The assumption here is that, without AI, none of that capital would have been deployed anywhere. That intuitively doesn't sound realistic. The article follows on with:
>In the last two years, about 60 percent of the stock market’s growth has come from AI-related companies, such as Microsoft, Nvidia, and Meta.
Which is a statement that's been broadly true since 2020, long before ChatGPT started the current boom. We had the Magnificent Seven, and before that the FAANG group. The US stock market has been tightly concentrated around a few small groups for a decades now.
>You see it in the business data. According to Stripe, firms that self-describe as “AI companies” are dominating revenue growth on the platform, and they’re far surpassing the growth rate of any other group.
The current Venn Diagram of "startups" and "AI companies" is two mostly concentric circles. Again, you could have written the following statement at any time in the last four decades:
> According to [datasource], firms that self-describe as “startups” are dominating revenue growth on the platform, and they’re far surpassing the growth rate of any other group.
I think it's more likely the assumption is you'd expect a far more diversified market. If we're really in a situation where the rational, good reasons move is to effectively ignore 98% of companies, that doesn't say good things about our economy (verging on some kind of technostate). You get into weird effects like "why invest in other companies" leading to "why start a company that will just get ignored" leading to even more consolidation and less dynamism.
But startups are thriving. That doesn't suggest decreasing dynamism to me. It suggests that there are abundant gains to be had by exploiting technological progress, and the legacy economy is not availing itself of these opportunities. A thriving tech startup sector is surely key to dynamism.
I'm just some guy with an opinion. I worked in startups for 20 years. Startups are called exciting or thriving or good bets because a tiny few are successful and even fewer are trying to compete with established companies. Capital is pumped into lots of little ones regardless of the technology de-jour or market opportunity. They generally don't last. Statistically, if you were an AI startup from 6 years ago, you're long gone and you made out with what you could scrape together on the way out. Startups are thriving by feeding off dreams of grandeur, with very few happening upon the right combination of personality, capability and market enough to last for a decade. Is that thriving or thrashing? Don't confuse the velocity of gambling with the volume of opportunity.
This metric has obvious problems, but 120/134 of YC's S25 batch are AI-based [0]. 90% is, I guess, better than 98%, but woof. So, depends on what you mean by "thriving", but if diversity factors in there at all then at least YC is proving you wrong.
If you are still using sleds after the invention of the wheel, you are making a mistake. If you are innovating with pencil and paper after the development of the computer, you're making a grave error. There was no virtue in still sending faxes in 2010.
Tech has to believe this. We're all victims of the Upton Sinclair "it's difficult to get a man to understand something when his salary depends on his not understanding it" thing.
A lot of tech is just straight up bad. The machine gun was bad. The nuclear bomb was bad. Social media was bad. TV was bad. Cars were bad. C++ was bad.
> If you are innovating with pencil and paper after the development of the computer, you're making a grave error.
This is a great example. Study after study shows that writing on paper helps retention and analysis vs. typing. Study after study shows reading on paper helps retention and comprehension vs. a screen.
> There was no virtue in still sending faxes in 2010.
You can send a document with an authentic signature. This might not seem like a big deal, but "wet ink" clauses prevent many pro-democracy agendas (voter registration, etc), and the good faith reason for them (the typical reason is voter suppression) is that documents are too easy to forge without signatures.
But also, were fax machines bad? Did they enable an even-more-sprawling bureaucracy? We're well past the point where "new tech = good" is even close to default true.
Even the acqui-hire route is ending, with BigCos just hiring the talent directly. I think this will be a drag on startup investment: why should I invest my capital just to build a throwaway resume builder for some AI engineers where I don't even get made whole at the end of it?
I agree, except that it seems dynamism is almost restricted to digital tech. I wish tech would spread its dynamism a little better into legacy industries, and give some productivity gains/disruption to those areas.
there's been plenty of disruption in traditional industries like retail, automotive, media, communications, etc.
part of the problem is that the remaining set of industries is pretty tough to make dynamic using technology simply because the explosive market size isn't there for various reasons. if you wanted to disrupt aviation, for example, a plane takes tens of billions of dollars to bring to market, and an airline requires outlaying billions in capex on planes.
