g'day
psa: i dropped my first youtube video for the year this week:
i had lunch with a friend yesterday. he is easily in the top 0.001% when it comes to AI adoption.
i expected optimism.
instead, he was seriously worried.
and specifically about australia.
he is already living 2 years in the future: spending his days writing new agents, and managing them on telegram.
he runs task markets, where unused compute bids for tasks and he pays them in stablecoins.
he’s even applied openclaw to his friend’s architectural firm, which has reduced the time and cost of tendering new business by 80%.
now… there’s plenty being said about the ai jobpocalypse.
but his insights were alarming:
australia is a service economy. 80% of our GDP.
knowledge workers are 39% of the workforce.
and agents hollow out services first.
his second insight provided slightly more hope:
there’s not going to be enough compute.
so either australia can lose 20% of our jobs in the next 3-5 years,
or we can hedge and get serious about supplying compute at an industrial scale.
we are exposed to the downside case of AI.
but we have the knowledge and security to take advantage too.
we’re at a fork:
path one is the australian specialty: keep debating. keep ‘reviewing’. keep doing productivity panels.
path two is to treat compute like a national industry. cheap power. fast approvals. shovels in the ground.
we’ve got the ingredients: political stability, land, talent, and some of the best energy potential on earth.
now we find out if australia can still build.
charlie
