Industrialising AI Development for Telco‑Scale Impact
The fireside chat explored how Swisscom is industrialising artificial intelligence through an AI factory model designed to accelerate development, reduce operational friction, and bring AI solutions closer to business value. The discussion highlighted the strategic, architectural, and organisational decisions required to operate AI at scale, as well as the lessons learned from Swisscom’s multi‑year transformation.
Strategic Drivers Behind the AI Factory Model
Swisscom’s approach is grounded in the recognition that AI demand is expanding exponentially across both internal and customer‑facing domains. The organisation faces thousands of potential use cases, each requiring rapid development cycles and consistent operational quality. The AI factory concept, described as a production line for advanced analytics, is intended to lower the barrier to building AI solutions by standardising infrastructure, tooling, and processes across the full lifecycle.
The conversation emphasised that very few individuals possess end‑to‑end skills across data engineering, data science, DevOps, and business context. This fragmentation slows delivery and increases dependency on scarce “unicorn” profiles. The AI factory model addresses this by providing shared platforms, automation, and reusable components, enabling business teams to develop solutions more quickly and with less specialised expertise.
Swisscom’s investment, including a $100 million commitment to NVIDIA infrastructure, reflects the scale of the opportunity. The company sees AI as a core enabler of customer experience transformation, cloud modernisation, and operational efficiency.
Architectural Foundations: Integrated, Hybrid, and Scalable
A central theme of the discussion was Swisscom’s decision to build an integrated ecosystem, combining data engineering, modelling, deployment, monitoring, and visualisation into a unified environment. This integrated approach reduces complexity, simplifies maintenance, and accelerates development.
Swisscom has adopted a hybrid architecture, combining on‑premises infrastructure with AWS native services. The on‑prem layer supports sensitive data categories, regulatory requirements, and existing data‑centre investments. The public cloud layer provides elasticity, access to advanced AI tooling, and the ability to scale rapidly. This hybrid model is designed to maximise flexibility, performance, and security.
The company leverages AWS services such as SageMaker and Bedrock to complement its internal capabilities, supported by professional services and cloud‑migration expertise. The fireside chat underscored that strong partnerships are essential for telcos entering AI at scale, particularly when navigating complex cloud ecosystems.
Scaling AI: Use Cases and Operational Acceleration
Swisscom is applying its AI factory to a broad range of use cases across customer experience, commercial optimisation, and network operations. Examples include customer segmentation, recommendation engines, churn prediction, ARPU optimisation, insolvency risk modelling, anomaly detection for infrastructure, and virtual assistants for both customers and employees.
The discussion highlighted how automation transforms the development cycle. Previously, a medium‑sized machine‑learning use case required around six months and thousands of engineering hours to move from concept to production. With the AI factory, Swisscom can now deliver similar solutions in four weeks, drastically reducing cost per use case. Given the scale of Swisscom’s use‑case portfolio, these efficiencies compound rapidly.
The AI factory also enables tighter alignment between AI solutions and business needs. By reducing friction and shortening development cycles, Swisscom can iterate more quickly, validate assumptions earlier, and deliver value faster.
Lessons Learned
The fireside chat concluded with reflections on the organisational and strategic lessons from Swisscom’s journey. A clear plan, combined with optionality and flexibility, is essential. Partnerships with hyperscalers and technology providers play a critical role in accelerating progress and de‑risking complex transformations. Choosing an integrated ecosystem simplifies operations and reduces long‑term overhead.
Cemmi emphasised the importance of staying connected to industry trends and best practices, as the AI landscape evolves rapidly and non‑linearly. Regular engagement with external experts helps validate decisions and avoid common pitfalls.
Ultimately, Swisscom’s experience demonstrates that operating an AI factory is not simply a technical challenge but a strategic shift toward industrialised, business‑aligned AI delivery. By moving from artisanal, handcrafted development to a scalable production‑line model, Swisscom is positioning itself to meet rising AI demand with speed, consistency, and measurable impact.