China’s ‘Sputnik moment’ hits Trump, Wall Street and Australia

AI has already cemented its place as the most important technology of the decade, upending industries, economies and daily life at an unprecedented rate. And China’s free, powerful chatbot is threatening to rule the roost.

For the newly re-elected Trump – and the entire Western world, including Australia – there’s much at stake.

Loading

Tech has become core to Trump’s power and his second presidency, and his dalliance with the likes of Elon Musk, Sam Altman and Oracle boss Larry Ellison offers a glimpse of the next four years. It’s a Trump tech “broligarchy”, with the spectre of China now hanging over everything.

Trump called DeepSeek’s arrival a “wake-up call” for US companies who must focus on “competing to win”.

Joan Westenberg is an Australian technologist and former start-up employee, and now operates independent publication The Index. For Westenberg, DeepSeek’s ascent represents an existential crisis for Silicon Valley and its executives like Altman and Zuckerberg.

She says rather than viewing it as an arms race, however, what DeepSeek has proven is that collaboration beats competition, and openness trumps secrecy.

DeepSeek’s cheap, open AI model sent a tidal wave through Silicon Valley but thrilled scientists, given it’s open for researchers to examine at will.

Published under an MIT licence, DeepSeek’s model can be freely reused, though technically, it’s not considered fully open source given its training data has not been made available. China has earned a reputation for secrecy but DeepSeek is anything but.

“The greatest threat to American AI dominance was never China,” Westenberg tells this masthead. “It was the delusion that dominance itself was sustainable.

“When you mistake money for momentum, you build castles on quicksand. And a trillion dollars vanished because the market finally saw through the theatre.”

‘When you spend billions convincing the world that only your path leads forward, watching someone take a different route … becomes an existential crisis.’

Joan Westenberg

American tech giants hoarded the narrative about AI’s future, according to Westenberg, and DeepSeek subsequently shattered it.

“DeepSeek is proving that innovation flows from many tributaries, and not a single divine, Trump-blessed spring in Silicon Valley. For years, Silicon Valley positioned itself as our shield against foreign AI supremacy. Then a Chinese start-up using open-source methods exposed the emperor’s new clothes: real innovation doesn’t need protection – it needs participation.

“The market crash was about trust … When you spend billions convincing the world that only your path leads forward, watching someone take a different route to the same destination becomes an existential crisis.”

DeepSeek’s rise raises important, existential questions for the Valley but also Australia, which has widely been perceived as an AI laggard as the likes of US and China sprint ahead.

Has Australia already been left behind? Or has DeepSeek’s rapid and impressive accomplishments, despite limited capabilities, mean something similar could yet emerge locally?

Eric Gao is the CEO of Boman Group, an institutional investor that manages funds of Chinese high-net-wealth individuals. For Gao, Australia currently lacks both the financial backing and deep talent pool needed to support a thriving AI industry.

“Australian corporations and customers tend to be hesitant early adopters, limiting domestic innovation and investment,” he says.

“However DeepSeek has opened the door for new AI start-ups worldwide to follow a similar playbook. We may soon see other ‘DeepSeeks’ emerging in different countries.”

Loading

Zac Altman – no relation to Sam – is an Australian start-up founder building Kantoko, a technology platform attempting to make ADHD care more accessible. He says Australia should have a clear plan to make every high school graduate AI-native, given students are already using AI for their assignments.

Australia does have world-class AI talent, he says, but the federal government is not moving fast enough.

“While the US and China can release $500 billion commitments overnight, Australia’s planning to deliver a plan later this year and won’t even have the budget until 2027,” he says.

“Start-ups are doing our part so it would be helpful to see the government step up and invest in our future. It’s embarrassing … Everyone else will be halfway through the race while we’re still picking which shoes to buy.

“I’d really hate to see us miss this golden AI opportunity.”

Others are more optimistic. For James Cameron, partner at one of Australia’s largest venture capital firms AirTree, the development should serve as a significant shot in the arm for Australian start-ups. AirTree has joined other investment funds in betting big on Australia’s AI companies, pouring millions of dollars into the likes of Canva, Build Club and Leonardo.ai.

Cameron says new open-source models like DeepSeek’s R1 will put downward pressure on the cost of building AI applications, meaning it should be easier for new Australian tech companies to flourish.

“Most Australian founders are focused on this application layer, where we see the greatest value being created as foundational models become increasingly commoditised,” he says.

“For start-ups vying to build the underlying foundational models themselves, DeepSeek shows that the cost curve for training these models may be dropping quickly. This paves the way for teams all over the world to build smaller, cheaper and more focused foundational models.”

James Tynan, partner at rival Australian venture capital firm Square Peg, agrees and says that right now, there is a unique opportunity for an Australian start-up founder to create the next global AI company.

“In over a decade of investing in founders building AI companies, we’ve never seen a wave of change like what we’re currently experiencing,” he says.

