Skip to content

Frontier compute: The future is how we combine these technologies

16 November 20264 min read
Guest Insights
Frontier compute: The future is how we combine these technologies

Frontier compute: The future is how we combine these technologies 

There is a tendency when we talk about frontier compute to focus on the next big technology. 

Today that might be AI. Tomorrow it might be quantum. 

But I think there is an equally important challenge in how we bring these technologies together as they continue to evolve and mature. 

The next generation of computing will not be defined by a single processor, platform or architecture. Organisations will increasingly have access to a mixture of compute resources, spanning CPUs, GPUs, HPC, AI accelerators, quantum inspired approaches and, eventually, quantum computers. 

The challenge becomes knowing which resource to use, for which workload, at which point in time. 

That changes the conversation. Rather than thinking solely about “when will quantum arrive?” or “do we need more GPUs?”, an additional question needs to be: “What becomes possible when we combine these technologies?” 

For me, that is where frontier compute gets interesting. 

We are already seeing computing architectures become more heterogeneous. Different types of processing, memory and acceleration are increasingly being brought together to address different characteristics of a workload. At Fujitsu, our work across supercomputing, advanced processors and quantum computing reflects this direction of travel. 

The real opportunity is not simply solving the same problem faster. It is the ability to combine and orchestrate different types of compute around the outcomes we are trying to achieve, including those that are difficult or simply not possible today. 

A complex problem might predominantly run on HPC, while AI could create surrogate models or analyse the results. Specific optimisation tasks within that same problem could draw on quantum-inspired techniques today and potentially quantum computing in the future. Rather than choosing one technology, different elements of the problem can be directed towards the computing capability best suited to them. 

But determining the right combination is inherently multidimensional. Performance matters, but so do cost, energy, security, data sovereignty, infrastructure capacity and the nature of the workload itself. 

And then comes a challenge: how do we make this growing range of compute resources work as one logical environment? 

It is not just about having access to different forms of compute. Organisations need to be able to place and move workloads and data between them appropriately, while managing the complexity underneath. The ability to abstract some of that complexity from the user could become just as important as the individual technologies themselves. 

That combination is also unlikely to remain static. As technologies mature, workloads change and new capabilities become available, the balance between different forms of compute will need to shift. Flexibility becomes critical. We need to create environments that can evolve as both the technology and the problems we want to solve change. 

Importantly, exploring these questions does not mean waiting for the next technological breakthrough. 

Organisations can start by understanding their workloads, where current approaches are reaching their limits and what capabilities they may need in the future. The investments being made today in AI, HPC, skills, applications and advanced compute can form part of the foundation for tomorrow’s hybrid environments. 

Frontier compute adds another dimension to how we approach the problems we are trying to solve. We have a growing range of computing capabilities available to us and the opportunity is in understanding how we bring the right combination together around what we are trying to achieve. 

The future is likely to be heterogeneous and increasingly dynamic, with workloads split and routed towards the computing resources best suited to each part of the problem. 

For organisations, the most useful question is not necessarily which technology comes next. 

Perhaps the question is: what do we want to do tomorrow that we cannot do today, and what combination of technologies could unlock it? 

That feels like the conversation frontier compute should be driving. 

Quantum computing by Fujitsu | Fujitsu Global