DeepSeek forces the world to face its end game as China takes the lead
In this brave new world, whoever controls AI wins. After this week, as Cold War 2.0 took a nasty turn against the West, it appears China is frighteningly close to doing just that.
The world changed this week. The release of the Chinese artificial intelligence DeepSeek chatbot wiped nearly a trillion US dollars off American technology stocks. But it did something much worse than that. It demonstrated that in one of the few areas of hi-tech where it was thought the US had a decisive advantage over China, it has not much advantage after all.
Cold War 2.0, a kind of a virtual Cold War fought out mainly in the shadows and in cyberspace, just took a nasty turn against the West. Russia’s Vladimir Putin claims that whoever controls AI controls the future. Donald Trump said the DeepSeek release was a wake-up call that should cause US companies to have a “laser-like focus” on winning.
That ain’t the half of it. In time, AI will affect every part of life. Medical cures and treatments. The management of critical infrastructure. The way you relate to all your devices. The way businesses run, not just their inventories and supply lines, but their routine work. Computers already do a lot of routine work. AI can perform much more complex tasks. AI will end up being part of every business, every organisation.
And, of course, AI will be central to militaries.
So whether Washington or Beijing leads in AI matters profoundly in ways we can hardly yet imagine.
But wait, there’s more. Much, much more.
The other technology Washington thinks it’s ahead in is quantum computing. AI is in its early stages of real world application. It will grow much more powerful. As yet, the world doesn’t have reliable quantum computing. Such a computer uses quantum mechanics to speed up processing beyond what we can even imagine now. Your smart phone has more computer capability than Apollo 11 had when it landed on the moon. Quantum computing will make your smart phone look like a Neanderthal cave tool.
Combine AI with quantum computing, and the results will explode our brains if not our cities. Militaries now are devoting a lot of attention to how birds and insects fly in swarms, staying in broad formation.
With even a feeble imagination you see swarms of relatively cheap but deadly drones, combining AI and quantum computing, overwhelming the defences of any ship, base or building. AI-powered drones that can work out what the defence will do. Quantum computing that reacts fantastically faster than any human being. Think of hypersonic missiles similarly empowered.
In the AUKUS Pillar Two agreement, the US, UK and Australia agreed to work on six key technologies: artificial intelligence, quantum computing, electronic warfare, advanced cyber, hypersonic missiles and undersea capabilities.
Beijing, Moscow and others are doing the same.
Many observers (including me) noted that DeepSeek recalled the dramatic first Cold War Sputnik moment of 1957. Communist Russia put a satellite into orbit before the US. It shocked and distressed Americans, but also galvanised them. They won the subsequent space race, the associated technology race – and, eventually, the Cold War.
So here are two early, curly questions:
Will the West, specifically Washington but also its allies including Australia, galvanise an effective response?
And what if Beijing gives us a new Sputnik moment in quantum computing?
Silicon Valley thinks it’s ahead in quantum computing and will be the first to launch and commercialise in the next few years.
But we don’t really know where the Chinese are up to in quantum. What if we get a series of Sputnik moments?
Way back in 2015, Beijing announced its Made in China strategy, under which it wanted to replace the US as the dominant force in hi-tech by, oh gosh, 2025.
Justin Bassi, head of the Australian Strategic Policy Institute, points me to ASPI’s Critical Technology Tracker. ASPI established this in 2023 and it’s an intensely useful tool (incidentally, isn’t it bizarre that the Albanese government apparently wants to muzzle, inhibit and ultimately destroy all this good work by ASPI?). It follows the most often cited research papers in 64 critical technology areas.
Bassi tells me the Chinese have now surpassed the Americans in 57 out of 64 areas. An ASPI paper concludes that the top five international institutions in machine-learning (a form of AI) are in China. Western firms once had a decisive advantages in electric vehicles, solar cells and smart phones, but all of these have been lost to China.
“What DeepSeek shows is that China’s trillions of dollars of investment in research are paying off in commercialisation,” Bassi tells Inquirer. Beijing approaches these technologies strategically and unites every national effort, commercial and government, defence and civilian, towards achieving dominance in them. Bassi thinks the West must similarly take a much more focused approach, with deep government involvement.