“These AI breakthroughs are driving a level of change akin to the birth of the internet … The race is on to create global winners.”

Australia’s new chief scientist Tony Haymet this week called AI a great opportunity for Australia, given the nation was “perfectly set up” to use renewable energy to run power-hungry AI data centres.

DeepSeek’s rise may indeed be a much-needed shot in the arm for Australia, which to date has struggled for AI relevancy, but it raises uncomfortable questions about the billions of dollars already invested by US giants to build and train their foundational models. Estimates suggest that Amazon, Microsoft, Google, and Meta are each spending between $US30 billion and $US60 billion annually on AI development.

The common belief was that building the most powerful systems required billions of dollars in specialised computer chips, but DeepSeek claimed to have developed its technologies with significantly fewer resources. The Chinese start-up says it spent just $US5.6 million ($8.9 million) on computing power for its base R1 model, relying on about 2000 less powerful H800 chips produced by Nvidia, which have been banned from export to China since October.

The events of the past week will bring the US’ export controls on microchips into sharp focus, given in this case they may have backfired, spurring China’s engineers to make do with less.

In an interview with Chinese media, DeepSeek’s founder Liang Wenfeng said that “AI should be affordable and accessible to everyone.”

Loading

Wenfeng, who is 40, has kept a relatively low profile despite his new role in leading China’s AI revolution. An engineering graduate who has never studied or worked outside of mainland China, Liang has been compared to OpenAI founder Sam Altman, who is effectively the face of AI in the US. DeepSeek grew out of Liang’s hedge fund, High-Flyer Capital, which uses AI to identify patterns in stock prices. The executive is assumed to own 51 per cent of High-Flyer.

“China’s AI can’t be in the position of following forever,” he said in a rare media interview in July last year.

“OpenAI is not a god and cannot always be at the forefront.”

Most of DeepSeek’s top researchers were fresh graduates from top Chinese universities, he said, and many were hired from outside the computer science field to diversify its abilities.

“More investment does not necessarily lead to more innovation,” he said. “Otherwise, large companies would take over all innovation.”

Maria Zhang, an associate professor at the University of Sydney, says DeepSeek is part of a new wave of Chinese start-ups defined by an ability to maximise limited resources.

She says Western observers have effectively missed the point. They often focus on China’s hardware limitations and US restrictions, she says, underestimating the ingenuity and resilience of Chinese companies like DeepSeek.

“Unlike their predecessors, who often prioritised quick profits under the pressures of venture capital, this new generation of Chinese tech entrepreneurs is defined by its dedication to foundational research and long-term technological advancement,” she says.

“Many of them are educated at China’s top universities – such as Tsinghua, Peking, and Zhejiang – and include top prizewinners in STEM competitions.

“And remarkably, even members of their management teams are often under the age of 35, bringing youthful energy and a fresh perspective to leadership. Having grown up during China’s rapid technological ascent, they are deeply motivated by a drive for self-reliance in innovation.”

Some in Silicon Valley remain calm despite the market bloodbath. James Alcorn is a partner at Silicon Valley venture capital firm Lightspeed, which is an investor in both OpenAI and fellow AI heavyweight Anthropic.

He thinks the claims about DeepSeek’s threat to US AI leadership are greatly overstated.

“Does this mark a profound shift in the distribution of geopolitical technology and power? Absolutely not,” he says.

“The kneejerk reaction that DeepSeek has commoditised AI; that hundreds of billions in capex at US labs is now rendered worthless; and that Western AI leadership has evaporated, is mistaken.

“The right way to interpret DeepSeek is twofold. Mind-bogglingly complex optimisation of AI’s status quo, and a blueprint of the new AI paradigm, where scale wins. DeepSeek models enable more use cases, which in turn creates more companies, more jobs, and ultimately, economic growth.”

DeepSeek’s position atop the app charts remains far from certain. The company has been fending off cyberattacks, and has been forced to temporarily limit user registrations due to its rampant popularity. Then there are the pressing questions about security. DeepSeek has quickly joined the growing list of Chinese companies causing unease for Western governments, including the likes of TikTok and Huawei, which have each been banned in various capacities.

DeepSeek’s AI model is censoring politically sensitive content and sending user data to Chinese servers: according to the company’s own privacy policy, it collects a user’s email address, phone number and date of birth, along with any user input including text and audio, then storing that data “in secure servers” in China.

When prompted to explain the 1989 Tiananmen Square incident, DeepSeek says bluntly: “Let’s talk about something else”, and when asking about Taiwan it evangelises “complete reunification” as “an unstoppable force”.

The federal government is yet to issue any formal guidance but Industry and Science Minister Ed Husic and Liberal senator James Paterson this week each urged caution.

Canberra, Washington and Beijing will each be watching closely. Whatever happens next will likely help determine the power dynamics of the rest of the decade, and beyond.