Skip to content
GLOBALANCE INTELLIGENCEThe economy ahead.
This Week Index 28/100 Sign in
Countries › USA
SwitzerlandGermanyChinaUSA
← All 14 pillars of the USACountries › USA › Pillar 07 of 14
PILLAR 07 · DIGITAL GOODS · USA

Data & AI as Commons

USA status: switches openUSA Module · Digital Goods

An innovation-friendly but patchy data regulatory framework; no federal data protection legislation equivalent to EU standards.

01 · STATUS QUO & FAULT LINE

Where does the USA stand, and where does it break?

STATUS QUO

The United States is the undisputed AI superpower: it is home to the leading foundation models, the majority of the world's computing power, and the deepest venture-capital markets on earth (Stanford AI Index). At the same time, a shared data governance framework is absent. There is no federal data-protection law comparable to the European General Data Protection Regulation. Instead, a patchwork of state-level rules prevails, led by California's privacy legislation, while many states have no comprehensive protection at all. Data and AI are regulated in a manner that is both innovation-friendly and full of gaps. The critical infrastructure — foundation models, data centres, and cloud — is held predominantly by a small number of private corporations with enormous computing power. At the federal level, AI policy shifts with each administration: in 2025, a new administration revoked its predecessor's AI executive order, adopted a deregulatory stance, and moved against state-level AI legislation by executive order. Counterweights include public initiatives such as the National AI Research Resource, which aims to give researchers outside the large corporations access to computing power and data. The underlying tension is between speed and leadership through private concentration on the one hand, and open, public-interest use of data and computing power on the other.

FAULT LINE

In the AI economy, data and computing power are the decisive factors of production. In the United States, they are concentrated in the hands of a small number of private corporations. Whoever controls the foundation models and the compute infrastructure sits at the source of the new value creation and the AI rent. Without a common data governance framework and without public access to computing power, the data commons — generated by everyone — risks becoming the private property of a few. The models trained on that data further reinforce the concentration of power. Patchy, innovation-friendly regulation is accelerating pace and leadership. Questions of liability, safety, authorship and market power remain unresolved. The state-by-state patchwork creates legal uncertainty, and the federal government oscillates between deregulation and the desire for uniform rules. The open question is this: will data and compute remain a privately controlled resource, or will a public-interest framework emerge that secures access, rights and competition without forfeiting innovation leadership?

02 · POLICY OPTIONS

Paths for the USA

A
Federal data protection and data rights★★★☆☆ 3/5

The United States enacts a federal data-protection law for the first time, replacing the patchwork of state-level rules and giving citizens clear rights over their data. This addresses the absence of a coherent data order directly, creates legal certainty, and strengthens individuals' negotiating position vis-à-vis the data corporations. California demonstrates that comprehensive data protection is achievable in the US. The weakness: federal legislation has foundered for years on the question of whether it should pre-empt stricter state-level rules, and the tech lobby resists stringent requirements.

+ Creates legal certainty and strengthens individual data rights
– Has foundered for years on federalism and the tech lobby

Trade-off: Uniform data framework versus state-level standards of protection

Feasibility: The conflict over pre-empting stricter state laws is the decisive issue

B
Public computing power and the AI commons★★★☆☆ 3/5

The state is building publicly accessible computing power and data resources so that research, start-ups and the public are not entirely dependent on a handful of large corporations. This strikes at the heart of the fault line by breaking open the concentration of compute and enabling AI development beyond a small number of players. The National AI Research Resource is a real beginning. The weakness: the scale is enormous. The computing power of the major corporations exceeds public capacity many times over, and funding must be secured on a lasting basis to remain effective.

+ Breaks open the concentration of compute, opens up AI development
– Enormous scale required; sustained funding essential

Trade-off: Open access versus fiscal and technical demands

Feasibility: The NAIRR is a start, but still too small in scale

C
Public participation in the data and AI rent★★☆☆☆ 2/5

The state secures for the public a share of the economic rent arising from collectively generated data — through a data levy, licensing models, or equity stakes — and distributes it as a dividend. This addresses the fact that the data commons is produced by everyone, yet its value is captured privately. Conceptually, it connects with the idea of an AI dividend. The weakness: valuing data, levying a charge, and establishing governance are all highly complex, and in the market-oriented culture of the United States such an intervention is widely regarded as hostile to innovation.

