Hacker Newsnew | past | comments | ask | show | jobs | submit | ac29's commentslogin

Airlines have major outages frequently, its a pretty fragile system

Oddly enough they didn’t used to and used to have rock solid computer systems with paper backup procedures. So did banks

Youre talking about gross margin (revenue - cost of goods sold). Operating margin, which includes other operational expenses, is still negative. Net margin, which includes all expenses, is deeply negative.

When people say Inference is subsidised, they obviously don't talk about the company itself being profitable. To be fair, I included training costs also.

I don't think that's obvious at all. The price charged for inference is subsidized in the sense that they don't charge a rate that actually makes sense if they want to be a profitable company whose main product is selling inference. If they did want to be profitable, obviously they'd charge much more for inference. Hence, people are saying it's subsidized.

No sorry I don't think what you are saying is correct. There's a clear difference between saying "companies are losing money on inference" and "companies are not profitable". One is about unit economics and the impossibility of ever becoming profitable.

Muse Spark Open is 816B-A42B: https://huggingface.co/meta-models

My current favorite kitchen timer is a Japanese one I got online. Its got 5 buttons: start/stop, reset, and add 1/3/5 mins (so for example a 12 min timer would be 5,5,1,1,start)

> The $2 for every $1 inference cost quote from Anthropic I presume includes all opex

No, it does not, its just revenue minus compute. Per their S1, Anthropic has never turned an operational profit.


That's simply not true. See page 160 of the S-1 - clearly shows positive operating profit in Q2 2026.

The large open models (GLM-5.3, Kimi3, etc) are far more expensive than their large closed equivalents (Opus, Sol), because the closed models are far more subsidized when using subscriptions.

If you see no difference at all in between Opus and Deepseek Flash for your work, that is great! Personally, I found the latter to be unusable for the projects I was working on.


Can you describe what you are doing where comparing the two is problematic?

For what it's worth, I use DS flash for statistical analysis, design (C++, Python), scripting, server maintenance. The same I used for Opus.


You can get amazingly far with an ESP32, RP2350, or other modern microcontrollers if Linux isnt a hard requirement.

The latest Deepseek Flash is much, much larger than Qwen3.8 Flash next

They say "our search results also include anonymized API calls to all major search result providers worldwide": https://help.kagi.com/kagi/search-details/search-sources.htm...

Google of course does not have a search API - Kagi is paying (IMO sketchy) SERP scrapers for these results. One of the reasons I canceled my subscription


Thoughtful enough to care about scraping, yet not thoughtful enough to place the blame on google.

The data published in their S1 says they took a $693M operational loss in the first half of this year.


That is from the increase in the share price, because they owe that in shares and the share price is higher now. On the other side is the IPO, it balances the books but it doesn't count as an asset today, whereas the promised shares count as a debt today.

They are cash flow positive, $3.4 billion in H1.

That means they're making money.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: