Sovereignty Without Infrastructure

Analytical Note No. 11 · Series: Governance Briefs · June 2026 Operationalises: Essay 9 — The Sovereignty Question and Essay 10 — The Infrastructure Question — Beyond Control: Theory of Limits of AI Governance · Русская версия

Contents

The sovereign declaration travels without the material capacity it presupposes.

I. The Declaration and the Substrate

On 10 March 2026, Presidential Decree No. PD-38 defined the Cybersecurity Strategy of the Republic of Uzbekistan for 2026–2030. Among its strategic directions is the development of digital innovations, AI technologies, and the strengthening of technological independence in the sphere of cybersecurity. A second direction specifies the development of a national infrastructure for digital identification and access management, along with domestic software and hardware solutions.

These are institutional commitments with significant operational implications. The Strategy mandates government agencies to fund cybersecurity activities from their own budgets from April 2026, establishes a National Coordination Council for Cybersecurity under the Security Council, and requires dedicated cybersecurity units to be created across priority ministries. The sovereignty claim is explicit, formally adopted, and now binding on state bodies.

The substrate on which that claim lands is equally explicit, though it appears in no founding document.

The AI and cybersecurity infrastructure currently operating in Uzbekistan runs on hardware produced outside the country, primarily through NVIDIA chips distributed via official channels, and cloud platforms operated by companies with foreign capital and infrastructure located outside Uzbekistan’s territory. No domestic fabrication facility exists. No domestic accelerator production is under development at publicly documented scale. No domestic cloud provider with the capacity to run frontier AI workloads is identified in the materials accompanying УП-38. The declaration of technological independence and the material means to exercise that independence are, at this writing, two categorically different objects operating within the same institutional frame.

The question УП-38 raises is not whether the strategic direction is wrong. The question is what institutional content the declaration carries in the absence of the material infrastructure that would make it something other than a declaration.

II. What the Declaration Assumes

A declaration of technological independence in AI has a specific set of preconditions. It assumes that the declaring state can, in principle, operate the AI systems that its governance documents claim authority over, without depending on the unilateral decisions of the states or corporations that produce, license, and maintain those systems. This is the operational content of independence as distinct from its formal statement.

УП-38 does not specify these preconditions. It does not name a timeline for achieving domestic chip production, a program for reducing cloud dependency, or a mechanism for exercising sovereignty over models whose weights, architectures, and training processes remain with their developers. This is not a criticism of the document’s authors: no such preconditions appear in equivalent strategic documents across the vast majority of non-producing jurisdictions.

What the absence reveals is the institutional grammar within which such documents operate. In this grammar, sovereignty functions primarily as a declaration of authority. It is not a property that must be verified against material conditions before it can be asserted. A document can commit to “technological independence” in the same sentence that mandates continued use of foreign systems, because the gap between the two is not visible within the declaratory frame.

Essay 7 of the Beyond Control series established that real enforcement requires three elements: halt authority, independent access, and consequences for misrepresentation. None of these is present in the relationship between a sovereign declaring AI independence and the infrastructure layer on which that AI actually runs. The declaring state cannot halt access to its own critical systems if the provider revises access conditions, because the systems are not under local material control. It cannot independently access the operational logic of the systems it deploys, because the logic sits with the developer. And when the gap between the declaration and the material reality becomes operational, no existing mechanism permits attributing that gap as misrepresentation by a party over whom consequences can be imposed.

III. The Structural Position

Kazakhstan is the clearest available comparator in the region, not as a critique but as a structural illustration of the same institutional dynamic at a more advanced stage of public declaration.

In May 2025, Kazakhstan received what its Ministry of Digital Development described as the most powerful supercomputer in Central Asia, built on NVIDIA H200 graphics processors and developed in strategic partnership with Presight, part of UAE-based G42 Group. At the official launch ceremony in July 2025, attended by President Tokayev, the Minister of Digital Development stated that the project would “strengthen Kazakhstan’s digital sovereignty and reinforce its leadership in the region in developing AI technologies.” A subsequent official statement elaborated that the infrastructure would “accelerate the adoption of artificial intelligence, reduce dependence on foreign IT resources, and ensure the country’s technological sovereignty.”

