DeepSeek, Deep Six…Head Fake?
Yesterday, the growth and technology markets were roiled by the announcement that DeepSeek, a Chinese tech company, created a large language AI model using less sophisticated technology and costing single-digit millions. If this is true, how can Nvidia and other tech darlings be worth so much? Should they be worth so much?
As an analyst, and a bond analyst at that, I approach things with healthy skepticism. What is different now? Is this new product or service really the disruptor it is purported to be? Those types of questions enter my head. When I read about DeepSeek yesterday, I harkened back to my father purchasing a Texas Instruments calculator in the 1970s for about $100, if memory serves me correctly. It was far less sophisticated than the calculator that comes standard on any smartphone, but it was new technology and that usually carries a pretty sporty price tax. For those of us old enough to remember when VCRs came out, they were priced at about $700 dollars. Sony had the likely better technology called Betamax, and that cost even more. Ultimately, VHS (I cannot even remember what that stands for!) won out and by the time DVD players came out, VCRs were priced at far less than $100.
Technology, once created, will engineer its own obsolescence when the next generation of that particular technology is launched. Further, there will be improvements that will change the way certain things are consumed. Think 8-track tapes to cassettes to CDs and now digital music. Apple made a fortune with iTunes until Spotify and Pandora changed the game. I think that AI will be exactly the same. The primary driver will be expensive because of the cost of computing power and all of the coding that is needed in order for the AI models to work. Once built, the copycats will offer it cheaper and the innovators will earn the next fortune by taking the AI concept to a new level, either faster, cheaper, or broader in scope.
Last week, I read an interesting quip about AI. The columnist posited that AI is not going to take your job. But someone in your field who knows how to leverage AI is going to be the one to take your job. Ultimately, AI is a tool. It can solve complex problems; it can learn and help drive output based on queries and input. But it is just a tool. I always come back to the old adage that trees do not grow to the sky and when investing in stocks taking a gain, trimming winners, and moving on to new ideas, is a better way to make money than putting all of your eggs into a single theme or single stock. I am not predicting the end of Nvidia and the AI darlings. I am saying that technologies improve and either Nvidia will take the idea to a new level, or another competitor will.