It’s a good time to take stock. Where does Australia actually stand on AI in early 2026? Not where the press releases say we stand. Where we actually stand.

I’ve spent the past two months pulling together data, talking to people across the ecosystem, and trying to build an honest picture. Here it is.

Adoption: Accelerating But Uneven

AI adoption in Australian business is genuinely accelerating. The percentage of Australian companies using AI in at least one business process has grown from roughly 25% in early 2025 to approximately 40% in early 2026. That’s meaningful growth.

But the number hides enormous variation. In financial services, adoption is above 70%. In healthcare, it’s around 45%. In construction, it’s below 20%. In agriculture, it’s growing rapidly from a small base.

Company size matters more than sector. Large enterprises (1,000+ employees) have adoption rates above 60%. Mid-market companies (100-999 employees) are around 35%. Small businesses (under 100 employees) are below 20%.

The usage depth varies even more. Many companies counted as “adopting AI” are using basic tools like ChatGPT for ad-hoc tasks. Fewer have deployed AI in core business processes. Fewer still have AI driving automated decisions. The pyramid narrows quickly from “using AI somewhere” to “using AI strategically.”

Policy: Progress But Still Behind

The federal government has made progress on AI policy since 2024. The voluntary AI Ethics Principles are established. The National AI Centre is operational. Mandatory guardrails for high-risk AI have been announced, though not yet legislated.

But the gaps are significant. We still don’t have:

• Legislated AI safety requirements for high-risk applications
• A mandatory AI transparency register for government AI systems
• Updated privacy legislation that explicitly addresses AI processing
• A clear regulatory framework for autonomous systems
• Standardised approaches to AI procurement across government

Other comparable nations are further ahead. The EU AI Act is being implemented. The UK’s AI Safety Institute is operational. Singapore’s model governance framework is mature. Canada’s AI and Data Act is progressing through parliament.

Australia isn’t dramatically behind, but we’re not leading, and the gap isn’t closing.

Research: World-Class but Underleveraged

Australian AI research is genuinely excellent. Our universities and CSIRO produce world-class work. Australian researchers are disproportionately represented in top AI conferences relative to our population.

The problem is commercialisation. Australian AI research produces papers and patents but struggles to produce products and companies. The research-to-market pipeline has improved but remains slower than in comparable countries.

Brain drain continues. Top Australian AI researchers leave for better-funded positions overseas. Those who stay often lack the compute infrastructure and industry partnerships needed to translate research into applications.

The positive development: applied AI research is gaining ground over pure research. Programs that partner universities with industry on practical problems are producing more commercially relevant outputs. CSIRO’s Data61 has improved its commercialisation focus.

Investment: Growing From a Low Base

AI-specific venture investment in Australia has grown but remains small by international standards. Total AI-related VC investment in 2025 was approximately $800 million, a meaningful increase from previous years but a fraction of comparable markets.

Government investment in AI, combining federal and state programs, totals roughly $2 billion committed over various timeframes. That’s significant for Australia but modest compared to the US, EU, UK, and several Asian nations.

The super fund involvement I wrote about earlier is the most interesting capital development. If super funds deploy committed capital into AI investments at the scale suggested, the capital constraint loosens significantly.

Corporate AI spending is harder to measure but growing rapidly. Major Australian companies are allocating between 2% and 8% of their IT budgets to AI initiatives, with the percentage growing annually.

Workforce: The Gap Is Real

The AI skills shortage in Australia is genuine and growing. Employer demand for AI skills outpaces supply at every level, from junior data scientists to senior ML engineers to AI leadership.

Universities are graduating more AI-skilled students, but the quality gap between theoretical knowledge and practical capability remains. Industry is absorbing graduates but reporting significant ramp-up times.

The AI translator shortage, people who bridge technical and business domains, is arguably more acute than the technical shortage. Companies struggle to find people who understand both AI capabilities and business context.

Immigration has been an important source of AI talent for Australia, but visa processing delays and competitive global markets make it increasingly difficult to attract international AI professionals.

My Assessment

Australia is in a solid but not exceptional position on AI. We have genuine strengths: world-class research, a stable regulatory environment, strong data infrastructure, and specific domain expertise in resources, agriculture, and healthcare.

Our weaknesses are equally real: limited compute infrastructure, a research-to-market gap, regulatory lag, uneven adoption across company sizes and sectors, and a persistent skills shortage.

The trajectory matters more than the current position. If current trends continue, adoption growth, policy development, and investment acceleration, Australia will be well-positioned by 2028. If any of those trends stall, we risk falling further behind countries that are moving faster.

The single highest-impact action would be accelerating AI regulation. Clear rules would remove the uncertainty that slows both adoption and investment. Every month of regulatory ambiguity is a month where Australian businesses either hesitate or make decisions they may need to reverse later.

The second-highest-impact action would be investing in domestic AI compute infrastructure. You can’t have AI sovereignty or a competitive AI industry without the computational foundation to support it.

We’re not where we need to be. But we’re not as far behind as pessimists suggest, and the path forward is clear if the will to walk it is there.