The statements are precise. The architecture to which they are attached is equally precise: NVIDIA processors, UAE partner, foreign system architecture, hardware produced under US export control regulations that govern where and how NVIDIA H200 chips may be deployed. The sovereignty claim is made at the moment of acquiring foreign infrastructure, through a foreign partner, using chips produced by a company operating under the legal jurisdiction of a third state. The claim does not contradict the acquisition. Within the declaratory frame, these two facts coexist without tension.

The structural position is not unique to Kazakhstan. It describes the condition Essay 5 named as the Global South as a structural category: consumption without design, subordination without representation, responsibility without control, sovereignty without material power. The category is not geographic. Kazakhstan and Uzbekistan occupy it by virtue of their position in the AI production stack, not by virtue of their territory. The declaration of sovereignty does not alter this structural position. It names a direction without possessing the material means to reach it.

Procurement Trap logic (Analytical Note No. 7) applies here in its cleanest form. Architecture is fixed before governance is resolved. The dependency pattern is locked in through the acquisition decision. What comes after, including the sovereignty claim, operates on top of a material structure that was already set.

IV. Why the Gap Is Not Temporal

The institutional interpretation most commonly applied to gaps of this kind is developmental: the material infrastructure that would close the gap is not yet present, but will arrive in time through investment, reform, and technological development. This interpretation is familiar, and it is incomplete as a structural explanation.

The material preconditions for genuine technological independence in AI — specifically frontier chip production, advanced fabrication infrastructure, training-scale compute capacity, and indigenous model development at competitive capability — are not closed by investment policy alone. They are constrained by the same structural conditions Essay 10 identified: chokepoint concentration in fabrication tooling with a single Dutch supplier for extreme ultraviolet lithography, multi-decade lead times for any new national fabrication facility, and energy requirements for frontier compute that remain far beyond the region’s currently documented industrial and energy base. The trajectory toward these conditions does not run through a strategic document. It runs through twenty-year investment cycles in industrial and energy infrastructure that no current document authorizes.

The gap between УП-38’s declaration and the material substrate on which it lands is therefore not a transitional gap. It is the same structure Essay 9 identified when sovereign override meets an enforcement architecture that did not account for it, and the same structure Essay 10 identified when material predetermination closes the option set before sovereign choice is exercised. What this Note adds is the internal view: both limits visible simultaneously in a single domestic institutional act, not as abstractions from a global governance framework, but as properties of a specific document operating in a specific regulatory context.

That is the structural finding. The sovereignty claimed in УП-38 is real as a formal institutional category. It is not yet real as an operational category, because the material infrastructure whose control constitutes operational sovereignty does not exist within the declaring jurisdiction’s independent reach. These are two different objects that the declaratory frame allows to share the same name.

The correction window matters here exactly as Essay 7 described it: the longer the gap between formal claim and material reality goes unmapped, the harder it becomes to establish a governance architecture that is honest about what it can and cannot do.

Implications

For institutions operating within the УП-38 framework, the structural gap has a specific operational consequence. Cybersecurity and AI commitments that reference “technological independence” as a standard will, for the foreseeable future, be measured against systems whose independence conditions the declaring state does not independently control. This creates an accountability structure in which the regulator asserts sovereignty over a domain and simultaneously depends on external actors whose cooperation is not guaranteed across all access scenarios.

This is not a reason to revise the declaration. It is a diagnostic: the gap between formal sovereignty and operational sovereignty in AI requires explicit institutional mapping if the governance architecture is to function with accuracy about its own perimeter. A framework that knows the boundary of its actual operational control is a stronger institutional actor than one that does not. The honest institutional position is not “we have achieved technological independence” but “we have declared a direction, and we are entering a dependency pattern whose terms we do not fully control.”

