3 ms·

It's a lot of money, but spread out over let's say 10 years it isn't crazy. Nvidia's market cap has increased by 1.2 trillion over the past year alone. That's

It's a lot of money, but spread out over let's say 10 years it isn't crazy. Nvidia's market cap has increased by 1.2 trillion over the past year alone. That's just a single firm. 10 years ago Apple's market cap was 469 billion. Now it's 2.86 trillion. Microsoft's was 350 billion. Now it's 3 trillion. Perhaps the amount assumes growth in tech will continue at a similar rate. In 10 years, Microsoft would be worth 30 trillion. It may sound crazy, but how crazy would it sound to tell someone in 2014 that Nvidia would be worth the price it is now? If the amount is describing the cost of the investment over a time period of that scale, it starts to make more sense, ambitious as it is.

jwilber 3y ago
These comparisons are still magnitudes off from raising 7 trillion. To be completely honest, the only thing I’m convinced after reading your argument is that you don’t understand how much money that is.
- atleastoptimal 3y ago
  Yeah I'm sure I, and likewise Sam Altman just don't understand how big the number 7 trillion is. If only he read the HN comments before saying that absurd number so he wouldn't be caught looking like a fool. I'm pretty sure he of all people knows the implications of different orders of magnitude of funding and investment, given he's become a billionaire via hundreds of investments. You're treating the 7 trillion dollars like it's a one time payment. Or that the number is just there to be left unchanged, not negotiated on, for no other purpose than a single payment to make chips under one firm and one ownership arrangement. Maybe it's all just a marketing stunt so that people talk about the scale of AI investment in the future more seriously? Until actual term sheets are being written out, all it is is just a number that exists in our heads. My first comment point still stands. This is the upper end of a range cited by a single source regarding a speculative venture that has yet to be confirmed or elaborated on by the source it purportedly comes from.
jjtheblunt 3y ago
> It's a lot of money, but spread out over let's say 10 years it isn't crazy. Yes, it most certainly is.
object-a 3y ago
I think there's a Valuation vs Expenses error here. Your interpretation compares market cap, which is a valuation, to the $700B per year that Sam is saying he'd need to spend. Over 700B is 3% of US GDP. It is also 2.5x Apples OpEx, 5x MSFT's OpEx, and 28x NVIDIA's OpEx. It's a pretty sizable number even if you spread it out.
- atleastoptimal 3y ago
  Which is why I don't think it would be 1/10 per year. I think he's imagining AI growing the economy at a massive rate even before the chip build up. The later years of higher growth and higher market expectations would carry the higher percentage of the total costs as thinks ramp up.
chefandy 3y ago
> It's a lot of money, but spread out over let's say 10 years it isn't crazy. Friend, that would would be a bit over 1.3 million dollars per minute for ten years straight. I imagine anyone who came up with a number like that preemptively budgeted quite a bit extra to feed their cocaine habit, but that much money could single handedly fund US Social Security for at least 4 years, including inflation.

jwilber 3y ago

These comparisons are still magnitudes off from raising 7 trillion. To be completely honest, the only thing I’m convinced after reading your argument is that you don’t understand how much money that is.

atleastoptimal 3y ago
Yeah I'm sure I, and likewise Sam Altman just don't understand how big the number 7 trillion is. If only he read the HN comments before saying that absurd number so he wouldn't be caught looking like a fool. I'm pretty sure he of all people knows the implications of different orders of magnitude of funding and investment, given he's become a billionaire via hundreds of investments. You're treating the 7 trillion dollars like it's a one time payment. Or that the number is just there to be left unchanged, not negotiated on, for no other purpose than a single payment to make chips under one firm and one ownership arrangement. Maybe it's all just a marketing stunt so that people talk about the scale of AI investment in the future more seriously? Until actual term sheets are being written out, all it is is just a number that exists in our heads. My first comment point still stands. This is the upper end of a range cited by a single source regarding a speculative venture that has yet to be confirmed or elaborated on by the source it purportedly comes from.

jjtheblunt 3y ago

> It's a lot of money, but spread out over let's say 10 years it isn't crazy. Yes, it most certainly is.

object-a 3y ago

I think there's a Valuation vs Expenses error here. Your interpretation compares market cap, which is a valuation, to the $700B per year that Sam is saying he'd need to spend. Over 700B is 3% of US GDP. It is also 2.5x Apples OpEx, 5x MSFT's OpEx, and 28x NVIDIA's OpEx. It's a pretty sizable number even if you spread it out.

atleastoptimal 3y ago
Which is why I don't think it would be 1/10 per year. I think he's imagining AI growing the economy at a massive rate even before the chip build up. The later years of higher growth and higher market expectations would carry the higher percentage of the total costs as thinks ramp up.

chefandy 3y ago

> It's a lot of money, but spread out over let's say 10 years it isn't crazy. Friend, that would would be a bit over 1.3 million dollars per minute for ten years straight. I imagine anyone who came up with a number like that preemptively budgeted quite a bit extra to feed their cocaine habit, but that much money could single handedly fund US Social Security for at least 4 years, including inflation.