What DeepSeek accomplished this week represents a watershed moment that historians may one day classify alongside Pearl Harbor or the Sputnik moment in the global AI arms race. Monday marked the worst single-day loss for any public company when chip manufacturer NVIDIA shed $600 billion in value instantly. In terms of national economies, it's as if the world woke up to discover a country with Sweden's GDP had vanished from the map.
But is the panic justified? The answer depends on where you start tracking this race.
Anyone surprised by China's achievement must remember that in the Western world, artificial intelligence only entered mainstream consciousness in the last two years (for most of us, only after exposure to ChatGPT).
However, in China, artificial intelligence has existed for decades, even if no one called it by that name. China, for better or worse, is a totalitarian state that tracks its citizens in ways that are difficult for Westerners to even imagine. A person wanting to travel by train from one city to another must pre-register with their ID and undergo a series of checks before and after boarding (and of course for flights as well); every citizen receives a "social score" based on their behavior toward authorities and other citizens, and based on this score they are either entitled to benefits or subject to restrictions. We all remember how during the COVID period, millions of Chinese were effectively imprisoned and under electronic surveillance day and night in their apartments for fear of violating strict lockdown orders. And of course, to this day China has entire populations under constant surveillance tracking their every move (the Uyghur minority for example). Foreigners visiting China are under physical and electronic surveillance from the moment they step off the plane at times.
How exactly do you systematically and electronically track one and a half billion people if not through models that monitor behavioral and speech patterns? In the West, this is done legally by collecting information from users who have consented and training the large models that enabled the major breakthrough in the field. In China, they simply take the information based on the unwritten agreement between citizen and state that allows the Communist Party to know everything about the citizen – including through street cameras that recognize your face and systems that analyze whether you're dangerous to society or not, down to whether you use imagery of "Winnie the Pooh" perceived as offensive to the Chinese president.
In the 1990s, when China began building an alternative internet that would disconnect its citizens from the West and allow government censorship of unflattering news, then US President Bill Clinton said it wouldn't work and described it as "trying to nail Jell-O to the wall, and good luck to them." But since then, China has actually succeeded in creating a virtual wall between its citizens and the West, and today Chinese citizens have no access to Google and Facebook, instead having a completely parallel world with applications like WeChat and Weibo and others that technologically match those in the West.
China's leap into artificial intelligence was likely just a matter of definition – instead of channeling the vast information authorities have into monitoring, security and censorship, resources were directed (at least partially) to building the AI engine that stunned the world.
So yes, the world was left jaw-dropped by the DeepSeek performance report. According to that data, China achieved the same results as ChatGPT and other AI generators using inferior means and a tiny budget of just a few million dollars despite China not having legal access to NVIDIA's advanced processors.
Regarding funding, we must assume the data coming from China underwent significant manipulation, and there was likely strong government involvement from the ruling Communist Party and access to NVIDIA's supercomputing processors through particularly unofficial channels.
Nevertheless, we cannot ignore that China has apparently managed to significantly narrow the gap between itself and the US and the rest of the world regarding AI capabilities, even if in the long run the US will likely find it much easier to break through to the next stage (which remains unclear what it will be).
While this isn't about war and casualties like Japan's attack on Hawaii in 1941, to some extent the established narrative is again that the world's fate hangs in the balance, even if this time it's not a battle for physical dominance but rather who will lead the revolution in the precious resource called artificial intelligence. But two things must be remembered: First, regarding AI, China started the race long before the West, even if they themselves didn't define it as artificial intelligence. Second, like many things China does, their ultimate goal is to integrate with the West (from a position of strength), not necessarily replace it.
Anyone who visits China knows that ultimately the Chinese government sees Western technology and the capitalist system as their way to preserve political stability (at the expense of civil rights of course). The Chinese government on one hand ensures the average citizen has the most advanced version of the internet, can get everything delivered home at the push of a button and can upload videos to social networks and communicate with friends, but on the other hand ensures the citizen is under surveillance at all times, and that artificial intelligence primarily serves the state (even DeepSeek according to recent tests refuses to cooperate with queries deemed offensive to Beijing).
Should the West fear China winning the battle over artificial intelligence? It's hard to know, partly because the West is currently battling itself over the right way to develop the field. As many said this week – including Elon Musk who is developing his own model at X – ultimately artificial intelligence isn't worth the investment if the models it trained on don't include data reflecting the real world. In other words, as long as China is under a regime that sees citizens as part of a collectively controlled from above, artificial intelligence, however good – will only manage to challenge the West – mainly in programming products and applications like DeepSeek. On the other hand, it won't be able to beat it in areas where artificial intelligence hasn't yet started a revolution.
The big question is whether China has essentially proven that the AI world in the West, carried by NVIDIA, is actually a bubble that doesn't justify investments of hundreds of billions of dollars (including in the Trump administration's flagship Stargate project). The relative silence from the Trump administration after the DeepSeek report may indicate shock, or at least an intelligence failure in Washington (including during Joe Biden's term) regarding the level of sophistication the Chinese reached in AI research. If President Donald Trump was looking for another excuse to raise the threat level against China, he found one, and here he'll likely gain sympathy from the world. Perhaps from this crisis the US can for the first time truly position itself in a leadership role together with the rest of the West against China, instead of isolating itself.
