The study, published in the journal Global Public Policy and Governance, was conducted by Shaleen Khanal of the Centre for Trusted Internet and Community at the National University of Singapore and Hongzhou Zhang of the S. Rajaratnam School of International Studies at Nanyang Technological University. Motivated by the theoretical framing of power-knowledge, the idea, associated with Michel Foucault, that the production of knowledge and the exercise of power are deeply intertwined, the authors set out to measure whether the concentration of innovation inputs in corporate hands is translating into a concentration of scientific output and agenda-setting power. The stakes, they argue, extend far beyond academic sociology: whoever controls the production of knowledge in critical technologies ultimately shapes the direction, safety and governance of technologies that will define economic and geopolitical power for decades.

To move beyond anecdote, the researchers assembled an unusually large evidence base. Using OpenAlex, an open-access index of scholarly works, they identified 3.28 million publications on critical and emerging technologies published between 2001 and 2024, spanning fields such as artificial intelligence, quantum computing and biotechnology. They then documented the share of these outputs authored or co-authored by scientists affiliated with large technology firms, using author-level affiliation data rather than publication-level labels alone, a methodological choice that allowed them to capture researchers who move between universities and industry. Within this corpus, they drilled down into artificial intelligence, the technology where concerns about corporate capture have been loudest, to assess both the extent and the character of Big Tech’s influence.

The headline result is mixed evidence. Across the full landscape of critical and emerging technologies, Big Tech’s footprint is significant but far from dominant, contradicting the strongest versions of the epistemic capture thesis. Public research institutions remain central to knowledge production in most critical fields, and corporate publishing varies widely across domains, reflecting the different economics of basic and applied research in each area. In artificial intelligence, however, the picture is different: the study documents an unmistakable and increasing presence of Big Tech-affiliated scientists over the two decades covered by the data, consistent with a broader literature showing that industry’s share of leading AI papers has climbed steeply since the deep learning revolution made massive datasets and computing clusters decisive inputs.

That escalation is unsurprising given the underlying resource asymmetry. Training frontier AI models requires computing infrastructure, specialized hardware and capital budgets that few universities can match. Previous scholarship has described a compute divide in which academic labs are progressively squeezed out of state-of-the-art AI research, and high-profile controversies, including the departure of prominent ethics researchers from major labs, have fed concerns that corporate priorities shape not only what gets researched but what may be published. Yet the new analysis complicates the gloomiest narratives. The authors note that emerging evidence suggests Big Tech’s influence on research has in some respects been overtly exaggerated, and their own data show that corporate-affiliated scientists have not colonized critical technology research wholesale, even as their presence in AI grows year after year.

One of the study’s most striking findings concerns collaboration. Publications co-authored by Big Tech researchers and other knowledge institutions, including universities and public research organizations, are associated with higher research impact than papers produced in isolation. This suggests that the flow of industrial resources, data and engineering expertise into joint projects is not merely extracting value from academia but genuinely raising the scientific quality and visibility of the resulting work. Collaborative papers also show a higher probability of including keywords related to trustworthy AI, indicating that corporate-academic partnerships may, at least in bibliometric terms, be pulling attention toward questions of safety, fairness and accountability rather than away from them. The finding challenges a simple capture story in which industry involvement necessarily narrows the research agenda toward commercially lucrative topics.

At the same time, the authors are careful not to declare the concerns unfounded. The theoretical framework of power-knowledge implies that influence operates through channels that bibliometrics can only partially reveal: the setting of technical standards, the funding of labs, the revolving door of personnel between firms and governments, lobbying networks, and the strategic control of research infrastructures such as cloud computing platforms. Other work by the same research team has documented how Big Tech firms act as super policy entrepreneurs, increasing their power in the policy process around generative AI, and how public-private relations in areas such as defense and security governance are being restructured around corporate technological capabilities. The new bibliometric evidence should therefore be read as calibrating the debate rather than closing it.

The findings carry direct implications for technological sovereignty, a concern that has moved to the center of policy debates in the United States, Europe, China and beyond. Governments increasingly treat critical and emerging technologies as strategic assets, screening foreign investment, restricting exports and funding national research programs precisely because control over knowledge production is understood as control over future power. If a small number of private firms, rather than public institutions, dominate the production of knowledge in these fields, then national strategies aimed at sovereignty must grapple not only with foreign rivals but with domestic corporate concentration. The study’s evidence that Big Tech is most entrenched in AI, arguably the most consequential critical technology, sharpens this dilemma: policymakers who depend on corporate labs for national competitiveness may find that the same dependence erodes public capacity for independent assessment, regulation and basic research.

For universities and research funders, the message is double-edged. The positive association between Big Tech collaboration and research impact creates real incentives to deepen industrial partnerships, and such partnerships have produced celebrated scientific advances, from protein structure prediction to large-scale language models. But the study’s framing also echoes calls from prominent university leaders that academia must reclaim independent capacity in AI research for the public good, ensuring that questions about societal risk, ethics and long-term safety are pursued even when they offer no commercial return. The risk identified in the broader literature is that corporate funding and infrastructure dependency can produce insularity, recency bias and quiet suppression of unfavorable findings, dynamics that citation counts alone will not capture.

Ultimately, the research offers an evidence-based middle position in a debate that has often been driven by rhetoric. Big Tech has not captured the production of scientific knowledge across critical technologies, but its presence in artificial intelligence research is growing steadily and its collaborative ties amplify both its scientific contribution and its agenda-setting influence. The privatization of critical knowledge infrastructure, the authors conclude, remains a live geopolitical risk, one that demands governance responses calibrated to the actual scale and character of corporate power rather than to either panic or complacency. As data, compute and capital continue to concentrate, the question is no longer whether Big Tech matters for frontier science, but how democratic societies can harness its capabilities while preserving the independence of the knowledge systems on which they depend.

Subject of Research: Big Tech's influence on scientific research in critical and emerging technologies such as artificial intelligence

Article Title: Power-knowledge and strategic control: how Big Tech is reshaping research in critical technologies

Article References: Khanal, S., & Zhang, H. (2026). Power-knowledge and strategic control: how Big Tech is reshaping research in critical technologies. Global Public Policy and Governance, 6(2), 199-226. https://doi.org/10.1007/s43508-026-00150-2

Image Credits: AI Generated

DOI: 10.1007/s43508-026-00150-2

Keywords: Big Tech, artificial intelligence, AI research, critical and emerging technologies, power-knowledge, bibliometrics, technological sovereignty, trustworthy AI, epistemic capture, industry-academia collaboration, strategic, control