4 ms·

I don't know. The moat doesn't seem to exist. Probably safer to stay out of the race for now.

alansaber 25d ago
Stay out of the race sounds insane when AI is, and will be, a common commodity. China is getting great bang for their buck on AI training, why not the same for Europe?
aurareturn 25d ago
I’m still looking for no-moatists to logically explain why there is no moat in SOTA LLMS. That paper released by a lone Google employee 3 years ago still misguiding people it seems.
- throwa356262 25d ago
  My reasoning is that we have seen multiple small labs come out of nowhere and create impressive models.
  - aurareturn 25d ago
    Impressive doesn't mean sustainable. We're also seeing behemoths like Google and Meta both fail to keep up with OpenAI and Anthropic. Microsoft has given up on SOTA training. Loads of once promising LLM training companies are no longer relevant at the SOTA stage such as Cohere, Mistral. I've said this countless times but SOTA LLM market looks like a classic monopoly/duopoly market over time. Each generation requires magnitudes more resources to train and you can only compete if you've made enough money on your previous generation.
    - throwa356262 25d ago
      On the contrary, small efficient models using recent improvements should be more sustainable than huge frontier models that require billions and years to train
      - aurareturn 25d ago
        Small efficient models are commodities.
    - r4rf 25d ago
      "Impressive doesn't mean sustainable." The frontier labs at the current rate are also not sustainable. You make some incredibly stupid assumptions - like assuming revenue growth rate of Anthropic and OAI will continue on. Elon-Musk type stuff. lmfao.
      - aurareturn 25d ago
        Why is it not sustainable when Anthropic has already had a profitable quarter?
        philipwhiuk 25d ago
        It's pretty easy to use accounting wizardry to have one profitable quarter.
        aurareturn 25d ago
        So what makes their AI business fundamentally unsustainable?
        philipwhiuk 24d ago
        The obvious draw-down on resources being used by companies following the token price hikes? The fact that in the testimonials both frontier labs quote the same company (who I'm led to believe is still tokenmaxxing). The fact that we've not seen actual productivity benefits or new products resulting from people outside the companies using it and a lot of things that in most businesses would result in a lot of investor skepticism and concern. The fervent development and focus on 'the next big thing'.
- Zambyte 25d ago
  The vast majority of people simply don't need SOTA models. I run most of my programming on a single 7900 XTX with Qwen 3.8 27b. I'm more interested in scaling out to run more instances of the model in parallel, than I am in scaling up to better models.
codingbot3000 25d ago
"Probably safer to stay out of the race for now." What kind of attitude is that?

Stay out of the race sounds insane when AI is, and will be, a common commodity. China is getting great bang for their buck on AI training, why not the same for Europe?

aurareturn 25d ago

I’m still looking for no-moatists to logically explain why there is no moat in SOTA LLMS. That paper released by a lone Google employee 3 years ago still misguiding people it seems.

throwa356262 25d ago
My reasoning is that we have seen multiple small labs come out of nowhere and create impressive models.
- aurareturn 25d ago
  Impressive doesn't mean sustainable. We're also seeing behemoths like Google and Meta both fail to keep up with OpenAI and Anthropic. Microsoft has given up on SOTA training. Loads of once promising LLM training companies are no longer relevant at the SOTA stage such as Cohere, Mistral. I've said this countless times but SOTA LLM market looks like a classic monopoly/duopoly market over time. Each generation requires magnitudes more resources to train and you can only compete if you've made enough money on your previous generation.
  - throwa356262 25d ago
    On the contrary, small efficient models using recent improvements should be more sustainable than huge frontier models that require billions and years to train
    - aurareturn 25d ago
      Small efficient models are commodities.
  - r4rf 25d ago
    "Impressive doesn't mean sustainable." The frontier labs at the current rate are also not sustainable. You make some incredibly stupid assumptions - like assuming revenue growth rate of Anthropic and OAI will continue on. Elon-Musk type stuff. lmfao.
    - aurareturn 25d ago
      Why is it not sustainable when Anthropic has already had a profitable quarter?
      - philipwhiuk 25d ago
        It's pretty easy to use accounting wizardry to have one profitable quarter.
        aurareturn 25d ago
        So what makes their AI business fundamentally unsustainable?
        philipwhiuk 24d ago
        The obvious draw-down on resources being used by companies following the token price hikes? The fact that in the testimonials both frontier labs quote the same company (who I'm led to believe is still tokenmaxxing). The fact that we've not seen actual productivity benefits or new products resulting from people outside the companies using it and a lot of things that in most businesses would result in a lot of investor skepticism and concern. The fervent development and focus on 'the next big thing'.
Zambyte 25d ago
The vast majority of people simply don't need SOTA models. I run most of my programming on a single 7900 XTX with Qwen 3.8 27b. I'm more interested in scaling out to run more instances of the model in parallel, than I am in scaling up to better models.

codingbot3000 25d ago

"Probably safer to stay out of the race for now." What kind of attitude is that?