One would think that those involved in innovation, above all walks of life should always be ready for… …innovation. However, shocking as it may seem, the tech giants at the forefront of AI development seemed to get blindsided by a Chinese minnow (DeepSeek) innovating a small scale, much more efficient, open source AI chatbot. Its metrics are impressive. It performed as well as, better, or close to on third-party tests of knowledge, reasoning, and particularly well on code and math, when compared to all the major AI tech companies.
The consequence of this was a sudden market fall in the Nasdaq and by extension in the S&P 500. Nvdia experienced the largest dollar loss in market capitalization, ever. At its session low it had lost almost $600M of its market cap value. Unsurprisingly, given its suddenness, this was followed by something of a recovery the next day, but was this perhaps the first sign of worry about the AI wave of promise which has contributed enormously to the lopsidedness that has been growing in the US market over the past two years since the unveiling of ChatGPT?
The funny thing about this market knee-jerk reaction is that the news of DeepSeek’s advance in efficiency is actually helpful in the development of AI and in scaling up its use and deployment across business in general (which should in the long-term be good for business). What the market got worried about, however, is not that AI itself is overvalued, but rather that the benefits of expensive exclusivity might be. Put simply, if you can do AI more cheaply then why spend more?
When a small unknown company claims that it has built an equivalent AI model for 10% of what it cost Meta to build theirs (n.b. ‘claims’ – given the expenditure figures DeepSeek released, this has inevitably been met with much skepticism). Then market analysts rightly start to get worried if the hoped for return on investment on some of the recent AI hype might not be as good as expected.
Wariness of Deja-vu
Though it may not often look like it, the market does have memory, particularly of negative events. The AI wave has caused market watchers to look for potential parallels with the Dotcom era. As for instance we did, when we explored in our November commentary, Intel’s pre- and post-Dotcom travails in comparison to Nvidia’s current trajectory. We noted that the ultimate long-term benefit to business of the internet did not preclude initial over-exuberance as the bubble inflated, followed by excess regret and negativity with its bursting. Over the long-term though, tech stocks were a winner.
This awareness of a potential bubble, in itself can contribute to creating one, or alternatively to seeing one when it is not there. The interpretation in this present case, that an advance in technology is perceived as a threat to value rather than something positive, says more about investors’ wariness than the usefulness of AI. It should be remembered that sudden market jitters do not necessarily mean that there is a bubble or that it has begun to burst.
To return to the benefits of innovation, a matter close to our hearts at Plotinus. We have strongly argued the need for smaller company innovation in AI, not, in our opinion focused on the broader aim of Artificial General Intelligence, but rather on specific ‘tools for the job’ as for example, in our case, AI Trade Decision-making. This kind of development covers all industries and it requires companies to tackle solving precise, situational problems with built for purpose AI agents. So looking beyond the whales of the large tech components of the Nasdaq, there is vast scope for developing approaches that are focused on processing efficiency, like for instance our own use of small data to develop our AI models.
President Trump’s immediate response to the news about DeepSeek was interesting, noting that it was a wake-up call for American industry and that it was also a good thing because its development involved spending less money. Perhaps in the back of his mind, with his eye on the lookout for a bargain, he was thinking that the recently announced $500 Bn for AI infrastructure might be able to be done for ten percent of the cost! ■
© 2024 Plotinus Asset Management. All rights reserved.
Unauthorized use and/or duplication of any material on this site without written permission is prohibited.
Image Credit: Ozalpvahid at Shutterstock.