25 Comments

This was a great article! Brings back memories of those days as a fingerling in the industry. One could see a lot of it for sure but hindsight always seems clearer.

I’m curious if you were working at the time and lived through all the craziness?

Expand full comment

Hahahaha oh no, I’m turning 30 next year. In 1999 I was in kindergarten.

Luckily for me, I love research and reading, I read almost 5 books to get up to speed.

Expand full comment

Cool. I could not believe that you actually read Patterson and Henessey! If only more industry analysts would emulate something like that.

Expand full comment

Look man, I barely understood SO much of it. But I like reading, and I find books my consumption of choice. You can learn so much in a single unified way. Just finish the book!

Expand full comment

THANKS!!!! I forwarded this to multiple mailing lists I am on, like internet history an my new nnagain list which has ol farts like me on it: https://lists.bufferbloat.net/pipermail/nnagain/2023-November/000344.html

Expand full comment

thank you - i did my best to try to get the whole thing. Saw your post that I missed part of it - but it was hard to decide exactly where to start / end it all.

Expand full comment

really great piece of research! Best thing I have read all month. My nnagain list has some good folk on it, and also the internet history list has founding people on it going all the way back to the 60s, if ever you want to talk to original sources: https://elists.isoc.org/mailman/listinfo/internet-history

If you have not read it, the story of iridium is great: "Eccentric Orbits".

Expand full comment

I have the book! I have been meaning to get to it, sadly lots on my list.

Expand full comment

I would like to chat with you about what the economic effects of fixing "bufferbloat" might be like, one day.

Expand full comment

If ever you do, there are a few I know that dispute that version of iridium's history somewhat. I have been trying to figure out where starlink will go... (another mailing list of mine)

Expand full comment

I'm very late to the discussion, but I have been thinking about the current AI bubble. This was an incredibly fun read, but I think that a simple summary is super helpful. So, here is mine:

a. Supply was too high because everybody listened to the forecast given by the Fox sitting outside the investor hen house. I do believe Munger had 60 years of wisdom when he said, "Forecasts (by the company) is worthless." By the way, we threw out the financial models, and we would say the words "New Economy."

b. Overbuilding is concave in the 2000s (using Taleb's terminology) because not only because you don't use the stuff, but it is rotting fish. You never buy a PC and let it sit on the shelf because in 3 years, it can be replaced at half the cost at double the performance. This the same for telecom infrastructure. Overbuilding was a disaster.

Finally, the productivity data use in the writing is a tertiary point. As Doug writes, there was a boost, but at the time it was about computers. The failure up to this time by computer investment was pointed out by Robert Solow, and received lot of attention.

The thing that dominated at the time was Web Browsers and the feeling of social change, and tech would change the world. It is hard to describe the feeling of using a Web Browser to those that haven't known a world without web browsers. You would go to the office with broadband, and then be able to visit web cams around the world.

Netscape did a '96 IPO at $2.5B, which just was inconceivable.

My point is that you cannot subtract the "telecom" bubble from the dot.com bubble. They were both outgrowth of the social impact of web browsers that showed a massive change in the world to anybody that was around the time. With that written, there were some that tried to start to figure out an ROI. The common term was "frictionless transactions."

The friction was real, but solutions were a long time coming. You needed a massive changing in every API. Or you were a shopping site, but found out you needed to build a distribution center to sell over the internet. So, you couldn't find a good ROI, but still the stocks went up.

So we invented a new term called "the New Economy." I am ashamed that I used these words once or twice and thought it was a good idea. However, I never moved industries.....

I have been using AI agents, and the difference is I can see this scaling very quickly to replace real human beings. It has already massively impact the creative community. Right now we are at a college programmer level. This increase is not concave, but more linear until we are "overachieving." This is the #1 thing to watch. Anybody that says that "AI needs to prove itself" has no idea of the tools currently being used by programmers to enhance their productivity. but will it scale?

Expand full comment

What a breathtaking piece.

Just qq on Nvidia roadmap: "I expect each generation to be about a 3x increase in performance for 50% more price." - is this based on historicals or was there some analysis driving this?

Expand full comment

I think that's just the rumors going around, and a bit of using A100 to H100 initial release as a guide. Would guess gross margins actually get a bit worse fwiw

Expand full comment

"The fake statistic of the internet doubling every 100 days made it seem like the world was improving at an exponential rate"

Doubling once a year is still exponential rate and saying 57% annual growth is measly is just disingenuous

Expand full comment

I don’t think you grasped the point of this. It isn’t measly at all, but if everyone is putting down supply supporting 100s of percentage of growth when actual demand is much less, then you have a problem lol.

Expand full comment

Excellent, Doug. It’s still not clear to me how to monetize AI such that it merits such large caped buildouts. The cost of training are high and hyperscalers are realizing this now. Check out this WSJ that few are talking about:

https://www.wsj.com/tech/ai/ais-costly-buildup-could-make-early-products-a-hard-sell-bdd29b9f

Expand full comment

oh yes - trust me. We are spending for a business to come :wink:

Expand full comment

This was a fantastic read. Thank you for sharing your research.

Expand full comment

Great piece Doug. I hear the fiber buildout analogy a lot lately so this dive into it is very timely.

I did have one question. Where can I read more about the NVDA fan out attempts you mention. "Moreover, Nvidia is working hard to increase fan-out ability, meaning the number of GPUs you can have in a fleet." What's limiting it now and what kind of things are they doing to address it?

Expand full comment

It’s about scaling out. There’s software to split a LLM into pieces for parallel and then there’s a scaling limit of infiniabnd itself

Expand full comment

Ah, ok. Thank you for that info

Expand full comment

Great write up.

Expand full comment

Great stuff Doug, thanks!

Expand full comment

Really impressive work here

Expand full comment

🔥👍🏽🔥

Expand full comment