Comment by daft_pink
1 day ago
I think when these companies IPO later this year, we’re going to see the reality of the PNL numbers and whether they are sustainable etc as all the financials will become public.
Rumor mill suggests that Anthropic might be profitable (but at what magnitude), OpenAI is not profitable, Google is mostly vertically integrated and has a low cost structure as they are have pre-existing data center buildouts, their own silicon and experience that suggests they will be able to operate at a very low cost, but they still have to justify their spend.
I think having to report numbers publically on a quarterly basis will bring the whole thing into reality.
> later this year
One can hope that reality intrudes before the bubble gets even more dangerously inflated, but how many years has Tesla had a ridiculous P/E ratio. Even after growth stagnated and market leadership was lost in Asia and Europe. Number still goes up.
I hate Musk, and I'm not going to justify Tesla's crazy stock valuation, but consider the following:
Outside of China, Tesla's probably the only company that can compete on battery prices. I don't know how accurate it was, but a news report was comparing the cost the manufacturer's pay to build the battery. Chinese companies were around $6000. Tesla was at $7000. Everyone else was around $12-15K. This is why a number of companies have exited the EV market - they just can't compete. This is why Ford lost money on every EV, despite the high MSRP. This is why the Ford CEO says "We're f####d" when he saw Chinese cars.
The only hope regular Japanese/American/European auto manufacturers have is if EVs do not gain substantial market share.
If the future is EVs, Tesla is the only non-Chinese company that has a chance.
It's depressing.
This assumes it is literally impossible for anyone else to reduce their battery costs to a level which makes them competitive with Tesla in an environment when battery prices are falling rapidly and the tech continues to evolve (and Tesla's EVs are not even unusually cheap or experimental in their battery supply chains)
That sounds less likely than the bull case Tesla is trying to make on "we're a robotics company now" or "one day all cars will be autonomous taxis controlled by us".
4 replies →
It's not depressing because it's not true. Tesla isn't investing in battery technology. Tesla also isn't developing new models at the same rate as VW, BMW, Mercedes, Hyundai Kia, etc.
9 replies →
Its interesting, practically all car makers seem to have factories in China, why cant they compete?
If anything though, this shows that at least tech driven hype bubbles can stay around way longer than we think if we are looking at it from a product POV.
This just means short sellers might have a hard time sinking a hype-category stock with reasoned research because the irrationality keeps it afloat.
"Markets can remain irrational longer than you can remain solvent." - John Maynard Keynes
1 reply →
Agreed, and assuming local open AI models start catching up, which they seem to be doing, the foundation models' hold on society gets a lot slipperier. If there's a "what to do about all this" from an engineer's standpoint, pushing the needle toward local models, whether in research, agents, or just using them, understanding how they work, and advocating for them when it makes sense (which is more often than they get credit for) is probably the best ROI.
While I agree with you on open models getting better, I have been starting to see how the value, the reason you pay for Claude, isn’t in the models.
For example, I just hooked Claude desktop up to my outlook to build a report for my timesheet then I used the chrome extension to fill it out automatically with that data. It could read Jira tickets if that’s where the information was.
A local model can’t do that for me because I have to get the rest of the integration software somewhere.
I also think this is why OpenAI is the worst positioned of the group of AI giants. Anthropic is trying to make a productivity operating system, while ChatGPT is basically just a website until recently.
Aside from filling in the timesheet (which I assume could be done from a CSV import - I appreciate this is another step), I have almost the same setup as you, without AI. I have bugwarrior pulling JIRA tickets and github PRs into taskwarrior. I have an integration from task warrior into time warrior and from there another hook back into JIRA to get titles and summarise.
All of that is done with two API keys and no AI. A local agent could easily put it together for you.
Now sure, I had to put this together and I lose the AI summary you have plus the auto filling, but what I'm trying to say is that I have 80% of this without any AI.
I also have a script parsing git from my emails and a little tui that translates them into git diff and a key binding to pull the PR but I find it a little cumbersome and don't really use it, and trying to parse todo from maildir is also a little useless so I accept that AI would be better there.
I also accept that your example is just one prompt of a dozen and I have to plan the solution for every one of your prompts, but I also don't find prompting to be terribly useful for occasions where I don't think the solution through- because it probably means I don't know what I want or don't really need it.
What I do find it really useful for is digging through Kubernetes and asking how two services are connected. Claude is better than local for that but there's nothing inherently non-local about that usage.
1 reply →
I get local models to drive applications through MCP (e.g. Google Chrome DevTools) via OpenCode all the time, and do things that would otherwise be very token-intensive and result in pointless meatspin. This is totally possible, and will become more so.
The real reason you pay for Claude _is_ in the models. The locally runnable models are impressive for what they are, but simply will not accomplish the task as effectively, incisively or quickly enough. I have to be willing to let OpenCode run in agentic loop on "download my bank statements"[1] for an hour and just walk away, and take a low-ish but profoundly nonzero chance that it will just fail. Claude can do it in 5 minutes, if I let it (I have), and it will not fail. Both are driving the browser via MCP and performing the same task.
[1] One of those difficult-to-use, modal-rich JavaScript-laden banking portals that seems quite intentionally designed to prevent this sort of downloading, or I wouldn't bother letting an agent loose on it in the first place.
So far there's no moat though. A lot of that kind of stuff is available open source too if you look for it (and was available before claude desktop). And for anything that doesn't exist, with coding agents now you can write one up in an afternoon.
It's kind of paradoxical in a way. By making writing software cheap, they've made it much harder to create a moat for themselves that involves only software. It'll be interesting to see how they respond.
3 replies →
OpenAI may be better positioned than you think with Codex being able to drive your applications without an API according to Nate: https://www.youtube.com/watch?v=2d9ZmA-4QzU&t=1093s
this is also possible with local models, MCP is an open standard
if you already have claude, you can even get it to set it all up for you too
the issue with local models is just the compute required, it's hard to compete with 100t/s cloud models
Unfortunately for Claude, you can have it set up local models to do that for you and then you don't need to renew it. Codex's computer use is better than Claude's, imo. I'm a Mac, no idea about windows but I know there's no Linux version. Haven't had enough time with Opus 4.8 but GPT 5.5 > Opus 4.6 and 4.7.
1 reply →
Google really needs to work on the user experience. The Google Cloud based approach to Gemini for coding is so clunky.
I thought most people used Antigravity to code with Gemini?
https://antigravity.google/
they really dont care I think. Only Sergey brin cared for a while and he is silent again
The current Gemini experience is "The Google Experience"; brilliant engineering underneath, disdain for customers on top.
Things went down why they hired anti gravity team. This thing is very had and isn't even opensourced.
Anthropic have only scraped together a month of profitability by cooking the books. They have an extreme compute discount from spacex that only applies for the first couple of months of the deal. By pushing the costs down the road they can make themselves look good before IPO. Even they have admitted publicly though that they don't expect profitability to last.
i think the narrative of “all white collar employment replacement in the near future” can sustain their public market valuation for many years, regardless of how profitable they are in the medium term.
> having to report numbers publically on a quarterly basis
They won't need to do that if the new rules come into effect.
> I think having to report numbers publically on a quarterly basis will bring the whole thing into reality.
That is a good reason that all companies (over a certain size, say in terms of gross expenditures) should have to report such numbers. There's no reason that huge companies should be able to distort the economy while not having to report anything just because they're not publicly traded.
you: insurmountable cost structure for LLM providers
also you: only three companies make LLMs people pay for
sounds like a 15 minute phone call to form a cartel is all that is between them and profitability? less money has been made on more complex schemes.