 |
| The East Is Red: Journalists and commentators are referring to the sudden and disruptive arrival of DeepSeek as a second “Sputnik moment”. (Sputnik being the name given by the godless communists of the Soviet Union to the world’s first artificial satellite which, to the consternation and dismay of the Americans, they successfully launched into space in October 1957.) |
“DEEPSEEK” is a pretty good name for an artificially intelligent Chinese chatbot. Not as good as Douglas Adams’ “Deep Thought”, which, as all readers of A Hitchhiker’s Guide to the Galaxy know, was the name of the super-computer tasked with answering the Ultimate Question – i.e. the meaning of “Life, the Universe, and Everything” – and which came back, 7.5 million years later, with the less-than-helpful, “42”.
That said, the numbers associated with the arrival of DeepSeek are nothing if not compelling. One trillion, for example, is the number of US dollars knocked-off the collective value of the companies listed on the “tech-heavy” Nasdaq index. $US600 billion of that trillion is attributable to the plummeting value of America’s leading AI chip manufacturer, Nvidia – the greatest single one-day loss ever recorded by a US listed company.
To rub salt in the American tech-lords’ financial wounds, the Chinese creators of DeepSeek claim to have spent just $US6 million on perfecting their chatbot, which they are generously offering to the world free, gratis, and for nothing.
No wonder journalists and commentators are referring to all this as a second “Sputnik moment”. (Sputnik being the name given by the godless communists of the Soviet Union to the world’s first artificial satellite which, to the consternation and dismay of the Americans, they successfully launched into space in October 1957.)
The success of DeepSeek, like the success of Sputnik, throws into serious doubt a whole host of American assumptions. Not the least of these being the typically brash and overconfident American assumption that the United States enjoys an unassailable lead in AI and is, therefore, destined to be the prime beneficiary of the “Fourth Industrial Revolution” this crucial innovation will, inevitably, usher-in.
The future is, thus, an American “lock”. Too far ahead of its rivals to be overtaken technologically, the economic and military hegemony of the USA will remain incontestable – a “golden future” indeed!
“Meh!”, says China, “We shall see.”
Viewed from the other side of the Pacific Ocean, the future of the USA does not seem so golden. Yes, America’s big, powerful, and well-protected – but so was Goliath.
Beijing observes a blustering bully in the White House, throwing his weight around, and expecting the rest of the world to be intimidated. But the Chinese President, Xi Jinping, is not afraid of Donald Trump’s tariffs, seeing them as proof, not of America’s economic strength, but of its weakness.
Beggar-thy-neighbour economics will not keep America at the top of the global heap. A superpower that raises its drawbridge and hides behind its walls is no longer a superpower.
It is now more than two decades since the USA, in full control of the UN Security Council, thanks to the support of both the Russian Federation and the Peoples Republic of China, was able to unleash shock and awe upon any nation that dared to challenge, or simply got in the way of the “New World Order” Washington was fashioning out of its total Cold War victory.
But, the hyper-globalised world that American neoliberalism produced succeeded only in enriching China and hollowing-out the United States as a coherent social and political entity. Trump, far from being the apotheosis of American success is the tawdry emblem of American failure.
The only redeeming feature of the 47th president’s wrecking-ball administration is its determination to implement Trump’s populist agenda. In doing so, however, it will not make America great: “Promises made, promises kept” will only make America bankrupt.
Trump’s long and lurid career has left him supremely indifferent to the perils of debt. But, there is a big difference between a media-savvy paper billionaire shrugging off his financial failures and striking out in a new direction; and the world’s largest economy, fast approaching the limits of its ability to print its way out of irredeemable insolvency.
The timing of DeepSeek, like most Chinese moves, shows every sign of being carefully thought through. It speaks of a regime that is happy to demonstrate, rather than brag about, its technological prowess. Such seemingly effortless (and cost-effective!) competence is likely to play better than America’s punitive tariffs and sanctions, especially on those continents where Chinese capital and expertise are, increasingly, on display.
