelligence Research Resource Task Force, 2023, p. 44.
[21] “Microsoft CEO on A.I. race: ‘Not a given’ that Alphabet or Microsoft are the only two games in town”, CNBC, May 16, 2023. Available here.
[22] John Thornhill, “Step aside world, the US wants to write the AI rules”, The Financial Times, 02/11/2023. Available here.
[23] Sheila Jasanoff, “Future Imperfect: Science, Technology, and the Imaginations of Modernity,” Dreamscapes of Modernity: Sociotechnical Imaginaries and the Fabrication of Power, University of Chicago Press: 2015, 4.
[24] On corporate innovation systems and science and technology predation see Cecilia Rikap, “Intellectual Monopolies as a New Pattern of Innovation and Technological Regime,” Industrial and Corporate Change, 2023, vol. 33, no. 5, pp. 1037–62.
[25] See e.g., Erik Brynjolfsson et al., “Generative AI at Work,” National Bureau of Economic Research, 2023. Available here.
[26] Daron Acemoglu, “The Simple Macroeconomics of AI”, National Bureau of Economic Research, Working Paper 32487, 2024. Available here.
[27] See Cecilia Rikap, “Dynamics of Corporate Governance Beyond Ownership in AI”, Common Wealth, 2024. Available here.
[28] Giovanni Dosi and Maria Enrica Virgillito, “Whither the Evolution of the Contemporary Social Fabric? New Technologies and Old Socio-Economic Trends,” International Labour Review, 2019, vol. 158, no. 4, pp. 593–625.
[29] Cecilia Rikap, “Dynamics of Corporate Governance Beyond Ownership in AI”, Common Wealth, 2024. Available here.
[30] See “National Security Strategy of the United States of America”, November 2025. Available here.
[31] Andrea Coveri, Claudio Cozza, and Dario Guarascio, “Blurring Boundaries: An Analysis of the Digital Platforms-Military Nexus”, Review of Political Economy 37 (4), 2025, pp. 1632–63.
[32] Cecilia Rikap and Bengt-Åke Lundvall, The Digital Innovation Race: Conceptualizing the Emerging New World Order, Palgrave Macmillan: 2021.
[33] See Cecilia Rikap, “The US National Security State and Big Tech: frenemy relations and innovation planning in turbulent times”, Review of Keynesian Economics, 12:3, pp. 348–364.
[34] Jack Aldane, “‘We have to do it faster and better than they do’: CIA chief sets out era of US-China ‘strategic competition’ in AI and beyond”, Global Government Forum, July 17, 2023. Available here; AJ Vicens, “Top FBI officials warn of ‘unparalleled’ threat from China and AI”, Cyberscoop, July 26, 2023. Available here.
[35] Microsoft Corporation Federal Award Recipient Profile”, USAspending.gov, 2025. Available here.
[36] Joe Miller, Rafe Uddin, and Stephen Morris, “Google to agree cloud discount as US government squeezes Big Tech”, The Financial Times, July 11, 2025. Available here.
[37] Stephen Nellis, Karen Freifeld, and Michael Martina “US to allow Nvidia H200 chip shipments to China, Trump says”, Reuters, December 9, 2025. Available here.
[38] “Trump administration pivots to buying stakes in critical sectors”, Reuters, January 26, 2026. Available here.
[39] Cristina Criddle, George Hammond, “OpenAI proposes handing Trump administration 5% stake”, The FT, July 2, 2026. Available here.
[40] Peter Wells, James Politi, “Donald Trump warns of 100% tariff on countries implementing digital services tax”, The Financial Times, June 26, 2026. Available here.
[41] Per requested remedies, Google is barred from entering any exclusivity agreement related to the distribution of Google Search, Chrome, Google Assistant, and the Gemini app, its source code and algorithms will be audited by an independent committee that will operate under a confidentiality agreement and Google will share a small sample of its search index and user-interaction data, though not ads data, with Qualified Competitors. This is just a partial opening of a subset of data to a few other intellectual monopolies like Microsoft. The emergence of generative AI chatbots that are used as search engines was presented as a reason for betting on the long term termination of Google’s monopolist position.
[42] Henry Farrell and Abraham Newman, “Weaponized Interdependence: How Global Economic Networks Shape State Coercion,” International Security, 2019, vol. 44, no. 1, pp. 42–79. Guillaume Beaumier and Madison Cartwright, “Cross-Network Weaponization in the Semiconductor Supply Chain,” International Studies Quarterly, 2024, vol. 68, no. 1. Available here; Lars Gjesvik, “Private Infrastructure in Weaponized Interdependence,” Review of International Political Economy, 2023, vol. 30, no. 2, pp. 722–46.
[43] Cecilia Rikap, The Rulers. Corporate Power in the Age of AI and the Cloud, Verso Books, 2026.
[44] Cecilia Rikap, “Heads I Win, Tails I Win: Why US Cloud Giants Benefit from DeepSeek and Other Chinese Companies’ AI Strategies”, Common Wealth, June 2025. Available here.
[45] Adam Satariano, Jeanna Smialek, “Europe’s Growing Fear: How Trump Might Use US Tech Dominance Against It”, The New York Times, June 20, 2025. Available here.
[46] See “FTC Secures Historic $2.5 Billion Settlement Against Amazon”, Federal Trade Commission, September 25, 2025. Available here.
[47] See “U.S. V. Microsoft: Proposed Findings Of Fact”, Antitrust division, U.S. Department of Justice. Available here.
[48] Kate Conger, “YouTube Settles Trump Lawsuit Over Account Suspension for $24.5 Million”, The New York Times, September 29, 2025. Available here.
[49] Giovanni Arrighi, Adam Smith in Beijing: Lineages of the 21st Century, Verso Books, 2009, p. 150.
[50] “Seizing the AI opportunity”, OpenAI, October 27, 2025. Available here.
[51] “‘Manhattan Project 2.0’: Bradburne Lays Out EM Vision to Support Nuclear Renaissance, AI Race”, Office of Environmental Management, U.S. Department of Energy, September 24, 2025. Available here.
[52] See “Microsoft Sovereign Cloud”, Microsoft. Available here.
[53] I have written extensively about this, among others, in Rikap, The Rulers. See also Rikap, Teoría de la Dependencia Digital.
[54] Rikap and Lundvall, The Digital Innovation Race.