The narrative around Australian AI talent has been consistent for two years: there’s a shortage, salaries are skyrocketing, the brain drain to the United States is real. That narrative isn’t wrong, but a closer look at the mid-2026 data suggests a more bifurcated picture than the headlines convey.

A read across recent industry surveys, Tech Council of Australia reporting, and conversations with hiring managers across the ecosystem reveals a market where some segments are genuinely tight and others are quietly oversupplied.

Where the shortage is real

Senior applied machine learning engineers with end-to-end production experience remain genuinely scarce. The combination of model selection, deployment infrastructure, evaluation, monitoring, and the ability to ship working systems is in short supply nationally. Salaries for this cohort have moved into the high-A$300K and frequently A$400K-plus range, with the top tier of Sydney and Melbourne offers landing between A$450K and A$550K for senior individual contributors.

Specialist roles in AI safety, evaluation engineering, and AI security are similarly tight. The pool of Australian-trained candidates for these is small, the international flow inward has been thin, and the demand has accelerated as enterprise AI moves into production at scale.

Research-grade ML scientists are scarce by definition. The leading research labs in the US and UK pay materially more than any Australian employer can match, and the structural pull is real. The handful of research roles being created locally — at CSIRO Data61, at the major universities, at a few well-funded startups — fill quickly when they’re posted.

Where the shortage is overstated

Junior-to-mid-level data scientists and analysts have not been in short supply for some time. The wave of graduates from bootcamps, master’s programs, and university data science streams has produced a substantial pool. Many of these candidates are perfectly capable but are competing for roles that companies have been slower to create than the talk would suggest.

LLM application developers — engineers building on top of frontier model APIs — are not in short supply. The technical skills required for competent application-layer work are well within reach of any experienced full-stack developer, and the pool of candidates is correspondingly large.

The “generalist AI consultant” category, populated heavily by former management consultants and former enterprise software people who’ve rebadged toward AI, has produced more supply than the market needs. The clients who can distinguish substance from positioning are filtering effectively. The clients who can’t are creating budget that won’t deliver outcomes.

The geography of the market

The Sydney and Melbourne markets account for the substantial majority of high-end AI hiring. Brisbane has built a respectable secondary market, particularly around the universities and a few growth-stage companies. Perth, Adelaide, and the regional centres have small concentrations but the overall volumes are modest.

Remote-first hiring has expanded the talent pool meaningfully for mid-market employers. A Sydney-based fintech can now realistically hire from Hobart, Cairns, or Albury without the relocation friction of five years ago. The flip side is that the top end of the market is increasingly competing globally — the senior engineers Australian employers want are also being courted by US and European firms running fully remote.

The internal Australian movement story is also more nuanced than the headlines suggest. Yes, some senior talent is moving offshore. The greater volume of movement is within Australia, between employers — and increasingly into specialist boutiques rather than into the big enterprises. A consultancy like Team400, or a handful of similar firms, has been hiring meaningfully from the big banks and from Big 4 consulting alumni networks, with the pull factor being interesting client work rather than maximum compensation.

What the compensation data actually shows

A few specifics from the most recent surveys.

Senior ML engineer median total compensation in Sydney sits around A$310K-340K for 6-10 years experience. In Melbourne, slightly lower, around A$285K-320K. Brisbane is meaningfully lower, around A$240K-280K, though the gap has been narrowing.

Lead and staff-level AI engineering roles cluster in the A$380K-450K range across Sydney and Melbourne, with significant variance based on whether the role includes meaningful equity and what the equity is.

Data scientist roles are flat year-on-year in nominal terms — slight real wage decline given inflation — reflecting the supply-side reality discussed above. Compensation for genuinely senior data scientists with statistical rigour and business judgment hasn’t compressed, but the broader category has.

AI product manager roles have been the most volatile compensation category. The premium for product managers who can credibly run AI product work is real but the definitions vary so widely between employers that comparing offers is difficult.

What the next 18 months probably look like

A few patterns I’d expect to continue.

The senior shortage at the genuinely-skilled end won’t ease quickly. The training pipeline for production-grade AI engineering is multi-year and the demand-side is still growing. Employers should expect compensation pressure at the top end to continue through 2027.

The mid-level oversupply will work itself out as enterprise demand catches up. The clients who finally start serious AI build work in late 2026 and through 2027 will absorb a meaningful chunk of the current oversupply, but the absorption will be uneven across geographies and sectors.

The specialist boutiques will continue to take a disproportionate share of the most interesting work. The economics of building a focused team of 30-80 specialists, rather than scaling out a generalist consultancy, are genuinely favourable for both the firm and the senior engineers who join them.

The honest summary is that the Australian AI talent market is healthier and more interesting than the shortage narrative implies. The premium for top-end specialist talent is real and earned. The mid-tier has more flexibility than candidates are sometimes told. And the geography of opportunity is broadening as remote-first arrangements mature.