Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GLM 5.2 and the coming AI margin collapse

martinalderson.com

Post reply on HN
Print view View original

161–170 of 495 posts

Re: GLM 5.2 and the coming AI margin collapse

#161

Post by zmgsabst »

mgambati wrote: post #87
iknowstuff wrote: post #81

Earlier quoted context omitted.

I don’t exactly see orgs lining up to switch (and train) their employees between claude desktop and codex and whatever copilot is doing. There’s probably some inertia to those harnesses/integrations on top of the llms themselves.

The inertia is legal and financial. People are paying Anthropic through AWS accounts because the simple reason of not dealing making new contract and legal agreements is enough of reason of the inertia. But, eventually, I’m quite sure that AWS will also provide open models with those contracts without any inertia. Copilot is already offering Kimi. My company has a deal with Devin and they provide new models all the t…

AWS already supports Llama and GLM in its Bedrock service for hosted models.

They’re much cheaper to run, eg, Llama 3.3 Instruct 70B is 5-10x cheaper than Sonnet 5.

https://aws.amazon.com/bedrock/pricing/

Say you have 20% of usecases that require the more expensive model — but in 80% you could just use Llama instead of Sonnet (eg, for basic queries of a document). That saves 80% of that 80%, or 65% of your total bill!

That is the kind of “swap” that’s likely to occur in automated tooling as pricing pressure kicks in — “can you save 65% on our AI bill by switching Bedrock over in 80% of uses?”

Re: GLM 5.2 and the coming AI margin collapse

#162

Post by pizlonator »

vmg12 wrote: post #112
calebhwin wrote: post #80

Earlier quoted context omitted.

Hard disagree. Two LLMs with the same numbers on important benchmarks could have vastly different behavior in actual deployment. Not sure if as hard to switch as Excel Libre but still not "cheap and easy".

This is just another example of the bitter lesson. In a year a model will come out that will make none of these model specific optimizations you made matter.

Yeah

But the point is that at any moment, there is friction in switching

Re: GLM 5.2 and the coming AI margin collapse

#163

Post by argee »

fragmede wrote: post #128
argee wrote: post #22

Earlier quoted context omitted.

Indeed, as it gets more commoditized it feels more like swapping electricity providers. Who cares whether you get your electricity from IBM or the state of Texas? An amp is an amp.

That's an interesting question. What if we did care? Is this amp from burning dinosaurs or from the sun or from fission? What if we could tag power as coming from oil vs renewables? how would that affect our habits?

We care indirectly through cost. Hydroelectric, solar, or wind power are often among the cheapest electricity sources, for example. Beyond that, no we don't care. That's why if people want change we leverage policy on cost, via subsidies, surcharges, taxes, tariffs, what have you.

To a consumer, an amp remains an amp — so they get the cheap one.

Re: GLM 5.2 and the coming AI margin collapse

#165

Post by aussieguy1234 »

yomismoaqui wrote: post #83
aussieguy1234 wrote: post #82

I would not be unsurprised if the US govt steps in to prevent this. They'll do anything to stop China getting ahead in the AI race. There's the sanctions already implemented, next step might be giving these companies government funding, just like they do with military companies.

Good luck trying to enforce that outside of the US.

I posted this some time ago https://news.ycombinator.com/item?id=48759668

Singapore seized a mansion due to Nvidia chip smuggling. So there are some countries that will enforce sanctions.

Re: GLM 5.2 and the coming AI margin collapse

#166

Post by vachina »

maxglute wrote: post #62

How fast is glm 5.2 in western hosts? It's doing everything I want it to, but going through PRC host it takes like 5-10 times longer. Not sure if that is nature of modest or PRC computer infra/routing.

GLM feels faster and more reliable in my experience. Anthropic and OpenAI models would hem and haw or straight up timeout during peak times.

Re: GLM 5.2 and the coming AI margin collapse

#167

Post by onion2k »

cogman10 wrote: post #31
fny wrote: post #5

I'm not convinced raw costs matter: 1. Compute costs collapsed since the advent of Cloud and yet hyperscalers still have fat margins. 2. Many open source office suites exist yet none compete with the ubiquity of gsuite or office. GitHub, Slack are similar examples. 3. Both Windows and macOS dominate the home desktop space despite free alternatives existing for a long time. 4. Many formerly open source infrastructure…

Unlike all your examples, switching out an LLM is both cheap an easy. So easy that every 3 months or so new models are released and people grab them and start using them. The UX is the same regardless the provider. You send in a prompt, it spits back an answer. In all your other cases, the cost to switch is losing support and a difficult transition period. But in the case of LLMs, there was no support to begin with.…

Unlike all your examples, switching out an LLM is both cheap an easy.

Rolling out AI access in a large business is still hard, especially if you're trying to do it safely e.g stopping people throwing all your company data including user PII into a chat for productivity reasons.

It's more a staff training and guardrails issue than a choosing which LLM to use issue, but I imagine picking an open model like GLM would make it harder because the 'enterprise stuff' will be missing.

Re: GLM 5.2 and the coming AI margin collapse

#169

Post by vachina »

AgentMasterRace wrote: post #103

I don't think the writer has used top tier models very much. I have subscriptions to basically every provider, the difference between glm5.2 and opus is not even close, the gap is huge. raw benchmarks glm is impressive , but in practice these models are lacking so much. I had fable create a detailed implementation guide that explained how to implement everything in immense detail, it included all the libraries to use…

Opus is good but not consistently good. That’s a problem. I’m paying the same but not getting the same results.

Re: GLM 5.2 and the coming AI margin collapse

#170

Post by shaewest »

altmanaltman wrote: post #144
shaewest wrote: post #98

Earlier quoted context omitted.

The companies don't necessarily need to make back $1T, the investors do, and those investors don't require $1T in profit to do so, they need an asset worth $1T. Considering leaks suggest Anthropic's ARR would be $47B, that'd be a 20x valuation, but it wouldn't shock me if Anthropic doubles their revenue in the next year or two, in which a 10x revenue could easily support a $1T valuation, and boom there's your ROI, bu…

Wait what? Why are you measuring valuation as 20x revenue here? If its a public stock (which is what anthropic plans to be soon), it doesn't matter. Otherwise spacex's valuation should be... 18.67 billion x 20 by your logic but its current valuation is over 2 trillion dollars right now. ARR literally doesn't mean much in terms of how these companies are valued by investors and it will mean little when it goes public.…

I was arguing a 20x ARR valuation based on a simple 'potential' justification for $1T.

If I was to go further into that, I'd say that Anthropic has grown from $9B ARR Dec 2025, to $47B at their Series H.

I'd say that Anthropic is still a growth stock, so their $1T valuation is based on expected ARR/growth over the next year, and if we assume a double in ARR (justified by their supply constraints as proof of demand), that's 10x Valuation to revenue.

We could consider valuing by P/E, but they're in a growth stage so that's a waste of time, hence why investors focus on growth, and hence ARR growth is hugely important. If they managed $100B ARR, the same P/E as other top software companies by marketcap, they'd fit in that lineup.

If Anthropic was to hit $100B ARR, they be in similar ratios of ARR:Valuation to Meta, MSFT, Apple, etc. If you assume per token price reduces, and 'per intelligence' prices to reduce, which bullish investors would, you'd also assume a good margin over time, (which rumours appear to support for Anthropic).

Post reply on HN