+ Gives the public a share in the collectively generated data rent
– Valuation and collection highly complex; perceived as an intervention

Trade-off: Public-interest participation versus innovation and market concerns

Feasibility: Conceptually compelling, but the hardest to implement in practice

D
Applying competition law to models and compute★★☆☆☆ 2/5

Competition authorities use merger control, interoperability obligations and access to infrastructure to prevent control over foundation models, data centres and key datasets from becoming entrenched among a small number of corporations. This addresses the concentration of power that lies at the core of the fault line. Antitrust proceedings against major tech companies are already under way in the United States. The weakness: cases take years, technology develops faster than the law, and political will oscillates between protecting competition and the desire to promote national AI champions.

+ Addresses the concentration of power over models and compute
– Slow-moving, technology outpaces the law, politically inconsistent

Trade-off: Open competition versus the promotion of national champions

Feasibility: Interoperability and infrastructure access are the most effective levers

03 · TIME HORIZON

Trigger-based phases

Phase 1: Leadership through concentration

Trigger: Control over foundation models, computing power and key datasets becomes further concentrated among a small number of corporations, while federal-level rules remain absent.

The United States retains its AI lead, but data and compute remain under private control. The state-level patchwork grows, and public initiatives such as the NAIRR remain small relative to private computing power.

Phase 2: Regulatory conflict

Trigger: The concentration of power among AI corporations and the legal uncertainty of the patchwork become a political and economic problem.

Calls for federal data protection, public computing power, and competition safeguards grow louder, even as the federal government oscillates between deregulation and a coherent regulatory framework. Individual states press ahead with their own AI legislation.

Phase 3: New data order or private status quo

Trigger: Lasting policy creates a public-interest framework for data and compute, or private control becomes entrenched.

The United States will determine whether data and computing power become an open, broadly accessible resource or remain the private property of a few. On this hinges whether the value generated by AI is shared or concentrated, and whether innovation leadership is preserved.

04 · PILOT PROGRAMMES

Real-world pilot programmes

National AI Research Resource (NAIRR)

Status: Pilot under way (public access to computing power and data)

Backed by the National Science Foundation, the NAIRR gives researchers outside large corporations access to computing power, data, and tools. It is the most concrete attempt to establish AI development as a publicly accessible resource and to mitigate the concentration of compute.

Open project

California's data protection laws (CCPA/CPRA)

Status: In Kraft (strengster Datenschutz eines US-Bundesstaats)

California grants citizens extensive rights over their personal data and functions as the de facto standard across the United States, as many companies apply it nationwide. The law demonstrates that robust data rights are both achievable and enforceable within a market-based US context.

Open project

Colorado AI Act and state-level AI legislation

Status: Enacted, entry into force deferred to 2026 (one of the first comprehensive AI laws passed by a US state)

Colorado is a pioneer in regulating the use of AI in high-stakes decisions, requiring transparency and protection against discrimination. The law serves as a policy testing ground for AI governance whilst also illustrating how regulation diverges in the absence of a federal framework.

Open project

05 · EARLY INDICATORS

What we are watching

Concentration of AI computing power

This metric measures who controls the decisive factor of production in the AI economy. The threshold matters because growing concentration draws value creation and power into ever fewer hands. An inflection point would mean that AI development outside the major corporations becomes barely feasible. For competition and participation, this is the central early indicator.

Today: The majority of global AI compute is held by a small number of private US corporations (Stanford AI Index)
→
Threshold: Continued rising concentration without growing public or open access
Progress towards federal data protection

This metric measures whether the United States arrives at a common framework for data as a resource. The threshold matters because, without a federal structure, legal uncertainty and unequal protection persist, and citizens' negotiating position vis-à-vis the major platform companies remains weak. A tipping point would mark the establishment of a coherent national data framework. Without one, private control will become further entrenched.

Today: No federal data protection law; a state-by-state patchwork, with California as the leading jurisdiction
→
Threshold: Persistent absence of a federal data framework as the patchwork continues to grow
Reach of public computing and data resources

This metric shows whether an AI commons is taking shape. The threshold matters because only publicly accessible computing power at meaningful scale will reduce dependence on large corporations. An upward shift would demonstrate that open AI development remains viable. If public capacity stays marginal, private compute alone will determine the direction of the AI economy.

Today: National AI Research Resource operational as a pilot, small relative to private computing capacity
→
Threshold: Expansion of public compute capacity to a meaningful proportion of private capacity
06 · OPEN QUESTIONS

What we do not yet know

1. Can a country without federal data-protection legislation arrive at a coherent data governance framework before the patchwork of state-level rules becomes a permanent burden?
2. Is it possible to break up the concentration of data and computing power without forfeiting the United States' lead in innovation?
3. Will the data commons generated by everyone become the private property of a handful of corporations, or will a model oriented towards the common good emerge?
← All pillars: overviewNext: 08 · Care, Health & Social Services →

What comes next