[Kon5ole]

It's hard to comprehend the scale of these investments. Comparing them to notable industrial projects, it's almost unbelievable.

Every week in 2026 Google will pay for the cost of a Burj Khalifa. Amazon for a Wembley Stadium.

Facebook will spend a France-England tunnel every month.

[peterlk]

I have been having this conversation more and more with friends. As a research topic, modern AI is a miracle, and I absolutely love learning about it. As an economic endeavor, it just feels insane. How many hospitals, roads, houses, machine shops, biomanufacturing facilities, parks, forests, laboratories, etc. could we build with the money we’re spending on pretraining models that we throw away next quarter?

[Kon5ole]

I have to admit I'm flip-flopping on the topic, back and forth from skeptic to scared enthusiast.

I just made a LLM recreate a decent approximation of the file system browser from the movie Hackers (similar to the SGI one from Jurassic park) in about 10 minutes. At work I've had it do useful features and bug fixes daily for a solid week.

Something happened around newyears 2026. The clients, the skills, the mcps, the tools and models reached some new level of usefulness. Or maybe I've been lucky for a week.

If it can do things like what I saw last week reliably, then every tool, widget, utility and library currently making money for a single dev or small team of devs is about to get eaten. Maybe even applications like jira, slack, or even salesforce or SAP can be made in-house by even small companies. "Make me a basic CRM".

Just a few months ago I found it mostly frustrating to use LLM's and I thought the whole thing was little more than a slight improvement over googling info for myself. But the past week has been mind-blowing.

Is it the beginning of the star trek ship computer? If so, it is as big as the smartphone, the internet, or even the invention of the microchip. And then the investments make sense in a way.

The problem might end up being that the value created by LLMs will have no customers when everyone is unemployed.

[lII1lIlI11ll]

>As a research topic, modern AI is a miracle, and I absolutely love learning about it. As an economic endeavor, it just feels insane. How many hospitals, roads, houses, machine shops, biomanufacturing facilities, parks, forests, laboratories, etc. could we build with the money we’re spending on pretraining models that we throw away next quarter?

This is a wrong way to look at it. The right way is to consider that AI investments generate (taxable) economic activity that your government can use to build "hospitals, roads, houses, machine shops, biomanufacturing facilities, parks, forests, laboratories".

[qaq]

Not many. Money is not a perfect abstraction. The raw materials used to produce 100B worth of Nvidia chips will not yield you many hospitals. AI researcher with 100M singup bonus from Meta ain't gonna lay you much brick.

[johnvanommen]

> How many hospitals, roads, houses, machine shops, biomanufacturing facilities, parks, forests, laboratories, etc. could we build

“We?”

This isn’t “our” money.

If you buy shares, you get a voice.

[mike_hearn]

FWIW the models aren't thrown away. The weights are used to preinit the next foundation model training run. It helps to reuse weights rather than randomize them even if the model has a somewhat different architecture.

As for the rest, constraint on hospital capacity (at least in some countries, not sure about the USA) isn't money for capex, it's doctors unions that restrict training slots.

[YZF]

It's not a zero sum game. We could build hospitals and data centers. The reason we are not building hospitals or parks or machine shops have nothing to do with AI. We weren't building them 2 years ago either.

[polski-g]

Google has zero expected build outs of "forests". They've never mentioned this in their 10k ever. There is no misallocation of Google's money from "forests" to datacenters.

[uejfiweun]

There is a certain logic to it though. If the scaling approaches DO get us to AGI, that's basically going to change everything, forever. And if you assume this is the case, then "our side" has to get there before our geopolitical adversaries do. Because in the long run the expected "hit" from a hostile nation developing AGI and using it to bully "our side" probably really dwarfs the "hit" we take from not developing the infrastructure you mentioned.

[mtrovo]

Remember the good old days of complaining about Bitcoin taking the energy output of a whole town.

[mjevans]

It has never _not_ been time to build all the power plants we can environmentally afford.

More power enables higher quality of living and more advanced civilization. It will be put to use doing something useful, or at the very worst it'll make doing existing useful things less expensive opening them up to more who would like those things.

[saalaa]

I'm a simple man, I just want these companies to pay taxes where they make money.

[WarOnPrivacy]

> I'm a simple man, I just want these companies to pay taxes where they make money.

The folks who bankroll elections work tirelessly to insure this doesn't happen.

[int_19h]

Especially when they make money off the free work of all the people who contributed to those training sets.

There really needs to be some kind of "commons tax" for that kind of thing.

[speleding]

Profits = revenue - costs. Costs include depreciations and write offs of investments, and capital costs (what they pay to service debt), before you get to a taxable profit for accounting purposes. It's going to be many years before they pay taxes, if ever.

They do have to pay Value Added Tax on their sales in many countries (all of the EU), but not in most of the US. (The basis for Trump's claim that the EU is robbing the US. Sigh.)

[fogzen]

It's incredibly sad and depressing. We could be building green energy, parks, public transit, education, healthcare.

[renewiltord]

We could, but perhaps people want AI data centers more than one more month of healthcare spending. Though it’s a tough thing choosing between 3.5 rail lines between SF and LA and LLMs.

[hunterpayne]

> building green energy

Fun fact, we have already spent about $10T on renewables and it still provides a very tiny amount of global energy. Learn about why before complaining about it in public. While you are at it, perhaps learn why health insurance is so expensive while also increasing the cost of healthcare. In many matters of public policy, lack of money isn't the problem. Its ignorance of what makes good policy that is missing and that isn't fixed by throwing money at problems.

[jiggawatts]

Sure, but the end-state of this isn't just chat bots! Bipedal robots are the real target application for this technology, and at least three big players have invested many billions each into base model training. The "GPT 2 moment" for robotics will happen likely later this year, next year at the latest. Then, the company that scales up from there to the equivalent of GPT 3.5 -- the first properly useful model -- can start selling androids by the tens of millions.

It'll be the next automobile, every well-to-do household will want one eventually. Every hospital, to assist/replace nurses. Every retirement home for the same reason.

Japan and China alone, with their ageing populations in dire need of nursing, will easily pay for the investment with that one use-case.

[bogzz]

Haters will say Sora wasn't worth it.

[jsheard]

Incredible how quickly that moment passed. Four months on it's barely clinging to the App Store top 100, below killer apps such as Gossip Harbor®: Merge & Story.

[mimischi]

Ok I’ll bite. Was it worth it? What have people missed that haven’t used it.

[askl]

What is Sora?

[baq]

Inflation adjusted?

[Symbiote]

Seems to be.

Facebook's investment is to be $135bn [1]. The Channel Tunnel was "in 1985 prices, £4.65 billion", which is £15.34bn in December 2025 [3], or $20.5bn at current exchange rate.

That is staggering.

(I didn't check the other figures.)

[1] https://www.bbc.com/news/articles/cn8jkyk78gno

[2] https://en.wikipedia.org/wiki/Channel_Tunnel

[3] https://www.bankofengland.co.uk/monetary-policy/inflation/in...

[jsemrau]

Its like we are transcending to a new state of the world.

[jaccola]

Not really your point but I think the skills to create these things are much slower to train than producing chips and data centres.

So they couldn't really build any of these projects weekly since the cost of construction materials / design engineers / construction workers would inflate rapidly.

Worth keeping in mind when people say "we could have built 52 hospitals instead!" or similar. Yes, but not really... since the other constraints would quickly reveal themselves

[ttoinou]

So, now you understand you can’t compare things with how much they cost