1. People aren't going to take on risk and deploy capital if they can't get a return.
2. If people think they can get an abnormally high return, they will invest more than otherwise.
3. Whatever other money would've got invested would've gone wherever it could've gotten the highest returns, which is unlikely to have the same ratio as US AI investments - the big tech companies did share repurchases for a decade because they didn't have any more R&D to invest in (according to their shareholders).
So while it's unlikely the US would've had $0 investment if not for AI, it's probably even less likely we would've had just as much investment.
> it's probably even less likely we would've had just as much investment.
I doubt it. Investors aren't going to just sit on money and let it lose value to inflation.
On the other hand, you could claim non-AI companies wouldn't start a new bubble, so there'd be fewer returns to reinvest, and that might be true, but it's kind of circular.
Correct - that's why you'd put it in Treasuries which have a positive real return for the first time in ~25 years - or, as I mentioned elsewhere - invest it somewhere else if you see a better option.
Sadly, this is exactly what the administration unironically desires. We're arguably in a recession and Trump wants to speed it up to a depression, ransacking the US and making off with the money.
we 1000% need all kinds of wealth taxes. The money's been hoarded for decades and I don't think that money is going to the next generation when the boomers kick the bucket. Inheritance tax, ultra high income tax, taxes on stocks if possible. The government wants to defund food stamps and healthcare while giving the robber barons trillions in tax breaks. Something's gotta give.
Sadly I think a wealth tax would make us all poorer in the end. Most of the wealth held by the top 0.01% is in the form of equities/investments in various industries. Taxing this would literally just suck capital out of the economy. I think taxing real estate holdings is a good idea though (something like Georgeism).
A big dynamic in this is that the business models for two of the biggest players are at stake.
Google is facing a significant danger that its search advertising business is going to just disappear. If people are using AI to find stuff or get recommendations then then aren't using google. Why should a photographer continue to spend $200 a month on ads when their clients are coming from openAI? Especially when OpenAI is using the organic search results. Meta is facing the same sort of issue, if the eyeballs aren't on insta then the ad $$$ go somewhere else.
So if they have money, and can get money, they will invest it to protect their businesses - all of it.
MS and VC's are doing the opposite, they are investing with the idea of taking the attention that Google and Meta currently have, but they are also following the "I'm scared" signal that Google and Meta have put in the market.
Every major consumer facing AI player (besides Anthropic and Apple) is incorporating advertisements into their product currently, or exploring how to, to include OpenAI.
So I’m not as optimistic that Google’s business of advertising is in danger, instead it’s just transforming.
> 1. People aren't going to take on risk and deploy capital if they can't get a return.
This doesn't seem to align with the behavior I've observed in modern VCs. It truly amazes me the kind of money that gets deployed into silly things that are long shots at best.
When you think about all of VC being like 1% of a mostly boring portfolio it makes more sense (from the perspective of the people putting the money in).
> > Without AI, US economic growth would be meager.
> The assumption here is that, without AI, none of that capital would have been deployed anywhere. That intuitively doesn't sound realistic.
That's the really damning thing about all of this, maybe all this capital could have been invested into actually growing the economy instead of fueling this speculation bubble that will burst sooner or later, bringing along any illusion of growth into its demise.
If the economy in my life has taught me anything, it’s that there will always be another bubble. The Innovators Dilemma mentions that bubbles aren’t even a bad thing in the sense that useful technologies are often made during them, it’s just that the market is messy and lots of people end up invested in the bubble. It’s the “throw spaghetti at the wall” approach to market growth. Not too different than evolution, in which most mutations are useless but all mutations have the potential to be transformative.
I took their comment to mean that we could have given school children free lunch or implement universal childcare. Like actual useful things that would unlock tremendous economic value, but instead I was given the lying machine and told to make it productive.
