Artificial Intelligence is reshaping the semiconductor landscape—not only as a transformative force within fabs but also as a driver of global computing innovation.

Internally, AI is revolutionizing fab operations: enabling intelligent automation, driving yield enhancement, reducing waste and accelerating R&D.

Externally, the impact of semiconductor technologies on AI is even more profound. While leading-edge nodes play the key role in scaling AI performance, More-than-Moore technologies are essential for enabling low-power, high-efficiency computing at the edge, for connecting embodied AI with the real world, and ultimately for creating a path towards sustainable datacenter growth. This keynote will spotlight the critical contribution of More-than-Moore process technologies in powering the Age of AI, from lowest-power in-memory computing at the very edge to sustainable large-scale computing at the cloud.

Benchmarking has long been a central practice in semiconductor manufacturing, yet many approaches rely on generalized metrics or incomplete comparisons that obscure the true levers of fab performance. As fabs navigate escalating technology complexity, rising capital intensity, and increasingly variable customer demands, a more rigorous and data-driven benchmarking framework is required.
This presentation will outline a structured methodology for benchmarking semiconductor fabs that emphasizes metrics most directly linked to operational excellence — including yield, cycle time, equipment effectiveness, and cost efficiency. Particular attention will be given to the interdependencies among these metrics, such as the coupling of yield and cycle time, and the influence on fab performance.
Key discussion points will include:
• Methodologies for designing benchmarking frameworks applicable across diverse fab sizes and technology portfolios.
• The importance of coupling and contextualizing KPIs to avoid misleading conclusions.
• Insights from fab case studies demonstrating how targeted benchmarking initiatives have revealed hidden inefficiencies and enabled measurable improvements.
• Practical pathways for translating benchmarking insights into sustainable competitive advantage.
By advancing from descriptive comparisons to analytically grounded benchmarking, fabs can move beyond performance reporting to develop actionable strategies that enhance resilience, efficiency, and differentiation in a highly competitive industry.

As AI compute demand accelerates, semiconductor manufacturing faces unprecedented pressure to deliver high-yield, cost-effective chips at scale. Traditional manual yield tuning struggles to manage the complexity of advanced nodes such as Intel’s 18A. This talk explores how Intel leverages a comprehensive AI/ML-driven approach encompassing Design-Technology Co-Optimization (DTCO), advanced process control, and yield & process analysis to transform yield optimization.

By integrating digital twins of facilities, equipment, process, and logistics with real-time factory automation data, Intel simulates complex fab operations to proactively detect and prevent yield-impacting issues. Advanced AI techniques identify critical correlations across hierarchical data levels from wafer to die enabling rapid root cause analysis and accelerating yield feedback loops. Turning vast amounts of data into actionable insights requires not only sophisticated algorithms but also deep manufacturing expertise and disciplined execution. Leveraging decades of manufacturing excellence and breakthrough innovations like 18A PowerVia technology, Intel has developed a scalable framework that drives faster innovation , cost control, and robust yield performance to meet the surging demand of AI compute.

Attendees will gain a clear understanding of the complexities inherent in advanced process nodes and how evolving data-driven methods such as AI/ML and digital twin-based simulations are essential to managing these challenges. These approaches are becoming increasingly critical in today’s AI-driven semiconductor manufacturing landscape.

This talk introduces Lean Automation for Cycle-Time Wins.In today’s semiconductor industry, cycle time is a decisive lever for competitiveness and market leadership. We explore how lean manufacturing principles, when combined with state-of-the-art AI and automation technologies, can deliver substantial gains in production efficiency.Key innovations include:

• Knowledge graphs for unifying heterogeneous shop-floor data and ensuring full lineage and traceability.
• Causal discovery via KG link prediction for interpretable, data-driven root-cause hypotheses.
• Graph learning and GNNs for intelligent route and resource optimization.
• Selective state-space sequence models for fast, accurate cycle-time forecasting and online rescheduling.

Together, these methods establish a closed loop from causal insight → optimized plan → executed intervention (in the digital twin or on the fab floor), driving reductions in both average and tail cycle times.

Presentation abstract (shorter):

This talk introduces Lean Automation for Cycle-Time Wins, showing how lean manufacturing principles can be supercharged by state-of-the-art AI. We highlight knowledge graphs for data fusion and traceability, causal discovery for interpretable root causes, graph neural networks for routing optimization, and selective sequence models for cycle-time forecasting. Together, these methods enable a closed loop from insight to action, delivering reductions in semiconductor fab cycle times.

To revolutionize how intel’s facilities operations run, the industrial automation group is implementing advanced analytics and AI to extend equipment life and reduce costs. In this presentation, we will show how intel deployed a variety of anomaly detectors to catch events before they escalate into downtime as well as show how the Facilities team is leveraging advancements in AI for efficiency gains and cost reductions in their operations.

Balancing sustainability goals with cost pressures is a growing priority in semiconductor manufacturing. This session presents a practical approach to integrating carbon footprint data into cost models, enabling more informed decisions across sourcing, materials, and operations. Using examples from fab-level implementation and equipment supply chains, we highlight how lifecycle data, supplier engagement, and process transparency can drive measurable improvements in both environmental impact and cost efficiency. The discussion will focus on key metrics, decision frameworks, and lessons learned from applying these strategies in real-world contexts.

In the AI-Driven Autonomous Factory of the Future, both AI techniques and Digital Twins play an important role for enhancing fab performance and maximizing fab out.
The presentation will elaborate

(1) How Digital Twins empowered by stochastic simulation can be used for dynamic capacity modelling of wafer fabs, and

(2) How AI techniques will empower the critical step from currently predictive to prescriptive analytics of fab operations by increasing the degree of automation when an AI optimisation agent has to navigate through the mind-blowing complexity of the search space associated with such fab operations.Such AI-driven, simulation-based optimisation enables essential capacity planning and WIP flow optimisation decisions such as
• Prediction of bottlenecks and prescription of actions for maximizing fab output,
• Dynamic capacity allocation based on real-time demand and constraints,
• Prioritization of operational opportunities to simplify or better manage queue time requirements, depending on impact on cycle time, WIP flow and potential excursions,

• Determination of the direct impact of OEE initiatives on overall fab output and prioritization of projects accordingly,

• Optimisation of advanced strategies for wafer start decisions and balancing output maximization with cycle time reduction through intelligent wafer release, taking into account predicted line constraints, capacity, and demand. How the foundation for this has been successfully laid will be showcased through the benefits that have been achieved by Micron through the deployment of the simulation-based D-SIMCON Digital Twin framework and a collaborative effort to further develop these use cases.

Manufacturers face a rapidly changing sales landscape where digital engagement is no longer optional—it’s essential. This session explores how Product Information Management (PIM), Digital Asset Management (DAM), and Product Experience Management (PXM) transform the way manufacturing companies’ market, sell, and deliver value.

This talk explores how Agentic AI is advancing semiconductor manufacturing—enabling autonomous decision-making, equipment self-healing, and zero-defect operations. Through real-world cases in semiconductors and advanced manufacturing, it examines proven impact and future approaches that drive leap changes in quality, efficiency, and industry autonomy.