2025 has seen major advancements in AI technologies, starting with the Battle of the "Reasoning" Models. Open AI released GPT-5 in August, merging thinking capabilities with speed and multimodality; Chinese lab DeepSeek caused disruption by releasing a series of highly efficient, open models (DeepSeek-V3, R1, V3.2) rivalling top U.S. models in reasoning benchmarks but at a fraction of the training and inference cost; and in November, Google responded with Gemini 3, introducing iterative reasoning for math and logic problems along with generative interfaces – as well as all training and inference being on Tensor Processing Units (TPU), a shift away from Graphics Processing Unit (GPU) architectures.
AI has also dominated discussion this year, as I think it should. This kind of progress and change in a single year is a landscape shift, not just in the technology industry but across all industries. It’s important to note that adoption trails the innovation significantly at the moment in both business and government, with a very high percentage of organisations struggling to turn proof of concept into scaled solutions. By comparison, at Datacom, we have scaled many projects, for our own use and for our customers, by applying the engineering discipline we are known for. We design solutions that make use of a mixture of software engineering and Large Language Model (LLM) functionality where appropriate, carefully managing the data needed to support the use cases and extensively testing before being released.
The example that most fired up customers’ imagination this year was Datacom’s AI-driven modernisation of legacy applications, which at the time of publication, we are delivering for a large Australian state government and a major New Zealand port. By using AI agents for critical tasks including coding, analysis and testing – and with critical oversight from our own expert engineers – we have massively reduced the cost of redevelopment, and across key phases of the project, we have delivered in days what previously took months. This wasn’t a chatbot experiment. It was a complete reimagining of the software development lifecycle, showing how AI can deliver tangible results when applied with purpose and when the right questions are asked to determine the problems needing to be solved.
We’re also working hard to integrate AI into major payroll processes of Datapay and our local government product, Datascape’s core processes. Among the many other examples, we’ve paired smart digital engineering with AI to develop the Business Analysis Accelerator, an AI-powered business analysis tool which won the 2025 IIBA Innovation of the Year Award. And, we’ve had some fun building a conversational AI tool for the Melbourne International Comedy Festival to analyse and interpret attendee preferences as a personal entertainment concierge.
But the question we can’t lose sight of is not about AI’s adoption, but about how we design AI solutions that serve our communities and our countries. Deliberately, with people and purpose at the centre, and with resilience in mind.
Our AI Centre of Enablement and teams across Datacom have worked alongside government agencies and commercial enterprise, not just to develop AI tools and solutions, but to help build AI guardrails and strategies that focus on accountability, transparency and fairness. And where we are building AI solutions and tools – both within our own business and for our customers – we are focused on opportunities to augment human capability rather than replace it. That means designing systems that record how decisions are made, not just the outcomes, and ensuring human oversight in high-stakes areas like healthcare, education and employment.
There is growing awareness of the importance of resilient in-country AI processing as we build the foundations of our digital future. AI is power-hungry and data-intensive, yet most processing still happens offshore, creating dependency risks we can't ignore – particularly in an era of increasing geopolitical uncertainty. This challenge speaks directly to self-determination for the countries we live and work in, which is core to our purpose and vision.
Sovereignty isn't about withdrawing from global partnerships – it's about making deliberate choices around infrastructure, partnerships and governance. It means investing in regional capability, from renewable-powered data centres to GPU-as-a-Service, so sensitive workloads can be processed locally under frameworks that reflect our laws and values. It's also about leveraging the shared strengths of Australia and Aotearoa to build resilience and competitive advantage in a world where supply chains and geopolitics are increasingly uncertain.
As a company that is born and bred on engineering new solutions for new problems (we’re Always Solving), we believe the debate about the ethics of AI from intellectual property through to implications for the workforce are fair, but the genie is definitely out of the bottle and it is our role to take the engineering talent we have to help our customers understand how to apply these technologies to be more productive, more competitive and at a country level, secure our economic futures.
We are deep in our Strategy 2030 to fully engineer better productivity into every piece of our core service delivery to our customers. With what the technology is capable of today we cannot fully predict the extent to the change but, we know it will require us to think differently and adapt what the tools can do for us - that was not possible before.
As we enter 2026, the decisions we make now will shape the next decade of economic growth, social equity and digital resilience. We have the talent, the values and the momentum to lead – but only if we act with urgency and intention. Let’s design a future that reflects the best of Aotearoa and Australia – open to the world, accountable to our people and built to benefit us all.
To find out more about what the Datacom team has been up to in 2025, I’d encourage you to read through the stories we’re sharing in our Year in Review.
