If enough climate systems collapse, lots of existing farms will no longer be viable. That means famine and migration, which means war, which means lots of death. I don't know anyone who thinks we'll see extinction (outside of possible "hothouse earth" scenarios, where we become a second Venus) but societal collapse is definitely on the table. Saying this is "not scientific" would just be you not understanding the science.

End humanity, maybe not. Causing very painful social changes, however, is absolutely on the cards.

"Climate conspiracy"? Like you, mean, the conspiracy of climate scientists to publish facts to the best of their understanding?

I don't know what exact strawman you're arguing against, although I'm sure you can always find some idiots saying something like what you say. But scientific consensus has long been that it will lead to increasingly extreme weather and mass extinction, which we seem to be on track for. Of course, we can't know for sure what the consequences are untill we do the experiment, which in this case means potentially destroying large sections of the biosphere and living with increasingly destructive weather patterns. Surely that risk is worth at least legislating that hyperscalers need to spend some of their billions on solar panels?

> The left wing version of climate conspiracy is that climate change will end humanity itself.

That is unrealistic, but also not what a conspiracy is. It’s also a red herring, as nobody serious is claiming that. They are talking about civilisation changes, which we are already seeing.

At some point you have to engage with the arguments besides shouting "no, it’s you".

No idea what you're talking about. My local utility lit up 100MW of solar over the last year alone. Everywhere I look is doing the same.

Sure you can. The datacentre builders just don't want the (fairly modest) extra expense to do so properly. Obviously some preparatory work is required before dumping a lot of extra capacity into the grid.

My state allowed the power utilities to charge a modest fee ($10,000 to $100,000, depending on project size) before a yet-to-be-built data centre could demand a large amount of electricity. The amount of planned data centres went down by an order of magnitude. The truth is that most of the data centre builders (not all) do not want to be responsible citizens and are simply extracting value and wealth from other people, including from the power utilities and grid operators.

I know that market, and people for that matter don't care, but the environmentalist in me questions the word "needs" in the context of using AI.

GLM 5.2 isn't quite modern Opus tier, as seen in this comparison where Opus 4.5 scores 4/5 on some coding tasks where GLM 5.2 scores 0/5: https://www.tryai.dev/blog/gpt-5.6-build-off-12-models But yes, GLM 5.2 is cheap.

But the real standout on price is DeepSeek V4 Flash, which competes, more or less, with models in between Sonnet and Haiku. From third-party providers, it costs around $0.09/M, $0.18/M out, compared to $3M/in, $15M/out for Sonnet and $1M/in, $5M/out for Haiku. To get the price of DSv4 (Flash and Pro) so low, DeepSeek did a lot of innovative optimization work that will likely show up in other open weight models in the future.

> There isn't much evidence at all that inference is "subsidised" (and by whom?)

None of the big providers are profitable. It’s subsidised by overly enthusiastic VCs.

> In any case, you can go buy a MacBook Pro M5 48GB or an AMD R9700 and run Qwen 3.6 35B-A3B (a very capable model) and the only "subsidy" is you plugging it in

Right, people could. But they won’t, because that’s a bloody expensive computer and they don’t need that to ask ChatGPT. That war is lost already.

Subscription to the big players’ services would need to increase massively for that to happen. And the computational cost is only part of the problem; these models also eat a lot of storage and RAM, which is not exactly getting cheaper.

I agree with using smaller models, it's just that the majority of people I know feel like they need the biggest, beefier, behemoth model possible (with the longest thought setting) and consume much more than necessary when a flash or smaller model would be OK. I would also like to be able to use a smaller model, but given ram prices I would have to sell a kidney to buy ram now

> There isn't much evidence at all that inference is "subsidised" (and by whom?)

Well... why else would the major providers now tighten the screws on per-token pricing?