As a professional historian operating "under cover" here, I very much appreciated this. The industry seems to me to be terrifically bad at weighing continuity and change, which I suppose isn't a surprise: it isn't part of the training.
Great article. The more growth you price in and longer your time horizon, the wider your range of possible outcomes needs to be - both to the upside and to the downside.
Investors are very optimistic around the top companies right now. Some of that is probably warranted, but I can't help but wonder if Nvidia is planting the seeds of its own destruction.
Assuming they keep their pricing power for any length of time, does the cost of the chips required to run AI make it impossible to make money from AI?
OpenAI recently said that at $200/month for each user, it still loses money on the pro plan. People will use AI for free, and they'll pay $20 a month for it, but there's no proof that enough people are willing to pay enough money to make OpenAI profitable.
Nvidia is basically dependent on a handful of companies to keep buying GPUs hand over fist for AI development. How long will investors allow them to keep piling money into a technology that doesn't look like it'll make actual profits any time soon?
History is full of new technologies that changed the world, but lost early investors piles of money: canals, railroads, the internet, the over-investment in fiber optic cables in the early 2000's...
I could be totally wrong, but I haven't seen much discussion around that sort of possibility.
My only struggle is that I've been reading about valuation danger since probably 2014. But I do think prudence will prove to be a virtue again at some point. When that is seems like a crapshoot, though.
Which is why I focus on the growth expectations, not valuation per se. If those expectations are met, then valuation isn't a problem. Some companies will grow exponentially in the next decade. But the odds are that they are not same ones from last decade.
As a professional historian operating "under cover" here, I very much appreciated this. The industry seems to me to be terrifically bad at weighing continuity and change, which I suppose isn't a surprise: it isn't part of the training.
Great article. The more growth you price in and longer your time horizon, the wider your range of possible outcomes needs to be - both to the upside and to the downside.
Investors are very optimistic around the top companies right now. Some of that is probably warranted, but I can't help but wonder if Nvidia is planting the seeds of its own destruction.
Assuming they keep their pricing power for any length of time, does the cost of the chips required to run AI make it impossible to make money from AI?
OpenAI recently said that at $200/month for each user, it still loses money on the pro plan. People will use AI for free, and they'll pay $20 a month for it, but there's no proof that enough people are willing to pay enough money to make OpenAI profitable.
Nvidia is basically dependent on a handful of companies to keep buying GPUs hand over fist for AI development. How long will investors allow them to keep piling money into a technology that doesn't look like it'll make actual profits any time soon?
History is full of new technologies that changed the world, but lost early investors piles of money: canals, railroads, the internet, the over-investment in fiber optic cables in the early 2000's...
I could be totally wrong, but I haven't seen much discussion around that sort of possibility.
Outstanding article, Mr. Peris!
My only struggle is that I've been reading about valuation danger since probably 2014. But I do think prudence will prove to be a virtue again at some point. When that is seems like a crapshoot, though.
Which is why I focus on the growth expectations, not valuation per se. If those expectations are met, then valuation isn't a problem. Some companies will grow exponentially in the next decade. But the odds are that they are not same ones from last decade.
In theory, there's no difference between theory and practice. In practice, there is. Watch $TSLA. Look out below!