Or all that money might have been churning around chasing other speculative technologies. Or it might have been sitting in US Treasuries making 5% waiting for something promising. Who knows what is happening in the parallel alternate universe? Right now, it feels like everyone is just spamming dollars and hoping that AI actually becomes a big industry, to justify all of this economic activity. I'm reminded of Danny DeVito's character's speech in the movie Other People's Money, after the company's president made an impassioned speech about why its investors should keep investing:
"Amen. And amen. And amen. You have to forgive me. I'm not familiar with the local custom. Where I come from, you always say "Amen" after you hear a prayer. Because that's what you just heard - a prayer."
At this point, everyone is just praying that AI ends up a net positive, rather than bursting and plunging the world into a 5+ year recession.
I agree, in any time in US history there has always been those 5-10 companies leading the economic progress.
This is very common, and this happens in literally every country.
But their CAPEX would be much smaller, as if you look at current CAPEX from Big Tech, most of it are from NVidia GPUs.
If a Bubble is happening, when it pops, the depreciation applied to all that NVidia hardware will absolute melt the balance sheet and earnings of all Cloud companies, or companies building their own Data centers like Meta and X.ai
And NVidia don't even make the GPUs! They're all made on a brave little island off the coast of China, while 7 huge US companies shuffle them around and exchange vast amounts of money for them.
Its also not fair to compare AI firms with others using growth because AI is a novel technology. Why would there be explosive growth in rideshare apps when its a mature niche with established incumbents?
I think the explosive growth that people want is in manufacturing. Ex: US screws, bolts, rivets, dies, pcbs, assembly and such.
The dollars are being diverted elsewhere.
Intel a chip maker who can directly serve the AI boom, has failed to deploy its 2nm or 1.8nm fabs and instead written them off. The next generation fabs are failing. So even as AI gets a lot of dollars it doesn't seem to be going to the correct places.
They're not going to get it. The political economy of East Asia is simply better suited for advanced manufacturing. The US wants the manufacturing of East Asia without its politics. Sometimes for good reason - being an export economy has its downsides!
Taiwan isn't some backwater island making low skilled items.
USA lost mass manufacturing (screws and rivets and zippers), but now we are losing cream of the crop world class manufacturing (Intel vs TSMC).
If we cannot manufacture then we likely cannot win the next war. That's the politics at play. The last major war between industrialized nations shows that technology and manufacturing was the key to success. Now I don't think USA has to manufacture all by itself, but it needs a reasonable plan to get every critical component in our supply chain.
In WW2, that pretty much all came down to ball bearings. The future is hard to predict but maybe it's chips next time.
Maybe we give up on the cheapest of screws or nails. But we need to hold onto elite status on some item.
> Taiwan isn't some backwater island making low skilled items.
Definitely not! Wasn't trying to imply this.
> If we cannot manufacture then we likely cannot win the next war.
If you think a war is imminent (a big claim!), then our only chance is to partner with specialized allies that set up shop here (e.g. Taiwan, Japan, South Korea). Trying to resurrect Intel's vertically integrated business model to compete with TSMC's contractor model is a mistake, IMO.
I think this is a gross oversimplification and an incorrect assessment of the US’ economic manufacturing capabilities.
The US completely controls critical steps of the chip making process as well as the production of the intellectual property needed to produce competitive chips, and the lithography machines are controlled by a close ally that would abide by US sanctions.
The actual war planes and ships and missiles are of course still built in the USA. Modern warfare with stuff that China makes like drones and batteries only gets you so far. They can’t make a commercially competitive aviation jet engine without US and Western European suppliers.
And the US/NAFTA has a ton of existing manufacturing capability in a lot of the “screws and rivets” categories. For example, there are lots of automotive parts and assembly companies in the US. The industry isn’t as big as it used to be but it’s still significant. The US is the largest manufacturing exporter besides China.
Indeed. Just now our kid's therapist told us they are moving out from current school district because some chemical plant is coming up near by. More than pollution it is the attitude that any kind physical product factory is blight on Disney-fied suburbia and its white collar folks.
>I think the explosive growth that people want is in manufacturing. Ex: US screws, bolts, rivets, dies, pcbs, assembly and such.
And the [only] way to get that explosive growth is robotics. That is the Post-Post-Industrial Revolution that we're stepping into - it is when manufacturing stops being separate from the knowledge-based economy and instead becomes a part of it aa a form of an output, specifically a physical-world form of output from the knowledge-based economy.