That mapping — of what is actually within operational reach and what is not — is the residual governance work that the declaration leaves undone. The following Note in this series examines what a minimum viable architecture for that work looks like.

Signals to Watch

The specification of any procurement requirement for AI systems in Uzbekistan’s critical information infrastructure that includes a domestic alternative or a sovereignty condition — not merely a security certification, but a condition on whether the infrastructure underlying the acquired system remains within the jurisdiction’s operational control in the event of access revision by the provider.

Any revision of the УП-38 implementing roadmap to include explicit benchmarks for reducing the gap between the declared direction of technological independence and the actual infrastructure composition. Whether the roadmap treats domestic hardware and cloud capacity as a measurable target rather than an implied eventual state.

The terms of any expansion of cloud service provision by foreign providers under Uzbekistan’s digital government programs after April 2026: whether those terms include conditions on data localisation, access continuity, and the treatment of dependent government systems in the event of provider revision of service terms.

Kazakhstan’s next major AI infrastructure acquisition following the Presight/G42 cluster: whether the “digital sovereignty” framing persists or evolves to reflect the architecture of what is acquired, and what this signals for the regional template in Central Asia.

Sources & Notes

[1] Republic of Uzbekistan. Presidential Decree No. УП-38 “On the definition of the Cybersecurity Strategy of the Republic of Uzbekistan and improving the system for preventing cybercrime.” 10 March 2026. lex.uz

[2] Republic of Uzbekistan. Appendix No. 1 to Presidential Decree No. УП-38. Cybersecurity Strategy of the Republic of Uzbekistan for 2026–2030. Strategic directions include: “development of digital innovations, AI technologies, and strengthening technological independence in the field of cybersecurity”; “development of national digital identification and access management infrastructure, along with domestic software and hardware solutions.” lex.uz

[3] The Times of Central Asia. “Kazakhstan Launches Central Asia’s Most Powerful Supercomputer.” July 10, 2025. timesca.com

[4] Caspian Post. “Kazakhstan Receives Central Asia’s Most Powerful Nvidia-Powered Supercomputer.” May 19, 2025. caspianpost.com

[5] G42 / Presight. “Kazakhstan unveils first national supercomputer in strategic partnership with Presight.” July 11, 2025. g42.ai

[6] Khodjaev, O. Beyond Control: Theory of Limits of AI Governance. Essay 9 “The Sovereignty Question.” April 2026. okhodjaev.com

[7] Khodjaev, O. Beyond Control: Theory of Limits of AI Governance. Essay 10 “The Infrastructure Question.” April 2026. okhodjaev.com

[8] Khodjaev, O. Beyond Control: Theory of Limits of AI Governance. Essay 11 “The Institutional Gap.” April 2026. okhodjaev.com

[9] Khodjaev, O. Beyond Control: Theory of Limits of AI Governance. Essay 7 “The Correction Window.” April 2026. okhodjaev.com

[10] Khodjaev, O. Beyond Control: Theory of Limits of AI Governance. Essay 5 “The Colonial Pattern.” March 2026. okhodjaev.com

[11] Khodjaev, O. Governance Briefs. Analytical Note No. 7 “The Procurement Trap.” June 2026. okhodjaev.com

[12] Khodjaev, O. Governance Briefs. Analytical Note No. 10 “The Alignment Ceiling.” June 2026. okhodjaev.com

[13] International Energy Agency. Electricity 2024: Analysis and Forecast to 2026. iea.org

Oybek Khodjaev — over 35 years of experience in banking, finance, public administration, and business in Uzbekistan and the CIS. Author of the essay series “Beyond Control: Theory of Limits of AI Governance.” okhodjaev.com

The author advises public institutions and financial organisations on AI governance, verification frameworks, and institutional readiness.

Published: