Window of Opportunity or the Bottleneck? An Interview with Monika Malik

Social drift, generational varnas and emerging digital castes reveal the truth of our era from a different angle. The world is moving toward a divide between those who can shape and govern models and those who merely use them. Kazakhstan has yet to formally declare, at the legislative level, the creation of its own engineering-cognitive elite – and there is a real danger that such elite will be imported. This risks a new form of colonization across all domains: Pipelineistan (extraction), Mineralstan/Lithiumstan (extraction and processing), and Cognistan/DishBrainstan (model training and deployment). The highest added value of the 21st century is generated at the stage of model training. If Kazakhstan remains only a supplier of data and energy, it will lose.

Continuing the discussion on Kazakhstan’s readiness for the coming era – outlined in the article “A Sovereignty Claim: The Generational Digital Frontier” – the SPIK.KZ editorial team turned to Monika Malik, Lead Data & AI Engineer at AT&T, for clarification and deeper insight.

From the Editorial Board: The Battle for Intelligence

2024–2032 is a unique window. Too many countries are only now beginning to build their AI architectures. In ten years, catching up will be impossible – too late and too expensive. For the first time, Kazakhstan is participating not in a game for resources or corridors, but in a game for mindshare. In a battle over whose algorithmic architecture will shape the destinies of future generations. The stakes are not simply high. These are the stakes that determine whether a country is a subject – or merely data inside someone else’s model.

The choice between becoming a producer of cognitive technologies or a consumer of them is not a sectoral question. It is the demographic trajectory of the nation. And this is precisely where the voices of international experts become critically important – not as external judges, but as navigators who see how different countries convert resource advantages into cognitive ones. That is why the SPIK.KZ editorial team initiated a conversation with Monika Malik. We aimed not merely for an expert interview, but for a mirror – an external perspective reflecting Kazakhstan’s current position. In our view, this is a window into the practical meaning of technological sovereignty in the age of AI: what conditions it requires, what mistakes countries make when they try to buy the future with money, and what mechanisms allow infrastructure to transform into capability – and capability into economic power.

Her insights are especially vital now, as Kazakhstan bets on sovereign compute, raises the question of national models, commits to hyperscale infrastructure, and attempts to construct a new system of managed growth. When Monika speaks about human capital, institutional risk, the strategic role of H200 clusters, opportunities for exporting “digital competence,” and the threats of “AI colonialism,” – this is not abstraction. It is an early roadmap for what kind of Kazakhstan will be handed over to generations Z and α.

About the Expert:[/b] Monika Malik is a Lead Data/AI Engineer at AT&T, specializing in building and operating production-grade AI systems across AIOps, secure LLM platforms, and responsible-AI governance. Her work and commentary have appeared in CIO.com, Built In, and IT Brew, and she frequently speaks on generative AI for document automation as well as the emerging risks of “shadow AI.” Prior to joining AT&T, she worked at Barclays, where she focused on translating policy into pipelines - from model-risk gates and data-provenance controls to zero-trust deployment frameworks.

Kazakhstan’s Readiness for AI. In her responses for SPIK.KZ, Monika noted that Kazakhstan’s readiness for AI looks promising but remains uneven. According to her, the country ranks among the global top 10 in the quality of online public services and holds 24th place in the UN E-Government Index; however, applied AI competencies and production-grade MLOps are still lagging. She emphasized that Kazakhstan is already building a real foundation for sovereign computing – in particular, thanks to the Alemcloud H200 supercomputer and Tier III data centers. She also noted that the country has an established political course: the national AI development concept through 2029 and several national platforms set the right direction, but execution and talent shortages remain bottlenecks. In her view, governance risks must also be considered given the country’s Corruption Perceptions Index score (40/100); the state should accelerate the development of procurement and data-use regulations to avoid losses and dependency.

On the H200 Acquisition and Its Viability. According to Monika, acquiring the H200 makes strategic sense — provided it is oriented toward real workloads. She stressed that these GPUs will enable fine-tuning Kazakh and Turkic-language LLMs, delivering secure generative services for government needs, and strengthening high-performance computing in energy, logistics, and climate modeling. However, she warned that GPUs alone are insufficient – without high-quality datasets, engineers, and well-managed allocation policies, computational capacity remains underutilized. Monika also emphasized that the cluster must become a platform rather than a “showcase trophy.” It should be wrapped in a service layer – GPU-as-a-Service – and accompanied by the creation of a sovereign registry of models and datasets. In her view, the momentum is already there: recent reports on expanded Nvidia shipments to Kazakhstan confirm the realistic potential for scaling, but outcomes will depend on management quality.

