Australia AI boom: why local startups are still missing out

An investor whose firm has backed Anthropic and OpenAI says Australia is still watching the AI boom from the cheap seats. In Australian Financial Review reporting, Boman Group chief executive Eric Gao said his firm, which manages about $1 billion, wants more Australian AI companies to back but is not seeing enough of them.

That complaint matters because it lands at a moment when Australia looks busy on AI from the outside. Canberra has put $70 million into an AI Accelerator and broader productivity measures, the National AI Centre says it is tracking more than 1,500 AI companies, and businesses are being urged by AFR reporting to buy local AI services instead of defaulting to imported tools. But the harder part of the stack is still company formation: turning AI demand, procurement and model usage into venture-scale local firms rather than a long tail of buyers and services shops.

Yet the loudest version of the founder-flight story risks flattening the problem. As Startup Daily argued in a comparison with New Zealand, zero capital gains tax on its own does not create a startup paradise. Australia can still attract capital, and recent rounds such as Arkeus’s A$25 million Series A show serious cheques are available. The deeper issue is whether the country has enough founder incentives, procurement pull and specialised talent to keep ambitious AI company-building here.

Gao’s warning and the week’s tax backlash describe the same bottleneck from different ends. Global investors see too few local AI champions; founders see policy settings that make the risk-reward equation worse just as model companies offshore race ahead. The question for Canberra is no longer whether Australia will use AI. It already is. The question is whether Australia will own enough of the companies that capture the upside.

Australia has AI demand, but not enough local winners

The strongest point in the Australian debate is not that AI is missing. It is that too much of the current wave is flowing into adoption rather than into new domestic vendors. In The Conversation AU’s analysis of the 2026 budget, the opportunity lies beyond generic chatbot roll-outs, in specialised tools, industry software and implementation capability. That fits Gao’s complaint. A market can be full of AI customers and still light on AI companies worth backing.

Kate Cornick, chief executive of the Tech Council of Australia, put the risk plainly in Guardian reporting on the startup backlash:

There is work to do to ensure Australia’s startup community doesn’t become collateral damage as a result of proposed changes.

— Kate Cornick, Guardian World

That does not read as a plea for subsidies alone. It reads as a warning that policy can make a market look vibrant while starving the pipeline that produces independent software firms. Recent AFR reporting on buying local AI services and on Australia’s rise of small language models point to real activity, especially in narrower enterprise tools. Still, those are early signals, not proof that Australia is minting its own OpenAI or Anthropic equivalents.

The budget helps on supply, but tax changes hit the founder equation

Canberra is not ignoring AI. The 2026-27 budget’s productivity package backs AI adoption and sets aside money for the accelerator. That is useful if the goal is more experiments inside existing businesses. It is less complete if the goal is to create more venture-backed AI companies, because founders and early employees care not only about grants but also about what happens to equity if the company works.

In Startup Daily’s reporting on the founders’ open letter, me&u co-founder Kim Teo said 47% of respondents would take the reforms as a sign to leave. The figure may be more mood than migration forecast, but it captures a live fear: the 50% capital gains discount would be replaced by indexation and a 30% minimum tax under Treasury’s proposed reforms. For startup teams, that changes the story told to engineers and product staff about why taking equity risk is worth it.

47% would feel like the Australian government are holding up a big sign for us to leave.

— Kim Teo, Startup Daily

Alfie Robertson distilled the issue in the same Guardian report:

The concern is not just about tax. It’s about incentives.

— Alfie Robertson, Guardian World

That is the hinge in the policy perspective. Treasury is trying to sell the reform as housing fairness and broader tax repair, while ministers have signalled consultation on startup settings. The most plausible landing point is not a full retreat. It is a carve-out that treats genuinely early-stage equity differently. Without that, the budget risks subsidising AI uptake on one side while dulling the incentives needed to create more local vendors on the other.

The offshore story is real politics, but bad economics if taken literally

It is tempting to turn every complaint into a simple offshore exodus story. That overstates the case. Startup Daily’s New Zealand comparison argues that zero CGT did not by itself create a stronger startup sector across the Tasman. Capital, customers, research depth and follow-on funding still matter more than any single tax lever.

InnovationAus recently carried a founder’s account of why the new settings make the startup gamble harder, which is better read as an incentives warning than a relocation census. And while ASX investors are clearly hunting for AI exposure in everything from data centre contractors to power and infrastructure plays, that is not the same as building a thick layer of application and model companies at home.

Arkeus’s recent A$25 million Series A shows capital will still back locally built deep-tech when it finds a sharp problem and a customer base. The problem is count and repeatability. One or two standout rounds do not solve Gao’s complaint if global capital still has to leave Australia to find the companies defining the model era.

This is why Gao’s critique hits harder than the week’s meme politics. Australia does not lack AI hype, buyers or budget line items. It lacks enough repeatable pathways from research, procurement and early adoption to durable local AI firms. The country can improve that without pretending tax is the only variable. Buying from local vendors helps. So does keeping accelerator money pointed at specialised products rather than generic chatbot pilots. But the fastest signal Canberra could send is that early-stage equity will not be treated like a windfall to be clipped after the risk is taken.

Until then, the likely outcome is not a dramatic stampede of founders to Wellington or San Francisco. It is something more Australian and more damaging: strong local AI demand, decent capital pools and plenty of technical talent, followed by the biggest upside accruing somewhere else.