Perhaps the Americans should ask DeepSeek to predict the future of their once “indispensable” nation. The answer is, almost certainly, not 42 – or 47!
This essay was originally published in The Otago Daily Times and The Greymouth Star of Friday, 31 January 2025.
4 comments:
As a data and AI engineer, my two cents on DeepSeek is that there's a few questions that need to be answered before we can start crowing about the collapse of capitalism as we know it.
1. If their claims about cost, speed, and hardware hold up?
DeepSeek’s reported efficiency in training its model is impressive—but unverified. There’s also plausible evidence they may have used stolen OpenAI data. And while they claim to have trained on H800 chips, they’ve also hinted at stockpiling 10,000+ export-controlled H100s, which raises more questions than answers.
2. Can they match global standards on data security and privacy?
The relationship between DeepSeek and the CCP is murky. Given AI is a strategic priority for China, it's highly likely they have state backing—if not direct funding. Can they operate with the same transparency and privacy safeguards required of non-Chinese firms?
There's also the question of them model's reluctance to discuss issues that reflect badly on Xi Jiping or the CCP.
3. Is their business model actually sustainable?
It’s one thing to make a splash, another to build a long-term business. DeepSeek still has to prove it can compete without opaque funding or regulatory arbitrage.
tldr:
Which is more likely?
1) A quantum leap in AI efficiency, seemingly from nowhere?
2)Or an aggressive market disruption from a new player in a sketchy
jurisdiction with questionable practices that just happens to want to be a major competitor as a matter of serious national policy?
The jury's still out.
At the moment the AI market is overheated and overhyped – basically a bunch of tech and crypto bros high on huffing their own supply. Nobody with an understanding of the technology and underlying problems of building and scaling AI is taking this particularly seriously.
For now, best we can do is treat it like the cheaper, knockoff ChatGPT it likely is.
Not much to disagree with from me Chris. Iv'e always thought Trumps tariffs were a weakness in both his decision making and the US economy. The selfish ignorance of his supporters are happy to take any short term gain and are blind to the fact that long term these policies will undermine the American economy instead of fixing it. The AI thing, well its a bit like first to the moon wins all, however the Chinese will need their Bot to attain the expectations it's customers are hoping for or it's world prowess will be short lived. From the little Iv'e read there is a lot of development needed yet with this technology. As Chris suggests the US economy is debt ridden and it's hard to see how Trump will MAGA this any time soon so how patient will the US population be with him. He is already shouting out irrational statements blaming the previous administration for the LA fires and now the tragic collision of aircraft at Reagan National airport. Surely if he keeps this up his people will see him for what he is, a nasty political opportunist. Or maybe not.
As someone with 40 years experience in IT this is the third time I've seen over-valued US IT companies have their bubble burst. NVidia's valuations were never going to hold up.
As for Deepseek, I never trust reports out of China, on anything, as this guy now shows
New report by leading semiconductor analyst Dylan Patel shows that DeepSeek spent over $1 billion on its compute cluster. The widely reported $6M number is highly misleading, as it excludes capex and R&D, and at best describes the cost of the final training run only. x.com/wallstengine/s…
$6 million to train an AI? Nah!
I remember talking about China on this site maybe as much as 20 years ago, where I said that when the Chinese start putting as much money into R&D as they have into manufacturing, then the West would have to look out. I was laughed at at the time if I remember correctly because apparently the Chinese have a "bankrupt philosophy". That may well be true, but it does look like you don't need a solvent philosophy to inspire creativity. This whole thing may well be a complete bust of course we don't really know yet, but it's had knock-on effects on stock markets all over the world, and even the price of uranium.
According to Robert Reich who probably knows what he's talking about, American tech firms have become fat and lazy, and yet the CEOs of these companies purport to love disruption. Well I hope they like being disruptees rather than disruptors.
Post a Comment