mxschumacher

valuation gains on investments are one-offs, not signs of sustained improvements in profitability that would warrant higher market caps

when those valuation gains are in turn the result of circular financing schemes (a bakery giving out money so that people buy bread from it), we're getting to a dangerous situation

there's still some lock-in here: data storage (iCloud), the broader Apple ecosystem, certain apps, user habits (I'm a lifelong Android user and am having trouble using other people's iPhones)

the difference here is that switching is trivial due to standardized APIs to the underlying LLM capability.

harness <-> gateway <-> inference provider

Easy to switch any of them and (mostly) possible to combine any with any

Replacing something like Excel is crazy-hard because of network effects, replacing an Enterprise CRM is akin to a removing a metastasizing cancer

it's astounding how much of human work are tasks like the one you have described here. Dancing around messy interfaces because nobody bothered to properly standardize and automate

please stop being so mean, they also create echo chambers to kill political discourse (i.e. the foundation of democracy) and show the most inflammatory material to their users to pit them into a shouting match against each other

not if it's a fantastic book

i have trouble understanding these situations, e.g. the AI itself would presumably make the suggestion to write a python script for such a task. It seems to me that there two huge problems right now * understanding which category of problems an LLM is an appropriate solution for (rather than throwing LLMs at any and all problems) * matching model capability (and therefore cost) to the problem at hand. You can easily overspend massively by using a model that's too powerful

not sure one would expect huge revenue increases from these internal tools, but maybe dramatic cost savings? Surely a lot of corporate processes could be automated?

here's an article on DeepSeek's strategy that I found instructive. Cheap & good inference in order to boost the hardware makers: https://x.com/bookwormengr/status/2057909493250539891

I'm still not entirely clear on the problem <-> capability matching. E.g. it seems like Kimi K2.6 with good context would already be able to solve a huge chunk of problems. What share of prompts require frontier models?

not just renewables, also massive nuclear capacity and huge modern coal plants. They can really crank up capacity if they want to. How long will it take to get a new nuclear power plant operational in the US?

if the unit economics are broken (strong competition from other proprietary model providers + open weight models; LLM token race to the bottom) it's not clear how high revenue growth translates to high profits. These companies are valued like monopolists, but the competitive dynamics make them more akin to tomato sauce makers. I understand that the technology is pretty amazing and can lead to significant productivity gains, but from a business perspective, the question is how much of that value Antrophic and others can capture over time.

that's a very impressive demo, I have never seen a robot move so smoothly before.

My copy of Breakneck arrived a few days ago and I'm rushing through the book, hard to put down, highly recommended

there are around 140 EV companies in china competing very aggressively, they have excess capacity and are flooding the world market with cheap EVs, tough for Tesla to have a healthy margin in that environment

and SpaceX has been a major buyer of Cybertrucks

and Tesla is valued at over 21x more than GM

there was a rush to buy electric cars in the US for as long as the $7500 incentive was in place, so the Q3 2025 number if inflated; it's a pull forward effect.

Sales have been flat for 3 years and the delivery numbers in Europe are catastrophic

on a fully diluted basis, the market cap is above $1.6tn, so at a PE of 20, they'd have to generate something like $80bn in profit per year - hard to do in an industry that is as brutally competitive and low margin as passenger cars.

what are some historical examples of bubbles that didn't pop?

Originally submitted here: https://news.ycombinator.com/item?id=46212259

you are giving a new meaning to the term "PR stunt"

i have to say I'm a little disgusted by these statements. LLMs are useful for many problems, but is there really a conceivable path of them making progress into fighting the countless cancers tormenting humanity?

are you paying them enough to have profitable unit economics? How hard would it be for you to switch from one provider to another?

openAI is far past the venture stage

that's what Carvana is doing with the car loans it securitizes : https://hindenburgresearch.com/carvana/

what is OpenAI's durable, competitive advantage that differentiates it against the numerous other LLM providers? Investing at a $500bn valuation for a company that's losing money and has bad unit economics this seems rather aggressive.

it's not about taxes, it's about fighting illegal activity. Terrorist financing, drug dealing, human trafficking etc - do you really think it's a good idea to let those actors exchange payments freely?

slower transaction processing is more profitable, because banks can profit from interest in the interim. It's not some law of nature that it can't be done faster.

thanks

I'd really love to see the technology be used more in long-haul trucking. You could use roads in the middle of the night when capacity utilization is low on highways.