3 ms·

Anthropic is almost purely a model company. They own close to no data centers. If models are becoming commodities, and the main bottleneck is actually serving

Anthropic is almost purely a model company. They own close to no data centers. If models are becoming commodities, and the main bottleneck is actually serving them at scale, Anthropic does not appear particularly well positioned. If AWS had a ~60-80% margin for decades, I see no reason why inference can't have a ~60-80% margin for quite some time. The problem is, if costs continue to drop ~90% for the same level of quality every 18 months, demand is unlikely to grow 10x to keep the revenue stable. Who knows. Jevon's paradox. But the cost/quality is dropping too fast that it's hard for me to imagine demand keeps up long term to keep revenues (and profits) GROWING.

pu_pe 2mo ago
Anthropic is perceived as the leading AI company in the world. Even if they started selling inference alone, people would prefer to pay them a premium for it rather than figuring out how to operate opencode or other replacements. Add to that government contracts, enterprise and other markets, and I think Anthropic would be totally fine even if models plateau. Transistor count has increased exponentially for decades and so did demand for compute. I think we will see a similar phenomenon with AI.
- wookmaster 2mo ago
  My company has started asking people to convert to open code all the sudden to avoid vendor lock in with anthropic
xnx 2mo ago
> I see no reason why inference can't have a ~60-80% margin for quite some time. Same as anything with low margin: Viable competition and low switching costs. Many sysadmins lazily do 100% AWS because they don't know any alternatives. CFO's might enforce using more economical inference providers if the cost is even 10% less.
- onlyrealcuzzo 2mo ago
  > Same as anything with low margin: Viable competition and low switching costs. So why didn't this happen to AWS?
  - xnx 2mo ago
    High switching cost.
    - onlyrealcuzzo 2mo ago
      Now do Google
      - xnx 2mo ago
        Advantages: Lowest inference cost due to scale, experience, advanced datacenters and custom chips. Huge customer base. AI automation will make it much easier to switch from AWS to GCP. Serious organization that doesn't let it's product unknowingly hack companies. Disadvantages: Top model is slightly behind the frontier for now. Bias against using their products because it's not cool/trendy.
        disgruntledphd2 2mo ago
        Their big disadvantage is their terrible, no-good, awful product management (which has always been their issue).
keeda 2mo ago
> If models are becoming commodities, and the main bottleneck is actually serving them at scale, Anthropic does not appear particularly well positioned. That has been my thesis ever since Dario went on about how difficult it is to forecast capacity on the Dwarkesh podcast right about the time Claude started having 9's comparable to GitHub's. Even as an outsider, seeing all the cloud providers bemoan their lack of capacity and their growing backlogs, I could tell compute capacity was already the biggest constraint on industry growth and that it would be very hard to come by in the future... which also probably made it the only moat. And now we see Anthropic making very expensive deals with competitors to secure the necessary capacity. Claude likely still has enough momentum to make it worthwhile. But I can't believe Dario, probably the most AI-pilled of them all, would under-estimate demand so severely.
- disgruntledphd2 2mo ago
  > But I can't believe Dario, probably the most AI-pilled of them all, would under-estimate demand so severely. To be fair, this mess up actually increases my respect for him, as he focused on making sure that they could continue to run the business under reasonable assumptions rather than YOLOing compute into existence like OpenAI.
  - keeda 2mo ago
    My surprise is rooted in the point that he seems to be the biggest believer in AI there is -- to the extent the label "zealot" gets bandied about at times -- and he still underestimated the demand they would get. I would not have classified Altman as a bigger believer than Dario. My guess is OpenAI, having the first mover advantage and the bulk of AI traffic for a very long time, could gauge the trends much better than anyone else.

pu_pe 2mo ago

Anthropic is perceived as the leading AI company in the world. Even if they started selling inference alone, people would prefer to pay them a premium for it rather than figuring out how to operate opencode or other replacements. Add to that government contracts, enterprise and other markets, and I think Anthropic would be totally fine even if models plateau. Transistor count has increased exponentially for decades and so did demand for compute. I think we will see a similar phenomenon with AI.

wookmaster 2mo ago
My company has started asking people to convert to open code all the sudden to avoid vendor lock in with anthropic

xnx 2mo ago

> I see no reason why inference can't have a ~60-80% margin for quite some time. Same as anything with low margin: Viable competition and low switching costs. Many sysadmins lazily do 100% AWS because they don't know any alternatives. CFO's might enforce using more economical inference providers if the cost is even 10% less.

onlyrealcuzzo 2mo ago
> Same as anything with low margin: Viable competition and low switching costs. So why didn't this happen to AWS?
- xnx 2mo ago
  High switching cost.
  - onlyrealcuzzo 2mo ago
    Now do Google
    - xnx 2mo ago
      Advantages: Lowest inference cost due to scale, experience, advanced datacenters and custom chips. Huge customer base. AI automation will make it much easier to switch from AWS to GCP. Serious organization that doesn't let it's product unknowingly hack companies. Disadvantages: Top model is slightly behind the frontier for now. Bias against using their products because it's not cool/trendy.
      - disgruntledphd2 2mo ago
        Their big disadvantage is their terrible, no-good, awful product management (which has always been their issue).

keeda 2mo ago

> If models are becoming commodities, and the main bottleneck is actually serving them at scale, Anthropic does not appear particularly well positioned. That has been my thesis ever since Dario went on about how difficult it is to forecast capacity on the Dwarkesh podcast right about the time Claude started having 9's comparable to GitHub's. Even as an outsider, seeing all the cloud providers bemoan their lack of capacity and their growing backlogs, I could tell compute capacity was already the biggest constraint on industry growth and that it would be very hard to come by in the future... which also probably made it the only moat. And now we see Anthropic making very expensive deals with competitors to secure the necessary capacity. Claude likely still has enough momentum to make it worthwhile. But I can't believe Dario, probably the most AI-pilled of them all, would under-estimate demand so severely.

disgruntledphd2 2mo ago
> But I can't believe Dario, probably the most AI-pilled of them all, would under-estimate demand so severely. To be fair, this mess up actually increases my respect for him, as he focused on making sure that they could continue to run the business under reasonable assumptions rather than YOLOing compute into existence like OpenAI.
- keeda 2mo ago
  My surprise is rooted in the point that he seems to be the biggest believer in AI there is -- to the extent the label "zealot" gets bandied about at times -- and he still underestimated the demand they would get. I would not have classified Altman as a bigger believer than Dario. My guess is OpenAI, having the first mover advantage and the bulk of AI traffic for a very long time, could gauge the trends much better than anyone else.