Says Bassi: “Too many Western governments have held on to the view that the free market will prevail. But the market has been overtaken. There’s a country which has bought the market, which has stolen the market, which has subverted the market. We keep on having wake-up calls and then we keep going back to sleep.”
Government has been central in Western technological advances. If the world is to have any chance of zero-emission energy it will need to expand nuclear. This energy source came about entirely because of allied efforts to beat Nazi Germany to nuclear weapons during World War II.
In 1958, Dwight Eisenhower created the National Aeronautics and Space Agency, partly in response to Sputnik. The US space program had immense technological spin-offs for the US and the world, not least the creation of the internet.
Technological breakthroughs come in all shapes and sizes. But Big Government, properly deployed, is an engine for technological change, especially if it works in partnership with lots of smart scientists and engineers.
Former Labor leader and long-time defence minister Kim Beazley even drolly argued there was one irrefutable argument for socialist economic development – the Pentagon. In recent IT breakthroughs, the Pentagon played a minor role. Silicon Valley were the masters of the universe. With the election of Trump, and with their long disappointments in China, these companies seemed at last to embrace something of their natural national security role.
But DeepSeek, though only one app, has blown up the assumptions of that world.
DeepSeek is disruptive in a host of ways. The Chinese company claimed it trained the DeepSeek chatbot for just $US6m ($9.6m), a fraction of the cost of US models, to which its performance is comparable. Nvidia, which supplies the sophisticated chips that were thought essential for AI, suffered a market capitalisation drop of nearly $US600bn in response to DeepSeek. It was thought everybody needed Nvidia’s chips to get to technology’s cutting edge.
The Biden administration had banned the sale of those chips to China. DeepSeek claims it didn’t need those chips at all.
Now, many, many folks in the IT world take both those Chinese claims with a huge grain of salt. That $US6m figure might be as reliable as China’s published defence budget. And it’s not hard to imagine China getting hold of Nvidia chips despite the sanctions.
But whether they did it by pure innovation, or “grey trading” in chips, the end point is the Chinese produced AI of comparable quality to the American product. This now presents countless secondary problems. If the Chinese can sell AI much more cheaply than the Americans, it could come to dominate much of global AI use.
That’s disastrous for three clear reasons. One, the Chinese state will have access to any data its AI companies come across. It will hoover up an enormous amount of data on everybody. This data in aggregate, and sometimes individually, has strategic implications. Using AI in business is not like doing a Google search. It means you’re putting a whole lot of probably confidential information through AI, which can steal all that information.
Second, Chinese AI will always promote Communist Party propaganda. With DeepSeek it’s blatant in many respects. Ask it about the Tiananmen Square massacre or human rights abuses in China and you’ll get pure propaganda.
People might know that and discount for it. But what if you’re innocently asking for facts about Australian or US politics? You might get superficially plausible answers that swing towards Beijing’s purposes.
The Chinese, Russians, Iranians and others already make huge trouble in Western societies through their manipulation of social media. Russian intelligence was highly active in propelling early iterations of Black Lives Matter. If the Albanese government ever musters the courage to announce the site of an east coast nuclear submarine base, Chinese bots will work overtime to run anonymous, or misleadingly sourced, social media campaigns against it.
Third, the more our societies use Chinese-made internet-connected products, the more vulnerable we are not only to surveillance and propaganda, but in the event of any hostilities, the take-down of critical infrastructure.
Anthony Albanese’s government seemed as bewildered as anyone by DeepSeek, but Science Minister Ed Husic rightly foreshadowed national security concerns about it. He thought it would raise the same issues as Huawei, which Canberra banned from involvement in our 5G network, and TikTok, which Canberra has banned from any government-owned devices.
But still there is the strongest sense that we are making ad hoc, reactive, uncoordinated, one-off responses whereas Beijing has a clear plan and bends every sinew of national power to that end.