>The dollars are being diverted elsewhere.
The dollars are going in exactly right direction - AI. After LLM the companies like NVDA and Google are making next steps - foundational world models and robotics.
>Intel a chip maker who can directly serve the AI boom
Intel is a managers' gravy train - just like for example Sun Microsystems was 20 years ago. Forget about it.
>Intel ... has failed to deploy its 2nm or 1.8nm fabs and instead written them off. So even as AI gets a lot of dollars it doesn't seem to be going to the correct places.
The dollars go to NVDA instead of Intel. Seems exactly to correct place.
While the USA spent a $trillion on LLMs, China spent the past two decades mass making chip shooters and pick and place machines like above.
Your LLMs aren't even fast enough to process or compete vs 15year old technology.
Now USA has pick and place machines and other such equipment. But we are slower and we don't have nearly as many. And despite better software here in the USA out setup costs are far higher (jlcpcb orders are prewired up with premade libraries because of huge volumes of orders. USA has fewer orders and thus requires a human to setup each line).
>assumption here is that, without AI, none of that capital would have been deployed anywhere. That intuitively doesn't sound realistic
For the longest time, capex at FAANG was quite low. These companies are clearly responding specifically to AI. I don't think it's realistic to expect that they would raise capex for no reason.
>a statement that's been broadly true since 2020, long before ChatGPT started the current boom
I guess it depends on your definition of "long before," but the ChatGPT release is about mid-way between now and 2020.
As for the startup vs. AI company point, have you read Stripe's whitepaper on this? They go into detail on how it seems like AI companies are indeed a different breed. https://stripe.com/en-br/lp/indexingai
The sunsetting of research tax breaks would explain why they threw everything into this.
They also view labor as a replaceable cost, as most accountant based companies that no longer innovate act. They forget that if you don't hire people, and pay people, you don't have any sales demand and this grows worse as the overall concentration or intensity of money in few hands grows. Most AI companies are funded on extreme leverage from banks that are money-printing and this coincided with 2020 where the deposit requirements were set to 0% effectively removing fractional banking as a system.
>The assumption here is that, without AI, none of that capital would have been deployed anywhere.
With this recessionary behavior, it might not be that far fetched an assumption. I'm not sure where else it would flow outside of being hoarded up in assets if there wasn't this big speculation everyone wants to take advantadge of. Especially when you consider that there's so much money flowing into AI, but AI is not as of yet profitable.
That's it, it's unusual times and the AI boom could be covering things up. I personally completely divested from US businesses since the start of the year but I can understand the allure of wanting to ride that particular gravy train
I think the money is chasing growth in a market that is mostly mature. Tech is really the only hope in that situation, so that's where the dollars land.
“The assumption here is that, without AI, none of that capital would have been deployed anywhere.”
The question is could the capital be deployed somewhere more productive. If that was just used park it into REIT’s buying single family homes, would you be happier?
The capital would have, should have maybe, been deployed in stock buybacks and dividends. Investment already doesn't happen in the US, and that's an expected thing for a well-developed country with a trade deficit.
Derek Thompson is not well suited to this kind of work. He is much better suited to his usual lame, predictable, centrist, lukewarm commentary on tired political and social topics.
The assumption here is that, without AI, none of that capital would have been deployed anywhere. That intuitively doesn't sound realistic. The article follows on with:
>In the last two years, about 60 percent of the stock market’s growth has come from AI-related companies, such as Microsoft, Nvidia, and Meta.
Which is a statement that's been broadly true since 2020, long before ChatGPT started the current boom. We had the Magnificent Seven, and before that the FAANG group. The US stock market has been tightly concentrated around a few small groups for a decades now.
>You see it in the business data. According to Stripe, firms that self-describe as “AI companies” are dominating revenue growth on the platform, and they’re far surpassing the growth rate of any other group.
The current Venn Diagram of "startups" and "AI companies" is two mostly concentric circles. Again, you could have written the following statement at any time in the last four decades:
> According to [datasource], firms that self-describe as “startups” are dominating revenue growth on the platform, and they’re far surpassing the growth rate of any other group.