HPC Summit Thailand 2026

Registration & Networking Breakfast

Welcome Address

James Loggie

Business Analyst

w.media

Keynote: The Future Compute Problem Delivering More Performance Without More Complexity

Future compute cannot be solved simply by producing a faster processor. Vendors must address the cost of moving data, memory limitations, energy consumption, deployment complexity and the ability to scale systems without making them impossible to operate. This opening keynote sets out the engineering challenges that will define the next generation of enterprise, cloud and high-performance computing.

Panel: The Server Is Being Reinvented. Will the Rack Replace the Server?

The traditional server was designed as a self-contained machine with its own processors, memory, storage, networking, power supplies and cooling. Future servers are likely to become more modular and workload-specific, combining CPUs, GPUs, AI accelerators, DPUs and specialised processors according to the application being supported.At the same time, the boundary between the server and the rack is beginning to disappear. Compute trays, shared networking fabrics, pooled memory, integrated power distribution and liquid-cooling systems are increasingly being engineered as one coordinated platform. Rack-scale systems can already operate as a single computing domain, while modular hardware standards are being developed to make server components easier to replace, upgrade and integrate.

Keynote: From Automation to Autonomy. How AI is Changing Infrastructure Operations

Automation has long helped data centers, networks and computing platforms execute repetitive tasks more quickly and consistently. Artificial intelligence introduces a different capability: the ability to analyse changing conditions, predict outcomes and recommend or initiate actions without relying entirely on predefined rules.As compute, storage and networking environments become more complex, vendors are combining automation, machine learning and real-time telemetry to improve workload placement, infrastructure utilisation, energy efficiency, fault detection and recovery. This panel examines where traditional automation ends and AI-driven operations begin, and whether future infrastructure can become genuinely autonomous without compromising security, reliability and human oversight.

Morning Coffee and Networking

Panel: The Memory and Storage Wall. Why Future Compute Depends on Faster Data Access

Powerful processors deliver limited value when they are unable to access data quickly enough. Innovations in high-bandwidth memory, flash storage, computational storage, memory pooling and composable infrastructure are changing how data is delivered to compute systems. This keynote explores how memory and storage architectures must evolve to prevent data access from becoming the next major performance constraint.

Panel: The Data-Movement Battle. Copper, Optics, and The Next Generation of Interconnect

As processors become faster, the connections between processors, memory, storage and switches risk becoming the limiting factor. Vendors are responding with higher-speed Ethernet, scale-up fabrics, silicon photonics, near-package optics and co-packaged optical systems. This panel explores which technologies can deliver the bandwidth, latency, energy efficiency and reliability required by future computing platforms.

Panel: From Chip to Rack. Why Compute, Power and Cooling Must Be Designed Together

Future computing systems cannot be engineered as independent collections of components. Processor selection affects memory, networking, power conversion, thermal management and rack design. This keynote examines how vendors are moving towards system-level co-design, where the entire computing platform is optimised around workload performance, energy efficiency, reliability and serviceability.

Networking Lunch

Panel: Can Future Compute Fit Inside Today's Data Centers?

Not every enterprise or data center operator can rebuild its facility each time a new processor generation arrives. Vendors must therefore develop systems that can bridge existing air-cooled environments, liquid-ready facilities, conventional power distribution and emerging higher-density architectures. This panel examines whether modular systems, hybrid cooling, sidecar infrastructure and phased upgrades can support future compute without forcing customers into complete facility replacement.

Keynote: Modular Compute. Scaling from an Enterprise Cluster to a Rack-Scale Platform

Future computing platforms must serve organisations with very different requirements. Some enterprises may need a small private cluster for engineering or inference, while cloud and research organisations require rack-scale systems. This keynote explores how modular servers, integrated compute trays and common management platforms could allow customers to begin with a targeted deployment and expand without replacing the entire architecture.

Keynote: Software-Defined Compute. Making Expensive Hardware Work Harder

Hardware innovation alone cannot guarantee business value. Future computing platforms will depend on intelligent scheduling, virtualisation, workload orchestration, observability and automated resource management to improve utilisation. This keynote examines how software can dynamically allocate CPUs, GPUs, memory and storage according to workload requirements, helping organisations generate greater output from every infrastructure investment.

Closing Panel: The Future Infrastructure Stack. Which Innovations will Enterprises Actually Adopt?

Technology vendors are developing faster processors, intelligent storage, optical connections, programmable networks and increasingly integrated rack-scale systems. However, not every innovation will be affordable, interoperable or practical for enterprise deployment. This closing panel identifies which technologies are approaching commercial readiness and what customers should prioritise over the next three to five years.

Closing Address

Event Networking Drinks