vannevar

They were, but then Anthropic got so much flack in the business press for omitting them publicly (while sharing them privately with potential subscribers) that the bad PR was as bad as the financials themselves.

The article raises some concerns about the CSP revenue (which could be gamed, since it's part of the circular revenue deals among AI labs and hypoerscalers), but it doesn't delve into the "organic" customer revenue. Fortunately, other sources have done this work and have revealed that 80% of revenue comes from just 1% of their customers, and VC-backed startups make up a large segment of that top tier, while also showing disproportionate growth. Many of these companies are backed by the same VC that back Anthropic.[1] So even the "organic" revenue growth is OPM, not cash flow from profitable businesses.

[1] https://www.wheresyoured.at/annualized-enron-rate/#ai-startu...

"Hyperscalers, by contrast, look like winners. Their cloud revenue (a stat Wall Street uses to measure AI demand) grew 50% in the second quarter. They have huge contract backlogs and the capacity to absorb higher borrowing costs."

And if that revenue was coming from end users, it would be great. But it isn't---most of it is coming from the very same debt that is the subject of the article. There's a huge difference in my financial outlook if my biggest customer is borrowing all the money they are spending with me vs earning it in the marketplace. In the latter scenario, my business is tied into the success of the my customer in the market (harder to grow), in the former it's just a question of how much money they can borrow (much easier to grow, especially if I'm Nvidia and I can guarantee their debt).

I see the business press make this fundamental mistake all the time, starting off by drawing a boundary around the AI industry as a whole and correctly assessing the staggering levels of debt, but then switching perspective to individual sectors and completely forgetting about where the revenue actually originates, and treating it the same as organic growth.

>...its just that there's so much demand, that people are paying these markups.

The catch is that the vast majority of that demand is from data center projects, not actual AI users. And data center projects are funded by private credit that is speculating about future demand (most of these centers will take years to build out). So a corporate user who would put the hardware into service now has to compete with a data center project that is hoarding GPUs for the future. If those GPUs end up being used AND the operative margins on AI SaaS are high enough, then the speculators may win. If either of those conditions are not met (and both look pretty shaky), then the data center bubble will burst and the GPUs will flood the market at pennies to the dollar.

Ha, now that's a word salad.^^^^^ Well done, I appreciate a well-crafted ironic troll as much as the next man.

Important context: this from a narrow study in Japan. And the explanation is pretty simple:

"The researchers found that the JCP's [Japanese Communist Party] open website and party newspaper were among the most-cited sources in the AI models’ recommendations. Unable to distinguish between editorially independent journalism and partisan content, the models treated the JCP content as a credible news source."

> is just bonkers word salad.

I think you're placing much too much emphasis on the "electricity" comment, which I admit was an overstatement. But the description of the economics driving data center deals is quite parseable for someone familiar with the financial structure of the AI industry.

As for the rest, if you point out the parts that you don't understand, I'd be happy to elaborate.

>AFAICT, this hasn't happened for any "cloud" compute company that has been in business for -say- at least five years.

Most cloud software has had moats. AI does not. And AI is eating software, which means those historical margins are going away for a lot of cloud software,too. That's what the SAAS-pocalypse is about.

>Assuming that you're not doing anything that needs exotic system administrators

Remember, these are also going to be AI.

>Assuming I haven't fucked up my math, [2][3] the cost of the hardware here is more than three times the cost of power.

This is a fair point, but it just means the COGS is higher, not that the margin is any better. I stand by the prediction that AI margins will be very slim.

AI compute is not the new oil. Demand for oil is very inelastic: it's very difficult to increase the thermodynamic efficiency of the physical processes that convert oil into energy. AI compute efficiency, on the other hand, is improving exponentially. AI will end up being just slightly more expensive than the raw electricity required to produce it.

Data center demand right now has much more to do with private credit's demand for new deals, since the binge they went on around the pandemic led to the accumulation of a lot of bad debt and depreciating assets. Data centers are a tool for extracting money from insurance trusts and pension funds to prop up private credit and create debt-driven artificial "revenue" for the hyperscalers and Nvidia.

It could be worse, you could have a grotesquely over-sized Triumphal Arch in Trafalgar Square, Number 10 could be decked out with gold cherubs and eagles everywhere, with a series of plaques honoring former Prime Ministers by insulting them, and Big Ben could be officially renamed "Big Andy".

Then you'd really know you've lost your mojo.

>Oh wait, I almost forgot, they don’t exist.

At this point, the Roadster doesn't exist either, at least as a production car. The sibling comment lists several potential competitors. My guess is that if the Roadster somehow draws a bunch of new pre-orders, there will be more by 2028.

You're right, this is a very fair point. But if you had a choice between two sports cars with similar performance specs and one had 400+ mile range, you'd probably be inclined to take that one, all else being equal. This is a highly optimized version of older battery technology, which will be even older by the time it goes to production; you'd think that a car with a rocket option would also include a cutting edge battery.

The new motor does seem like good tech. But not enough to overcome the other shortcomings, IMHO.

The "other side of that bet" would be betting that ordinary consumers will be driving it in 2028, which is what Elon claimed. A lot can happen in ten years---if Tesla ends up selling a clean-sheet sports car called something else by 2036, would that still count? Or are we talking about this specific model with these specs?

The specs of the base (non-rocket-powered) car are pretty mediocre: RWD-only with 320 mile range. That's not terribly impressive in 2026, by 2028 it will be obsolete on delivery. The styling can charitably be described as generic. I agree, I don't think they have any intention of making this car.

