ianm218

For the time being this thesis hasn’t worked out though. H100’s are only going up in price and becoming more scarce [1]. At what point would you update if now going on 3 years into H100 lifecycle this doesn’t seem close to happen.

I’ll send this to the team at my company. No one likes feeling vendor locked in on artifacts but they are great at being like “hey I had an agent cook on this topic over night, here are the results, and what we need to make a decision on”.

Just some feedback for corporate use - it’s not clear to me from the landing page if it’s really intended for internal company use and if it’s hard tied to CloudFlare. I.e. if there was a “here’s how you can deploy it on AWS in 2 minutes” option I would just toss it on our dev env and show people a POC.

With open source projects we’re increasingly worried about supply chain type risks - it would be good if there was some boilerplate on “this is how you know this can be deployed internally and have zero risk of phoning home etc”

Do you think they would actually sell compute to anthropic? Relative to working with other labs that feels like conceding the overall AI race atleast outside of consumer.

For me the tight integration with Claude Artifacts (and whatever OpenAIs copy of those is) is super value-able for me. The biggest bottleneck for me is gathering all my thoughts and analyzing what needs to be done.

I was a long term user of the TUI and tools like cmux / herdr but now I find it much easier to actually focus

I don’t personally live like this to be clear! Just that it’s quite achievable on a normal non tech/ finance W2 salary.

One of the best aspects of “youth” is your health to be fair which can often be enjoyed cheaply - I.e. gyms/ hiking / healthy food doesn’t require a ton of lifestyle creep.

The median wage for an adult 24-35 who has a college degree is $84k-91k [1] and around 40% of 30 year old have college degrees. If your making 90k you can even save 7k a year in HCOL places like NYC and Boston if you live well below your means. I.e. lots of people would just use that disposable income on traveling/ eating out/ concerts etc but if your a homebody and live with an SO or don't mind commuting and cooking most of your food 90k goes quite far with no kids.

https://www.bls.gov/opub/ted/2025/median-weekly-earnings-by-...

A lot of it is local politics too. The Bay Area has chosen very hostile to economic growth policies especially around zoning/ affordable housing etc.

So it’s like if they keep voting for people who are anti growth it seems like the expected outcome to be left out. I.e. if new apartment buildings are largely banned, you don’t get the opportunity to open a coffee shop or work as a building manager.

Isn’t this just selection bias - you don’t see the people who are out making AI deliver boring value for people unless you use their products right?

I.e. Ramp does a pretty good job of using AI to make expense management easier/ CloudFlare makes it a lot easier to configure stuff properly on CloudFlare but is just like boringly useful.

I’m excited to try this. I got Instinct after hearing all the hype and it is basically completely useless. Fails at the most basic browser navigation type tasks.

Right now the consumer agent players are trying to pretend they have decent functionality that will exist 3-6 months so they can get marketshare. So hopefully Anthropic is sticking to their underpromise and over deliver.

Lots of use cases are more speed than cost sensitive. I.e. yes if you are paying a bunch of engineers in Boston who graduated from MIT to work on robotics $165 is a rounding error if you aren’t producing at scale.

Basically anything aerospace and defense related will care more about speed and compliance than cost.

They have $80 billion in debt, they are basically going to be operating like a private equity run business. They'll be converting any cash flow they can to debt payments.

Housing prices in urban areas going down is not really the end of the world right? Seattle is incredibly expensive

it definitely feels like sooner than later there will be “Mythos level” cyber threats available on open source / uncensorable models. I think the only thing that can be done is continued work on hardening anything that is remotely close to “critical infrastructure” and software supply chain.

On one hand, self hosting will always be niche so you could pass a law like “if your serving inference at scale you need to have safeguards/ classifiers in place” since > 95% of inference will go through hosted services in the long run … but that could turn into thought policing dystopia real quick.

This feels like kinda of a useless comparison if we don't know about the tax rates they pay, cost of capital, adjusting for geographic location and local regulations etc. I.e. the largest increase was seen in NY where it is really expensive to do anything and high taxes.

> Ireland’s planning and zoning framework is widely regarded by economists and housing experts as one of the most restrictive, bureaucratic, and hostile-to-development systems among developed common-law nations [1]

So Ireland made it almost literally illegal to build anything and it sounds like that led to bad behavior? Yes Ireland should fix corruption around building but they should also allow people to build homes.

