Bloomberg Opinion: Be Wary About the Rapid Rise of Nvidia

In her column last week, Palmi argued that NVIDIA is a classic case of the bigger they are, the harder they fall, and that other tech company leaders should be looking at what's happening with NVIDIA and wincing over how rapidly it's taken the AI world. Palmi, good timing to have you on the programme. Explain your thesis to me. Well, I think a lot of people look at NVIDIA as this kind of bellwether for a is success, but it's also a reflection a a lot of the hype in AI. You know, it's kind of profiting from the short term success of businesses who are buying its chips for their servers, cloud companies like Microsoft and Amazon. But if you look further downstream at the actual businesses who are buying the generative AI tools from the likes of Open AI or Microsoft Azure, we're seeing signs of discontent. There are businesses who are saying, and I've seen multiple surveys coming out just in the last few months saying that they are not really getting the productivity they were hoping for. They're not quite sure how to use it. There's been a decline in plans to spend on AII think there's been a decline from 93% last year to 63% this year, in spite of the fact that these tools are getting better. So I think a big reason for this is that these generative AI tools from the likes of Microsoft Open AI and Google have been marketed as general purpose tools, like a Swiss army knife of technology that's going to make your workforce more productive. But of course, these tools aren't necessarily general purposes. They're, they're certainly capable in some areas, but they make mistakes, they make hallucinations. There are issues with data security. Strangely, our whole perception of computers in AI in how it was kind of marketed to us by science fiction as being these kind of robotic fact based machines isn't really how it is in reality. They're actually very good at artistic endeavours and generating images and poetry and stories. Actually really not so good at generating facts that you can rely on. And this is something that businesses are realizing sometimes to their detriment. And so I think this has been this kind of reigning in of spending. And I think this we're starting to see that reflected potentially in the decline in shares in NVIDIA. A lot of investors, even if they can't do the math on valuation or correctly forecasting or predicting sales, they'll say, Oh, well, Nvidia's supply constrained, right? Demand right now for AI accelerators greatly exceeds their ability to supply. There was one other piece of your column that I found so interesting. We've both covered technology for quite a long time. And you make the point that historically in phases of technological progress, all companies have a very clear North star, something to work toward. But you also make the point that with AI, it's kind of become such a grand vision led by specific individuals that it's lost all meaning. What do you mean by that? Well, this whole race, this arms race for AI was sparked by two men, Sam Altman of Open AI and Demis Hassabis, the founder of DeepMind, Google DeepMind Now. And both guys were trying to create artificial general intelligence, which is this hugely ambitious goal to create a machine that can essentially do everything. It is more knowledgeable than humans. It can surpass our own cognitive abilities, and it has general knowledge, meaning it can do creative things. But it can also do mathematical calculations very well and better than humans. And their objectives were nothing less than curing cancer and solving climate change. There was a sense that if you had this machine, this almost God like machine that could solve everything in a general capacity, that it could fix all problems. And I think when you have a vision like that that's just so grand and then it trickles down into the marketing and sales channels of your tech company and then they're going out and trying to sell this to businesses. Then your end customers have this kind of sense that they're getting this tool that has this general purpose ability. They are, they're almost left with a sense of paralysis. I mean, what do you do with technology that can do everything? Where do you even start? So I think the mistake that some tech companies have made in marketing AI has been not necessarily in, in saying that the capabilities are too high because the, they're very capable, but it's been in marketing them as being general purpose because they can't do Everything Everywhere all at once. They, they can only do a few things very well. And that's why I think businesses need time when they buy these tools to experiment with them. You know, it's a little bit like with the mobile revolution, people, individuals who worked for companies brought in their smartphones and they told the IT people, can you just set up my corporate e-mail to my BlackBerry, to my iPhone? And they were sort of forced. Companies were forced to do that because people were doing their own personal experiments with them and using them as Productivity Tools for themselves. And I think right now, this kind of top down approach to let's force the entire staff to use these tools is just a recipe for failure. Because when technology is so new and cutting edge, it really needs time to percolate and for individuals to just sort of find the way and how these tools will work for everybody else.