OpenAI chief executive Sam Altman’s prediction that the world’s first billion-dollar startup built by a single person with a laptop is coming soon is an unsettling vision of progress.
But progress for who? This seems like a small-minded ambition.
OpenAI’s chief strategy officer Jason Kwon, who was on a charm offensive in Sydney and Canberra last week, repeated the vision, patting the local startup sector on the head as he said Australia might even produce the first uni-unicorn.
Well, I suppose if the point of this talk of a blue sky is to point to a low-employment future, then Sam’s vision is quite illustrative.
And now, in Australia, we have the latest Startup Muster report that points us in the same job reduction future.
The headline numbers point to an Australian ecosystem in which startups are each employing dramatically fewer staff.
Here are the headlines. More than half of all startups in Australia are building AI products, and two thirds of startups use AI for core business functions.
The average number of full-time employees of local startups fell by nearly half in the past year – from 8.2 employees in 2024 to 4.2 this year.
The average number of planned hires for the next 12 months dropped by 30 per cent, from 5 employees to 3.5 employees.
It is worth noting that while 25 per cent of startups using AI reported “some of their AI in Australia”, more than 60 per cent of AI users in startups weren’t sure where their AI was hosted.
Only 11 per cent could name any safety standards for AI use in Australia.
In a week in which Australia’s social media age restrictions are being put in place as a response to the damage done to children and others by the policy decisions of more than a decade ago, you’d hope there are alarms going off somewhere.
The fact that most Australian startups are using AI tools to build their products and don’t know where that AI is hosted is really quite something. The data does point to the US AI as utterly dominant, with DeepSeek, Qwen and Mistral usage negligible.
OpenAI’s ChatGPT remains the dominant AI among startups, but Google and Anthropic have made massive market share gains in the past year, in a growing market.
For Startup Muster’s ever optimistic and agreeable founder Murray Hurps, the 2025 numbers are an accurate portrayal of a remarkable transition.
Startups are changing. “They have smaller teams. They’re seeking less funding and less grants,” he said.
“The size of teams getting smaller, the capability of individuals getting larger. The upside of that is hopefully we end up with more startups that are being started,” Mr Hurps said.
“On a simple read, you could look at it and say ‘Okay, that’s not great for employment’.”
“But I think the more likely outcome is we’ll end up with more companies being started by smaller numbers of people because of these barriers to entry are being taken out.”
There is a lot of data in this report that point to some pretty clear overall trends. It I certainly worth spending some time on – particularly for policymakers, deciding where to put their support dollars.
The self-described ‘deep tech’ startups are cause for optimism.
Mr Hurps points to a bifurcation between AI/software startups and deep tech.
“You’re either very small, using AI a lot and have a very low cost base, or you’re a deep tech, which very capital dependent, needs a larger team and longer runway,” he said.
“And that’s becoming a notable divide.”
You can read the 2025 Starup Muster report here.
Do you know more? Contact James Riley via Email.