Sovereign AI Takes Center Stage as Nations Weigh Build, Buy, or Partner Strategies

Sovereign AI has moved from buzzword to balance sheet line item. At a Breakout Stage session during TechCrunch's Disrupt 2026 conference in San Francisco, a panel convened by Invest in Türkiye laid out a case that national AI independence is less about owning a proprietary large language model and more about controlling the layers beneath it: compute, energy, data centers, talent, and secure cloud infrastructure.

That framing matters for anyone sizing up the sovereign AI investment theme. The term has become shorthand for a geopolitical scramble as countries seek to reduce dependence on foreign cloud providers and foundation model vendors. But the panel, which included Bekir Polat, VP at Invest in Türkiye, Suresh Babu, SVP and GM of HPC and AI go-to-market at Hewlett Packard Enterprise, and Ozgur Altun, Director of Infrastructure at Turkcell, pushed back on the assumption that sovereignty means reinventing the wheel from the silicon up.

Why it matters

The stakes here go beyond national pride. Governments and corporations are making multi-year capital commitments based on where they believe data, compute, and AI capability should reside. If sovereignty is wrongly equated with building a domestic foundation model, countries risk sinking capital into redundant, undifferentiated infrastructure that competes poorly against hyperscaler economics. Get the definition right, and resources flow toward the layers where smaller markets can actually compete: regional data centers, energy-efficient compute, and applied AI tailored to local industries.

Türkiye's pitch, as presented at the session, is a hybrid one. Rather than isolating itself from the global AI stack, the country is positioning itself as a node within it: a market that builds domestic capacity in compute and data infrastructure while staying integrated with global technology partners and supply chains. That is a materially different bet than the all-in approach some nations have taken toward building their own frontier models.

What sovereignty actually requires

The panel's working definition of sovereign AI spanned several distinct components, each with its own capital intensity and risk profile. Compute capacity tops the list, physical GPU and accelerator infrastructure that a country can access without relying entirely on foreign cloud regions. Energy supply is inseparable from that calculation, since AI data centers are power-hungry assets and energy availability increasingly determines where compute gets sited.

Data centers themselves represent a separate investment category, one where HPE's Babu brought direct relevance given the company's role supplying high-performance computing and AI infrastructure to enterprises and governments. Talent development is the slower-moving variable: building a workforce capable of deploying, maintaining, and innovating on AI systems takes years, not quarters, and tends to lag behind infrastructure buildout in most emerging markets.

Secure cloud infrastructure and applications round out the picture. For a telecom operator like Turkcell, infrastructure and cloud security sit close to the core business, which explains why Altun's presence on the panel carries weight. A national operator with existing network infrastructure is arguably better positioned to pivot into AI infrastructure than a market with no incumbent telecom or cloud backbone to build from.

What the session did not offer was a foundation model race. No claim was made that Türkiye intends to compete head-on with the large US or Chinese labs on frontier model training. Instead, the thesis rested on capturing value in infrastructure, applied deployment, and regional data residency, areas where the capital requirements are lower and the competitive moat is narrower but still defensible.

Key risks

There are real constraints worth weighing before treating sovereign AI as a straightforward growth story. Emerging markets pursuing this strategy face capital intensity that dwarfs most other infrastructure categories. Data centers built for AI workloads require power density and cooling capacity that strain existing grids, and the energy transition needed to support that demand is itself a multi-year, multi-billion-dollar undertaking in most markets.

Talent remains the slower-moving constraint. Infrastructure can be built in 18 to 36 months with sufficient capital. A workforce with deep AI engineering and operations expertise takes considerably longer to cultivate, and competition for that talent is global, not local. Countries that build data centers without a parallel talent strategy risk ending up with capacity they cannot fully operate or monetize domestically.

There is also a dependency risk baked into the hybrid model itself. Pairing domestic infrastructure with global technology partnerships means remaining exposed to the strategic decisions of hyperscalers and chip suppliers who control critical inputs. A country that builds sovereign compute but still relies on foreign-made accelerators, foreign-trained models, or foreign cloud orchestration tools has only partially decoupled, and that partial decoupling may not satisfy the security and data residency goals that motivated the sovereign AI push in the first place.

The opportunity

For investors and infrastructure providers, the opportunity lies in the gap between what large markets can self-fund and what smaller, resource-rich markets need. HPE's participation signals that established infrastructure vendors see commercial opportunity in supplying HPC and AI systems to governments and enterprises pursuing sovereignty, rather than competing with hyperscalers for the same cloud workloads.

Turkcell's involvement points to a similar dynamic on the telecom side. Operators with existing network assets, data center footprints, and regulatory relationships are natural candidates to extend into AI infrastructure provision, particularly in markets where sovereignty concerns favor domestic operators over foreign cloud entrants.

Türkiye's broader pitch, positioning itself between East and West, connected to global supply chains but building domestic capacity, is not unique in concept. Similar framing has emerged from Gulf states, India, and parts of Southeast Asia. What differentiates any given market will be execution speed, energy availability, and the depth of partnerships with established infrastructure vendors like HPE.

Investment Implications

Sovereign AI is best understood as an infrastructure and services opportunity, not a foundation model arms race for most countries outside the US and China. Investors should track capital deployment in data centers, energy capacity, and HPC partnerships rather than expecting emerging markets to produce competitive frontier models. Watch for telecom operators and infrastructure vendors positioning themselves as intermediaries in this buildout, since that is where near-term revenue is most likely to materialize, well ahead of any payoff from domestic AI research programs.