Alastair MacGibbon, of CyberCX and the former head of the Australian Cyber Security Centre, tells Inquirer we’ve moved from an era when things we frequently used were made in China, when China was the world’s factory. Now, many things we use are “controlled in China”. These new appliances are always connected to the net, need routine software upgrades which they get automatically, or remain in communication with the manufacturer. They’re all potentially vulnerable to manipulation by Beijing.
Says MacGibbon: “The issue in banning Huawei (from 5G) in 2018 was not concerned mainly with surveillance, but what if China one day just blocked our 5G through some software upgrade. The era of things controlled by China may have passed a tipping point in terms of the safety and survivability of our society. What if all the driverless vehicles just cease functioning? All these electric cars – they are a danger for us.
“Everyone just talks about price. The government doesn’t have the wherewithal to discuss it (the security issues) and just leaves it alone.”
MacGibbon, like many others including Bassi, believes we need to rethink fundamentally some of our ideas about technology: “The free and open internet is a fiction. China has shown it can cut its own people off from that but then spew its toxicity into it for everyone else. We fight with both hands tied behind our backs.”
MacGibbon, who knows these issues as well as any Australian, has reached a profound and sobering conclusion: “We have to have two internets – theirs and ours. That is a heretical concept.”
It certainly is. It contradicts every happy assumption and glad, optimistic cliche ever uttered about the internet. In the current US discussion about TikTok, MacGibbon points out that ownership really is a furphy, so is the idea that American information from TikTok is stored in America and thus safe from Chinese intelligence authorities, or that Australian information might be safely stored in Singapore. “The question is – who gets access to the information? The engineering, and the algorithm, will still be in China.”
MacGibbon believes we should clearly and publicly distinguish between different countries in our regulations. Suppliers of concern should be restricted – and, in some areas of technology, banned outright. The problem is that governments are typically scared of offending China.
Everything depends on the US. If Washington bans a Chinese technology, its allies feel they can do the same. As every new Chinese device or commodity with any kind of security connection comes along, the US and its allies go through extended political misery before banning them, at least from government. Australia banned Chinese-made drones from the Defence Department, ripped out Chinese made surveillance cameras from Australian government properties, banned Huawei from 5G, banned TikTok from government devices. But the security vulnerability which private citizens face from such devices is huge.
This isn’t just from the Chinese state. Cheap technology is easily hacked by anyone. There is a roaring business on the dark web of vision from hacked children’s bedroom cameras. On the dark web you can be specific in requests, asking, say, for continuous vision from the bedroom camera of a two-year-old girl of Asian background.
But the threat from the Chinese government is, of course, much greater. Former FBI director Chris Wray testified to congress that the Chinese had placed “malware” in critical US infrastructure such as water utilities, oil and gas pipelines, power grids and much else. This was designed to knock out such infrastructure in any conflict.
Senator James Paterson, the opposition’s home affairs spokesman, agrees with MacGibbon on the need for fundamental change, saying: “We are connecting far too many things to the internet with no serious security.”
The sheer range of Chinese cyber attacks, malware insertion, systematic hacking attacks, data hoovering, information manipulation, phony social media accounts used for activist purposes, means that effective action will need to be large scale, co-ordinated and Washington-led.
Says Paterson: “The landing point we’ll probably end up with will be a bifurcated tech world, one US-led and one China-led. That’s where we’ll end up but it will take leadership and time to get there.”
Mike Pezzullo, the former secretary of the Home Affairs Department and before that a senior Defence official, sees the enormous military consequences of the new technologies: “Quantum computing combined with advanced AI will rule the world.” He nominates weapons such as underwater unmanned vehicles, empowered by AI and quantum computing, and hypersonic missiles similarly enabled. Potentially most dangerous of all: biotech. AI and quantum computing could lead to cures for many diseases, but as Pezzullo speculates, they could “also develop deadly viruses preloaded with an antidote so it only affects certain peoples”.
This is a Brave New World. It’s vital for us, and for human civilisation, that the US and its allies dominate these technologies, or at least stay equal with any other player. The Americans are making some effort. But this was a bad week. There’s not much sign any US allies, assuredly including us, have much of a clue. You can rest assured, in Beijing they work on it day and night.
More Coverage
