rsalus

The conclusion that this is a bubble seems arbitrary and disconnected from the several true facts outlined in the article. Anthropic's growth _is_ slowing, albeit still strong at ~47%. They do have enormous compute obligations, and it will require a lot of growth to meet them. However, they also increased revenue per dollar spent on compute from $0.40 to $2.08 in five quarters and hold $131.7bn in cash.

Keep in mind this growth has mostly been sustained by low-level business automations, coding, and chatbots. With the adoption of products like personal agents, e.g., Meta's Muse, I am pretty optimistic there will be demand and growth around this level will be sustained for a few quarters. Doubtful we will ever see another ~150% growth quarter.

general knowledge perhaps, but certainly not intelligence in any traditional sense

I mean it would make sense if the training using that data resulted in a general set of equations/patterns, it's still really surprising though...

I'm convinced all engineers in the O365 org are call center techs in disguise

speaking as a FTE at Microsoft... yup

does this happen for Claude Desktop too?

I think it's safe to assume anything associated with PDFs is radioactive.

Definitely ease of use...

but the investment is increasingly _fueled_ by debt, regardless if OpenAI is the direct borrower. just look at the bond market. you think we'd be seeing these rates if corporate bonds weren't directly competing with government-issued bonds?

they've borrowed so much they're directly competing with national governments in capital markets.

> I fully expect to see situations where AI generated code becomes so hard to work with that even LLMs prefer to throw it away and start fresh again.

You know, I've never thought about that. It's really interesting to consider, especially since refactoring code is now cheaper than ever.

it is unarguably problematic when only 10% of your demographic is able to meaningfully participate in your economy outside of fixed head costs...

not to mention, the marginal propensity to consume for the top 10% is 5-6 times _less_ than the bottom 90%, so I don't know how you can claim the current consumption rate is good?

well, for one, the top 10% households drive most discretionary consumption in the US because the other 90% are broke.

I doubt they have the compute capacity to support a commercial source control platform. I'm sure you've seen how insane GitHub's traffic has become.

didn't say they invented them.. I'm just saying they follow them rather dogmatically.

I assume he's referring to the long-standing Microsoft engineering principles of "zero trust": https://learn.microsoft.com/en-us/security/zero-trust/zero-t...

and "least privilege": https://learn.microsoft.com/en-us/entra/identity-platform/se...

Don't even get me started on PowerBI...

Besides being obviously AI editorialized, the idea that a few fishing boats were the reason Somali piracy stopped and not because of an international coalition of navies is... quite a stretch.

I feel like half the game now is marketing though, so I can see why they'd be attractive to an investor. Maybe if they scale they can come up with something.

+1, I'm definitely running _way_ more kinds of CI tests than I had before AI.

super interesting. I have so many half-considered questions... like sandboxing (it seems like it is BYO). would love to have some kind of policy engine.. perhaps an integration with https://github.com/NVIDIA/openshell in the form of an extension?

also, I see most of the durability promise comes from persisting JSON documents locally and minimizing the amount of context/data kept in-memory, even during SQLite mode. while this makes sense, my own experiments with a process that relied on a JSONL-based event store have led me to prefer keeping things in-memory to avoid all the friction with I/O.. am I crazy for preferring just a straight .db file being persisted?

the latest 07-28 MCP spec is quite different than the previous iterations of MCP, so I understand the delay there tbh.

the evals? setting those up and empirically proving them is genuinely a lot of work.

I think the new 07-28 spec is quite decent

do you mean NativeAoT? it's mentioned a few times in the post

these blog posts have really turned me into a dotnet evangelist. love the framework & language.

> I guess you didn't understand my comment ;)

yeah, touch grass buddy.

what a patronizing response. I'd be embarrassed.

exactly, and realistically if you test the crap out of a given module, you'll have a pretty damn accurate idea of how it behaves.

Yeah, by the time my thought is fully formed I've gone through an entire conversation.