Game's Up Folks
The Market Ticker - Commentary on The Capital Markets
Login or register to improve your experience
Main Navigation
Sarah's Resources You Should See
Full-Text Search & Archives
Leverage, the book
Legal Disclaimer

The content on this site is provided without any warranty, express or implied. All opinions expressed on this site are those of the author and may contain errors or omissions. For investment, legal or other professional advice specific to your situation contact a licensed professional in your jurisdiction.

NO MATERIAL HERE CONSTITUTES "INVESTMENT ADVICE" NOR IS IT A RECOMMENDATION TO BUY OR SELL ANY FINANCIAL INSTRUMENT, INCLUDING BUT NOT LIMITED TO STOCKS, OPTIONS, BONDS OR FUTURES.

Actions you undertake as a consequence of any analysis, opinion or advertisement on this site are your sole responsibility; author(s) may have positions in securities or firms mentioned and have no duty to disclose same.

The Market Ticker content may be sent unmodified to lawmakers via print or electronic means or excerpted online for non-commercial purposes provided full attribution is given and the original article source is linked to. Please contact Karl Denninger for reprint permission in other media, to republish full articles, or for any commercial use (which includes any site where advertising is displayed.)

Submissions or tips on matters of economic or political interest may be sent "over the transom" to The Editor at any time. To be considered for publication your submission must be complete (NOT a "pitch"; those get you blocked as a spammer), include full and correct contact information and be related to an economic or political matter of the day. All submissions become the property of The Market Ticker.

Considering sending spam? Read this first.

2024-04-29 07:21 by Karl Denninger
in Technology , 424 references Ignore this thread
Game's Up Folks
[Comments enabled]

Not in AI on a general sense, but in the "deep fake / generated identity" sense, yes it is.

There are a lot of people who have the belief that the fact that Microsoft has demonstrated this can be done with one PC-grade card (albeit one that costs about $2,000) that in turn means demand will be driven higher and firms like Nvidia will go (further) toward the moon.

Nope.

Why not?

Because this is demonstration that commoditization of the space is now in the final stages.  Remember that a 3060 can be had for under $300 and has about 1/3 of the performance.  That means that three of them are about $1,000 (assuming you have the supporting CPU power to drive them) and about half the 4090 price.

Do you see the gradient there?  Those two cards are about 18 months apart in terms of time which means in another 18 months the 4090 will probably be $300 and there will be a new "king of the hill" that is $2,000 and three times faster.

Maybe.

Here's the rub with that: I had a 1080TI in my desktop machine because I do some video editing.  That is, the editing and rendering required a dedicated video processor.  I bought the 3060 (which is what I have now) because the price for the performance improvement was reasonable.

Today I'd buy the 4060 which is about half the speed of that 4090 and is the same $300!  In other words for the same money I am now at one-half instead of 1/3d of the 4090s price which means I can buy two of them and for $600 have that $2,000 card's performance, assuming what I'm doing can be partitioned up across both cards.

Note also that the 4090 has a TDP (power dissipation) of 450W while the 4060 only requires 115W, so if you need two of them you're also saving half the power budget at the same time!

This is what always happens with technology and the wall comes when it does for the companies in the space.  I can buy a newer PC or laptop today that is "spec-faster" than what I have now.  But unless I have a use for the other machine I'd be crazy to do it as there's nothing wrong with the current one and from a user perspective I will not get any more productivity out of it; in the PC space the marginal gains for each new generation have dropped to near-zero with the exception of some niche uses.

In fact the last such upgrade I did on my desktop motherboard was several years ago.  I just did a "step" upgrade on the server processor here which gave me 50% more cores at the same speed for $100, in other words 50% more processing for almost no more money, other than the fact that I had to buy a better heat sink as the original, while perfectly-adequate for an 80W TDP at full power was marginal at 120W so there was another $50 involved to do that (although I still have the other one and its perfectly-suitable for another machine with a CPU in the sub-100W TDP range, should I ever need it.)  I did the same thing with the previous generation of board in that server (which had dual Xeons from a far-earlier era) in that as newer versions came out I was able to get two much faster ones in the older generation for almost no money and the cost of swapping them out was $20 for some CPU heat paste to re-apply to the top of each (which I still have, and is enough to do another half-dozen CPUs.)

Look at Intel's price chart all the way back to 2000.  During the peak crazy of "more-more-more!" they hit $75; where do they trade now?

Remember, this was "all Internet, all the time, everywhere and everything" which is a whole lot more penetration than "AI anything."

If you believed Intel was going to $500 your backside is quite sore, especially adjusted for inflation over that 25 years.

Now look at Nvidia at $850 and tell me what you think after it more than tripled in the last year.

I'm not saying they're a bad company -- they aren't, and I both like their products and use them.

But if you think they are not likely to trade sub-$100 in the next few years you are betting that somehow unlike every other commoditization cycle in the technology space this one will be different.