I think the event was a very Elon Musk way to say, we're not making a Roadster. Which makes sense, Tesla is leaning into completely self-driving cars and the Roadster as originally envisioned would be the opposite of that.

Sort of. This is not a production car in any meaningful sense of that term. They'll make a handful of them to sell, just to say they did. But it sure seems like the actual mass-production Roadster they've teased for years is dead.

I listened to Andreesen's appearances on Lex Friedman. He does not strike me as a guy who thoughtfully considers the evidence, but rather someone with a lot of hard opinions and more than a bit of narcissism. And while I think there can be constructive criticism of DEI, most of the anti-DEI sentiment is simply re-branded racism for people who don't like to think of themselves as racists.

>It's not really possible.

It's almost certainly possible, just not very likely. You have to understand that there were probably thousands of devs working on solo passion game projects during the same timeframe that are not widely known. This one hit a social percolation cluster and became wildly successful. Making a good game is a necessary but not sufficient condition for success.

Which is also kind of an admission that maybe we won't really need all these data centers.

I keep hearing this, but the retail price of AI compute keeps dropping, which seems counterintuitive for a commodity that is supposed to be in such tight supply. It suggests that AI demand is extremely elastic, and belies the urgency of deploying new data centers. AI is something that AI labs think everyone should find incredibly valuable, but the market seems to be saying otherwise.

Like Russia's efforts to aid Donald Trump in 2016, it's hard to tell how much actual impact the propaganda is actually having on the outcome. There are many, many legitimate reasons to oppose data centers, from the local nuisance they create to the energy they consume to their contribution to an already critical level of global warming.

But I think the most damning argument against them is that they aren't needed. The cost of serving models is dropping faster than data centers can be built, and that trend seems likely to continue---AI is such a new technology that we have barely scraped the surface in finding efficiencies. And most of the innovations in efficiency that have emerged have come from the smaller players, because the big AI labs believe they can simply scale their way out of capacity constraints. An increasingly troubling trend is that they are not scaling their way out with cash flow, but with massive amounts of borrowed money. Seen in that light, data centers are not only a physical liability, they could become a catastrophic economic liability.

I think we're well past the time when we should have created an actual public commons instead of pretending that a patchwork of private online services constituted one. The commercial internet sits on a foundation of attention and deception, neither of which is hospitable to truth. I'm not sure that nationalizing Google would change that.

>I think you may be confusing what "real" income means.

Median income is exactly what it sounds like, the median income.

>The deflator used to correct incomes over the years includes housing and education.

That may be the intent, but it is clearly not working, because housing, medical, and education costs have in fact risen as a percentage of income.

Maybe you're confusing the terms "income" and "disposable income"? The latter would presumably include the items you mention, but the former does not. The (unreferenced) numbers you cite are almost certainly income, not disposable income. Or perhaps you're assuming that all costs have increased only by inflation, which is not true.

Median housing cost in 1960 was $11,900, which was about twice the median income at the time. Today it is nearly $400K, which is over 4 times as much as the median income. Medical and education costs have similarly risen faster than inflation (and faster than wage growth). You're right that it's not that complicated, though.

I would take that bet. Let's start with housing, which has almost doubled as a share of income since 1960. (See e.g. [1]). Same thing with medical care (including insurance premiums, since more medical costs were out-of-pocket back in the 50s and 60s).[2] Groceries are a bit cheaper as a share of discretionary income, but only by a few percentage points.[3] And of course we're assuming no one goes to college, which wasn't true even in the 50s and 60s. College costs have soared since then.

1. https://ti.org/antiplanner/?p=18800

2. https://www.healthsystemtracker.org/chart-collection/u-s-spe...

3. https://ourworldindata.org/grapher/food-expenditure-share-fa...

I'm pretty sure that one of the ideas that Western societies still promote is the notion that a claim should be consistent with the evidence, and that if it is not, it is probably wrong. And if it's very obviously inconsistent with the evidence, it may in fact be crazy. And there's a ton of evidence that raising a child in 2026 is substantially more expensive today than it was 75 years ago.

>it will cost less as a proportion of people's incomes today.

Do you have any evidence for this assertion, or are you relying on intuition here? A cursory search yields a number of references similar to this one---https://reno.nyloffices.com/resources/articles/starting-a-fa..., which indicate that the cost of raising a child has gone up significantly, at least in the last 40 years. And many of these references do not include the cost of college, which has skyrocketed. I'm not sure what you mean by "bringing up kids the way they were brought up in the 50s and 60s." They certainly had housing in the 50s and 60s, and that single expense alone is probably enough to make today's kids more expensive.

>...the 'taxing progressively higher any property a person directly/indirectly owns after first N properties' are seriously very much needed...

I don't see it in the article, but the progressive nature of the tax is important. It makes it harder for large landowners to pass along the tax, since they must compete with smaller landlords. The net effect should also stabilize land appreciation, which will no longer rise with rapidly increasing wealth disparity as it has in the past couple of decades. And we won't need clunky, patchwork solutions like the ban on home ownership by private equity.

This is pretty much the explanation---the decline in alcohol revenue mirrors the rise in cannabis sales. The alcohol decline started in 2012, when the first states legalized marijuana.

https://www.visualcapitalist.com/cp/daily-cannabis-vs-alcoho...