Lessons from Countries that Transformed Natural Resources into Cognitive Capital. Monika highlighted three models. In Norway, she said, resource revenues are treated as “patient capital” protected from politicization, governed by strict standards, and coupled with investment in human capital – including AI competencies at the board-of-directors level. In Singapore, she emphasized, investment is accompanied by operational certification of AI reliability – for example, via AI Verify and specialized AI sandboxes, which enable safe and scalable deployment. In Estonia, she noted, the critical factor has been the state-as-a-service model: digital identification and harmonized data exchange form a foundation on which AI fits naturally as a native layer rather than being bolted on afterward.

The Impact of AI on the Economic Geography of Central Asia. Monika noted that AI can shift the region from an extraction-based model to one centered on exporting cognitive services. According to her, Kazakhstan is already moving toward becoming a regional model-ops hub thanks to its large-scale data centers (up to 200 MW) and an upcoming Tier IV facility. She also highlighted a linguistic advantage: Turkic- and Kazakh-language LLMs can generate durable intellectual property that could underpin service exports, aligning with the national goal of achieving $1 billion in IT exports.

The Sovereign Wealth Fund and the Risk of “AI Colonialism.” Monika believes that Samruk-Kazyna has a dual impact. With assets of $70–80 billion, the fund can either serve as an anchor for sovereign compute and open ecosystems, or increase dependency if it pursues exclusive agreements with global providers. She emphasized that mitigating these risks requires multicloud portability, FRAND-based APIs, backup storage for the weights of critical models, and joint investments in open-source tools – from evaluation frameworks to data trusts.

Balancing ESG, Technological Sovereignty, and Pragmatism. Monika notes that sovereign AI can be developed sustainably if computational workloads are co-located with renewable energy sources, if energy-efficiency targets are enforced, if heat reuse from data centers is implemented, and if the energy base is diversified, including future nuclear generation. She underscored the importance of a “trust infrastructure” – applying the international AI management standard ISO/IEC 42001 and the NIST AI RMF across all government procurements to standardize risk governance. She also considers it effective to link capital to measurable outcomes – for example, citizen wait times or export revenue per kWh of compute – to align ESG priorities with economic growth.

Monika’s responses make one point unmistakably clear: Kazakhstan is not merely facing a technological choice – it is standing at a critical juncture of historical self-definition. Her analysis reinforces the central thesis of the article: in the emerging era of Cognistan/DishBrainstan, capital is being transformed into the ability to train models and manage cognitive infrastructure. Monika stresses that even sovereign compute (H200 clusters, hyperscale data centers, national platforms) does not, by itself, guarantee sovereignty if the country fails to build human capital, MLOps capabilities, and institutional frameworks capable of keeping value inside the nation. This aligns fully with the core warning of the “new great game”: the owner of the future will be the one who owns not the resources, but the architecture of data and algorithms.

Monika Malik’s perspective allows us to call things by their proper names: Kazakhstan may build compute infrastructure and still remain a “client state” within the global platform ecosystem. She states this directly. Compute without data governance, without sovereign models, without an open ecosystem of universities and companies, ceases to be a strategic asset and becomes a trophy – “hardware without thought.” This is why she emphasizes governance risks, transparent mechanisms for managing national data, and multicloud independence. These are not bureaucratic details; they are tools for preserving agency. They determine which reality generations Z and α will inhabit – their own, or an imported one.

Her recommendations on building a sovereign stack – from FRAND-based access to model escrow mechanisms – effectively form a checklist for Kazakhstan’s statehood in the 21st century and beyond. If Pipelineistan required pipelines and Mineralstan required processing, then Cognistan requires institutions. And her answers articulate with striking clarity the most important insight: sovereignty is not the product of a political declaration but a function of engineering maturity, an open innovation environment, and a country’s ability to convert compute into data, data into models, and models into exportable cognitive competence. This is the cycle that can move Kazakhstan from being a “resource player” to becoming an “architect of cognitive systems,” at least at the regional level.

Finally, Monika’s view reveals just how high the stakes are for generations Z and α. She effectively describes a window of opportunity for countries that have not yet secured their place in the global AI hierarchy and are now passing through the bottleneck of proving their viability. If Kazakhstan uses this moment to build an engineering core, risk standards, an open compute market, sovereign models, and institutional guarantees, then generations Z and α will inherit a country in which they are the creators of algorithms. If not, they will grow up inside an architecture designed by others. Thus, Monika’s responses become not just an expert position but the final confirmation of the central conclusion of the previous article – that the New Great Game is a race for the cognitive agency of future generations, and Kazakhstan has a rare opportunity to participate in this game as an author rather than a service appendage.

[b]Rafael BALGIN