[1]. https://www.centreforcities.org/reader/a-zoning-system-for-e...

I don’t think so - those companies are atleast modestly pro growth. They want to do more projects. In NYC and Boston local residents talk about “evil developers” who want to replace vacant lots with homes and businesses

Housing is largely restricted by local politics and zoning not corporate interests. There’s no conspiracy on that beyond crappy local politics.

I've been thinking for awhile it is very strange that claude/ openai etc google "What was the inflation rate last quater" instead of just having a tool to query BLS datasets and that sort of thing.

So hopefully most of these statistics move towards first party sources.

I think we just fundamentally disagree here. I'm sure they know the current state and that it doesn't work. I think it's likely that they need entirely different people knowing about these problems and working on them in a totally different reporting structure to get the job done. It's a hard job and they're not up for the task.

I think the reason they can't make that change is that Microsoft is a slow moving rigid culture and doesn't have the kind of org that can adapt to a 10x or 100x increase in usage compared to companies like Google or the AI Labs that handled much more difficult scaling problems.

Have you seen this legendary email thread from 2003 https://www.osnews.com/story/19921/full-text-an-epic-bill-ga...? This isn't a new problem at Microsoft.

My guess is that the problem is internal politics and incentives. Yes I think it is a fully solved problem for a company worth trillions to have an infinitely scalable "pull requests" page.

For example, Microsoft is known to pay significantly less than their competitors. I am guessing they just don't have the systems internally to say "This problem is worth literally billions to us in revenue per year, so we will pay millions to IC's to get it fixed and clear internal roadblocks".

It's not "just" small tweaks it is likely that people who are entrenched need to be moved and they need people with the right incentives in place to get it done.

I think you are giving them way too much leeway. For the service they provide there’s no technical reason it can’t be infinitely scalable. For example their “pull request” page constantly goes down. Serving that kind of workload at high burst load is a fully solved problem.

If they are genuinely compute constrained on CI they should obviously reduce the free offerings rather than take down all their paying customers constantly.

Why do these benchmarks often show models running on Max i.e. Opus 5.5 (max) is on the frontier? I would think it is never really economical to run max

> I’ve now applied for over 3 dozen jobs and I’ve heard back from 24 of them with a rejection.

I wouldn’t encourage anyone who wants to do something else to stay, this just read as strange to leave to me. If he wants to find a job in tech surely he could reach to many of the people he’s worked with in the past or connected with at various events before throwing in the towel.

Supermarkets very efficiently drive down the cost of food so that food is abundant - basically everyone in developed counties has access to low cost food. Part of that process is not wasting overhead on having people manually checking various gauges.

It’s a very low margin business if you don’t work efficiently you’ll get outcompeted by people who do like Aldi/ Trader Joe’s etc

So their assets/ brand/ people are worth hundreds of millions or billions to strategic buyers like nvidia but we should just make them go bankrupt? Why?

The existing search on macOS was almost completely useless to be fair

I think the thing is engaging with “ML is mostly math and science engineering, AI is getting better at autonomous math and science, if it gets fully super human the rate of change could be mind boggling” is hard to fully dismiss in the same way you can fully dismiss weird Bay Area people. So that’s why we are seeing the pivot towards the personal attacks on these people.

It seems like to me it would be more effective to focus on the object level aspects of their arguments rather than their weird lives? Their argument isn’t “look at my credentials, I’m an expert you must believe me” it’s that “ML research is largely math and science engineering, as AI gets better at math and science it can use that to speed up the cycle at which point something bad could happen”.

These Bay Area people dedicated their lives to AI risk often before AlexNet even showed deep learning could scale, by definition they are eccentric people.

At the same time, people like the AI 2027 guys (who that article speaks poorly of)… have been basically fully on the money so far.

AMD grew their revenue 50% last quarter from the year prior, for 11.5B in revenue. If this team can help continue that or grow that then yes - the amount of money involved here is insane.

Most LVT advocates would be in favor of letting these kinds of policies be more market driven.

But say there is an urban area that is 60% parking, by default LVT would probably make that uneconomical and the land owners would be incentivized to redevelop or sell.

A concrete example is I used to work at a warehouse in Boston as a teen that had hundreds of parking spaces… in one of the most housing constrained cities in the whole country in relatively prime real estate. I hope that lot gets turned into condos or something it would be a huge win for the city.