As Cambodia accelerates its digital economy, establishing data sovereignty and developing localised artificial intelligence (AI) have emerged as essential pillars for national economic resilience. From jurisdictional conflicts to cyber threats, relying entirely on foreign cloud servers carries significant operational and legal risks to domestic security and enterprise investment.
In an exclusive interview with B2B Asia News, Dr Prohim Tam, Instructor under the Faculty of Digital Technologies at the American University of Phnom Penh (AUPP), and Dr Rottana Ly, Research & Development Director and Lecturer at the Cambodia Academy of Digital Technology (CADT), outlined the structural risks of foreign tech reliance and the strategic imperative of custom Khmer language AI models.
Defining Data Sovereignty and Managing Foreign Jurisdiction Risks
Data sovereignty refers to the principle that data generated or collected within Cambodia from its citizens must be managed securely, transparently, and with full accountability.
Dr Prohim Tam highlighted three primary risk vectors associated with external data dependency:
- Jurisdictional Overreach: "One issue is when foreign jurisdiction can replace local jurisdiction," warned Dr Prohim. "This concerns whether or not our local jurisdiction can still overpower foreign jurisdiction – the jurisdiction applied to foreign owners or countries where the physical servers are located."
- Loss of Authority During Security Incidents: Another risk is cyber attacks, cyber incidents and data breaches. When facing such incidents or issues, it’s important for a country to have ownership and sufficient authority to collect all digital evidence.
- Hindering Domestic Investment and High-Level Workforce Talent: Dr Prohim noted that an overreliance on external cloud and data centre services will hinder the development of high level skills within the local workforce. He stressed that Cambodia should strive to see workers in the IT sector not only work as programmers or developers, but also senior managers, researchers and even Chief Technical Officers. He positively noted that there are local cloud and data centre providers like ByteDC, Daun Penh Cloud and Chaktomuk Data Centre helping the local economy by training high-level professional human resources.
"Local Problems Require Local Solutions" – The Khmer AI Imperative
The global surge in artificial intelligence adoption has brought data ownership to the forefront of national policy. Foreign large language models (LLMs) frequently fail when deployed in Southeast Asian enterprise settings due to structural linguistic barriers.
"When we constantly rely on foreign tech or foreign data, we cannot have full sovereignty over those assets ourselves," explained Dr Rottana Ly. "In research, to develop AI, we generally require three main pillars: human resources, digital infrastructure, and, finally, data. Therefore, we need to have our own data so that we can address our own problems. As the saying goes, 'local problems require local solutions'."
Dr Rottana further noted that offshore data sources cannot capture local market realities or linguistic nuances:
“When we use foreign data for AI training, it cannot reflect actual reality or the specific needs we have. In practice, when we use AI applications in the Khmer language, we often encounter errors or difficulties because Khmer is relatively more complex than foreign languages and has many special features regarding writing. Therefore, we need a large volume of Khmer language data to improve these AI models so that we can use AI more effectively.”
Developing robust Khmer technology requires extensive research across multiple modalities, including voice, writing, fonts, and text parsing, to build custom engines across enterprise, healthcare, and agricultural sectors.
Smart Regulation to Foster Digital Innovation
While strict data governance is critical, overly rigid frameworks risk stalling enterprise progress. Dr Prohim argued that Cambodia must reframe the traditional market perspective on compliance:
There is a saying that 'regulation kills innovation'. There are many use cases and scenarios in the United States, like companies moving headquarters from San Francisco, California, to Texas due to regulation and tax issues. We want to reshape 'regulation kills innovation' into 'good regulation fosters innovation'. Regulation shouldn't act as a wall that isolates innovation, but should be a protective shield.
To achieve this balance, Dr Prohim advocated for a risk-based, tiered data strategy, which would allow sensitive sovereign data to remain strictly protected locally while enabling commercial enterprises to scale globally.
"Data classification is crucial: sensitive data, national data, and other data that should be stored locally – we should maintain backups and follow proper procedures for storing such data in a separate tier,” he said. “Meanwhile, commercial data needs to expand to the global stage to increase revenue. Commercial data should have separate regulations granting freedom to exchange data through open standards, open APIs, or hybrid models, giving them access to up-to-date, state-of-the-art services to compete internationally.”
Overcoming Human Capital Bottlenecks and Digital Literacy Gaps
Building digital infrastructure and AI models is ineffective without a skilled workforce to manage them. Demographics present both an opportunity and a challenge for Cambodia.
"Looking at national data, the median age in Cambodia is 26.4," stated Dr Prohim. “This means half of the population are youth under 26, and among those who drop out of school, half do so at age 17. Looking at this reality, digital literacy is extremely low, even before discussing the university level, so a bottleneck exists right from the foundation level.”
He further identified several compounding friction points within higher education:
- Foundation Gaps: Students entering university often lack prior computer science exposure and face English language gaps that restrict self-learning.
- Educator Capacity: Rapid technological changes challenge the capacity of educators to maintain modern instruction standards.
- Curriculum Alignment: University course syllabi must be continuously updated alongside industrial advisory boards during recruitment analysis.
"We cannot expect students who finish Grade 12 without computer science exposure to succeed rapidly in university," Dr Prohim added. "We must strengthen their mindset and skill sets so that by the time they reach university, they can self-learn in English and access quality external resources. Curricula should be updated with an industrial advisory board analysing skills gaps identified during recruitment interviews. Some students have strong technical skills but lack communication skills to engage recruiters effectively, universities must address this."
Dr Rottana noted, however, that good progress is being made to develop STEM competencies within the Kingdom.
“The Ministry of Post and Telecommunications launched a roadmap for essential digital skills to encourage students who currently lean heavily towards social sciences rather than STEM,” he shared. “Since STEM enrollment remains low, we provide digital scholarships to encourage participation in digital and AI technologies.”
He also detailed what CADT is doing specific to AI human resource development:
AI expertise has only surged in recent years. CADT plays a role in human resource development by focusing on undergraduate and master's education, while partnering with foreign universities for PhD-level education. Beyond that, we train civil servants on AI fundamentals, data literacy, and AI literacy to accelerate digital transformation within the government. Our research team actively studies Khmer language technology, gathers Khmer data, and applies AI to address challenges in education, agriculture, and healthcare.
Ecosystem Collaboration: Government, Academia, and the Private Sector
To further bridge existing gaps, institutional alignment across government bodies, research institutions, and private industry is essential.
Dr Rottana Ly emphasised that academic institutions cannot work in isolation:
Currently, most research work is still done independently within academic settings without much engagement from the corporate sector. Private companies possess resources, data, and real-world problems. I think they can collaborate with research or educational institutions to submit research project proposals. This will consequently drive human resource development and yield solution-oriented technology and AI to address specific challenges.
He also noted the need for more developed digital infrastructure, such as advanced data centres and GPUs for researchers, while stressing the collaborative role all stakeholders must play to contribute data for AI training and development.
“We are currently constructing national data centres and national AI centres,” said Dr Rottana. “Importantly, everyone can contribute quality Khmer data across sectors to drive AI adoption.”
Through balanced regulation, localised data collection, and joint public-private workforce development, Cambodia is building the foundation for a secure and competitive digital economy.