Sam Altman’s assessment of the new Chinese AI chatbot spoke volumes – “impressive, particularly around what they’re able to deliver for the price”. The OpenAI chief executive remembered to add that his firm would be launching “better models” soon, but his praise for DeepSeek’s R1 version would seem to clear up one question for AI luddites: Is the Chinese product as good as the pundits are saying? Answer: if Altman is saying so, yes, it probably is.
Other mysteries remain, of course. Has DeepSeek really developed the thing on the comical shoestring budget of $6m? Or is the model running on Nvidia chips, just using fewer of them and of the older variety? If so, could it be good news for Nvidia in the long run that the AI game has new entrants? But does a chunk of spending by others on computing power for the AI arms race already represent wasted money?
Don’t expect quick answers to any of the above. For the time being, we can probably only say that, in terms of the stock market, the main significance of the DeepSeek development is that investors didn’t see it coming. But that is the nature of technology revolutions – they tend not to work in ways that are easy to predict. Everyone can sense something is happening that will change the world, but knowing where to place your investment bets is another matter.
Many are drawing parallels with the late-1990s dotcom bubble for understandable reasons. The level of market concentration within a handful of stocks is even more extreme than it was then; Nvidia alone, even after shedding nearly $600bn of value on Monday, is worth slightly more than all the companies in the UK’s FTSE 100 index added together. And the whole show is infused with the same belief in first-mover advantage – in other words, that the early leaders will scoop all the financial prizes.
But, as demonstrated a quarter of a century ago, real life doesn’t work out so simply. Fund manager Terry Smith last year assembled a strikingly long list of tech pacesetters that ended up as also-rans: Intel in chips; AOL in internet service provision; Nokia in mobile phones; Yahoo in search engines; Myspace in social media; and Research In Motion, the Blackberry people, in smartphones. It would be “a break with tradition”, he argued, if the market had discovered an ability to spot winners at the outset.
There is also the possibility, as Smith suggested, that “the adoption of AI may lead to a situation where everyone has it, so no one has any advantage.” The DeepSeek news would seem to nod to that idea. If “impressive” and cheap Chinese versions of AI are available on an open source basis, the cost of entry to the party could plunge. Competition would flourish, especially among companies developing applications for AI, undermining the prospect of super-normal returns.
A right-minded capitalist would have shot down the Wright brothers when they took to the skies at Kitty Hawk in 1903. Warren Buffett famously remarked about US airlines’ long habit of going bust and losing money for their investors. It would be absurd to predict that AI will turn into a similar tale of too much capital chasing too few returns from a society-transforming new technology. For starters, Nvidia made pre-tax profits of $58.8bn in the first nine months of its current financial year, which is more impressive than anything achieved by Smith’s list of eventual losers.
But the wider economics of the AI revolution remain guesswork and, post-DeepSeek, they are slightly cloudier than they were last week. That’s par for the course.
