Australia is a Deer in the AI Headlights

Australia’s stall and crawl approach to embracing AI won’t give us a green, “caring economy” no matter how often the Prime Minister rambles about it. The neglect is national self harm

The global tech giants have long loved doling out shock and awe and in the 21st century AI arms race there seemed none bigger than last week’s $US 500 billion Stargate project announced by a newly installed President Trump after his first full day in office.

The joint venture between American companies OpenAI and Oracle and Japan’s biggest tech investor, Softbank to accelerate the build out of AI data centre infrastructure in the United States was hailed as a clear signal to the world that the US was deadly serious about maintaining its lead over rival China for the critical technology.

When the founders of the joint venture partners, OpenAI’s Sam Altman, Oracle’s Larry Ellison and Softbank’s Masayoshi Son (Masa) all heaped praise on the President, trumpeting that they couldn’t “have done this without him,’ it was more than just platitudes or knee-bending.

Stargate Announcement: President Trump; Masayoshi Son, Larry Ellison, Sam Altman Source AP News

Trump’s repeal of a flurry of clean energy, AI safety and Diversity Equity and Inclusion (DEI) regulatory blocks mandated by the Biden administration represented a great unshackling for the US tech sector. The clear signal from Trump is to go all out to achieve AI supremacy over rival China – to develop AI systems “free from ideological bias, engineered social agendas” and clean energy restrictions.

But even relative to the eccentricities of President Trump and the characteristic overpromotion in the AI era, there was a strange “ad hoc” vibe to the announcement. Lots of vision and huge investment, but little detail. It was if this highly complex deal was cobbled together just for inauguration primetime.

Only hours later, Trump’s “best buddy” Elon Musk was the first to pour cold water on the deal when he tweeted on his X platform, “They don’t actually have the money.”

It led of course to a back and forth between Musk and his rival, Altman, which Silicon Valley venture capitalist, Benchmark Capital’s Biil Gurley described as “Better than an episode of Succession.”

But that turned out to be a mere distraction compared to what was being revealed on the other side of the world, and it should shake Australia out of its AI complacency.

In Davos Switzerland, the buzz at the annual World Economic Forum was all about a new Chinese breakthrough from AI lab, DeepSeek which has many commentators and US AI insiders claiming has leapfrogged the world.

Put very simply, the latest version of China’s open-source Large Language Model, (LLM) DeepSeek, which is funded by little-known Chinese quant hedge fund operator, Liang Wenfeng, is game changing for several reasons.

First, testing reveals it outperforms US models like OpenAI’s ChatGPT-4 and Meta’s open-source model, Llama and requires a tenth of the energy of those US models. DeepSeek also claims it cost less than $US6 million and only two months to develop compared to OpenAI’s years of development and $US5 billion annual spend.