Hacker News

I think the AI industry needs intelligent skeptics that keep the hype in check and ground us in reality.

But Ed Zitron is not it. Here's an example [1] of him fumbling on simple arithmetic. He's also perpetually bearish without any sense of principles on his message.

This is what he wrote in 2024 [2]

> You can fight with me on semantics, on claiming valuations are high and how many users ChatGPT has, but look at the products and tell me any of this is really the future.

I think the industry really needs someone better with principles.

[1] https://x.com/binarybits/status/2034376359909130249

[2] https://www.wheresyoured.at/never-forget-what-theyve-done/

Edit: here's another example https://x.com/blader/status/2031216372169191678

I get that people make mistakes but it really does seem like there are no principles behind the guy. It seems like he can write whatever.

Someone please correct my math.

The article says 240 Gigawatts of capacity is allocated for AI datacenters.

New York City draws about 10 Gigawatts in the hottest months of the year due to extra load from AC use.

So am I understanding correctly that these people want to foist upon the power grid 24 NYCs?

I'm at a loss as to how some of these projects got funded in the first place. Anyone funding these should have had the perspective to see that there isn't enough power for them. Anyone funding them should have had the perspective to see that by the time power could come online for even a significant fraction of them, the depreciation and interest costs should have murdered the company trying to do it, especially if their solution to that problem is the oh-so-21st century solution of "solving" the problem of losing money by levering up. It does no good to go out of business entirely in 2027 to make the phat buxx in 2030, which seems to be the best case scenario for this space as a whole.

The other question I have is... who exactly is doing all of 1. Using AI right now 2. Making substantial money on it or getting real value and 3. Capacity constrained? Who is actually going to productively soak up all this capacity? It seems to me that bringing all this stuff online can't really make things much cheaper than they are now because the fixed costs aren't going anywhere, and if anything, trying to jam so many projects through all at once just raises those fixed costs even higher. It's not like they triple data center capacity (and increasing AI capacity by, what, 10x? 20x?), stick them full of AI systems, and into that 10x+ greater AI capacity they can sell it at the prices they are now. Higher capacity would crash the selling price but the costs would be as high or higher than now.

I am at a complete loss as to how the numbers are supposed to work here. You can't build a company in 2026 on the economy and tech infrastructure of 2036 anymore than it worked to build a company in 1999 on the economy and tech infrastructure of 2019, no matter how rosy the numbers look on the projections based on conveniently ignoring the fact the company passes through "death" in a year and half. Everything promised in 1999 happened, but trying to artificially accelerate it onto Wall Street's time line burned money by the billions. I'm sure 2036 will have lots of AI in it, but you can't just spend money to bring it forward 10 years by sheer force of will. It has to happen at its own pace.

Something I heard a person say recently:

> Isn't it weird how there is no huge industry pushback on all this new AI datacenter power need, as there was about electrifying vehicles?

An analysis of datacenter commitments and GPU purchasing through how much power they will demand vs how much is available.

As someone who only has a passing interest, there isn't anything distilled enough in this article for me to comment on as the central point. Everyone seems to be reporting impossible numbers, and buying dramatically more hardware than they can install in a reasonable timeframe given the pace of the industry.

Very good points.

My current model for understand for how AI will scale out is that we'll move through the following choke points:

AI chip makers -> Data center infra and construction -> regional power companies

Right now we're firmly in the "AI chip makers" part of the expansion, with everything else in the beginning stages. AI is useful, but whether it's hyped or not, it's hard to deny that not being able to build and power data centers will impact how this plays out.

i see him brag about how long these astroturf joints are on bluesky. had been a while since i'd tried to actually scroll through one and it makes sense now: so much more space to spawn subscription promos. better offline indeed

Railroads, e-commerce, and AI - all useful, all were (or may be) credit/stock bubbles. Railroads however have a much better depreciation schedule than GPUs.

Pointed and excellent.

But I wouldn't hold my breath waiting for introspection from that camp. It seems that AI maximalists, like so many other players these days, see it as end-game time. There are no bounds or rules: pick a side, and go. And then eat the rest.

Sure, not everyone sees it this way. There are highly competent, human actors working in their joy toward a better way forward with all of it. But I don't think you'll find that spirit unbridled inside any profit-seeking corporation of any significant standing (though I would be happy to be proven wrong). If it existed there, it is being choked out by selfishness and survivalism.

And then there's Thiel and ilk waxing eschatological, adding a whole other layer to the scheme.

The article takes an odd turn in the second half and seems to veer from a very interesting deep-dive into how a lot of backlogged US data center production may correlate with GPU "slippage" via questionable resellers and GPU rental outfits to China

The lying is not even subtle anymore. The gap between the demo and the product has never been wider and people are starting to notice.

I don't think Ed has made a single correct LLM prediction, despite posting in a fury probably monthly since ChatGPT was released. Grifters gonna grift

Ah yes, now that the rails are all built what could we possibly do next?