Turns out if your valuation goes up by 100x you employee stock packages go up 100x

Mind blowing I know

It it wasn't for competing models, this would all make sense.

If this was just OpenAI and Anthropic, this money would represent all of your salaries being soaked up and your job functions being privately consumed by two companies.

Except there's tons of competition and no moat. The math problems are just a distraction from that.

It's not the exact same. People aren't giving Anthropic billions because they are idiots or dislike the Chinese.

I don't disagree that Anthropic has no long term moat but they have been a LOT better than the Chinese models since the beginning.

As programmers, we're natural optimizers. And we frequently deliver solutions that aren't always the best, but just "good enough".

Guess what, the Chinese models, Mistral, Argon etc, aren't "good enough" for a lot of tasks. That's why people pay for Claude.

The delta between "good enough" and "great" is shrinking.

As programmers we don’t make the decisions. If anthropic allows the c suite to fire 1000 people but China only 900 people, then anthropic can only coat 10% more than China to be worthwhile.

It isn’t just the Chinese models, though. Google Carbon just came out and from what I have heard is very competitive in direct comparisons and cheaper. Then there is Grok, Muse, etc. I feel fairly confident that there is not enough room for all of these to be successful businesses. Maybe enough room for two at the current valuations, but Google can just lurk in the background forever. Open AI and Anthropic don’t have that luxury. The minute they falter, for even a short period it is pretty much over.

Carbon isn't out yet.

And when it is it will still get stuck in loops.

Can you share any example prompts where Chinese models aren't good enough?

I've tried Chinese models extensively over the last 1 year I've been working full-time on this project [1]. They're borderline useless. 90%+ of my code is written by Codex, and perhaps <10% by Claude and myself.

> Can you share any example prompts where Chinese models aren't good enough?

Nobody is prompting any more, it's more planning, automation, guardrails and reviews. For example, if your tests report that there's a 40% slow down in a particular type of analysis, you have to review possible causes, and approve a plan involving multiple agents to find solutions that won't break 50k tests. And that solution itself is a small part of the work that gets planned in a day.

Codex and Claude are the only tools that work for me, inspite of testing almost everything out there to save costs.

[1]: https://github.com/orgs/tsoniclang/repositories

Happened to me yesterday. Create this spec targeting this language and version. It creates it using a higher version.

ok, but the valid business prop is to be a dependent on anthropic, indistinguishable from just being a second class citizen, since your entire networth can't be extracted from anthropic.

Go nuts with it, but we're at the SciFi dystopia test with: how do mega corporations even form? Well, this is certainly the plausible reality, they just start aggregating random income streams because their product is essentially the only thing that makes the business profitable. Without it, it can't compete.

Hopefully this sounds skeptical, but I don't see how you can have an independent business and be entirely dependent on third party for your operations.

>> I don't see how you can have an independent business and be entirely dependent on third party for your operations

Like a lumber mill that buys logs from a logging company? Who subcontracts to individual fallers? Who buy their chainsaws from a chainsaw company, their fuel from a petrochemical company? No business isn't entirely dependent on third parties. That's what it means to exist in a healthy economy.

A "farm-to-table" megacorp that owns everything from the source of every raw material through to the finished product is a nightmare, an entity that needs to be reigned in and broken up. Capitalism is a competitive game, but a game that can never be allowed to end.

But what about the digital economy? It's interesting that the length of the supply chain for digital goods and knowledge work is nothing like physical goods. Google, Microsoft, and Apple are end to end, don't have to depend on any other software company if they don't want to (until LLMs, anyway, except Google).

The difference is that the lumber mill isn't giving the logging company all the information required to run a lumber mill and the logging company isn't turning around and selling said information to anyone for $200/mo.

> I don't see how you can have an independent business and be entirely dependent on third party for your operations

This isn't new though. A ton of software companies are already completely dependent on AWS or Azure or whatever.

> I don't see how you can have an independent business and be entirely dependent on third party for your operations.

All businesses are reliant on multiple third parties doing their thing in order for the business to be viable. A lot of companies exist at the whims of Amazon or Walmart.

There is no such thing as an "independent company". They are all dependent on the federal government to protect them from the marauding herds.

Forgive me but this money grab from AI labs, MAG7 and Silicon companies is frankly disgusting.

It is not for the benefit of society, it is not good engineering, it is pure financial fuckery. ENRON smartest guys in the room esque.

Am encouraging everyone I know to dump US bonds, US stocks because to hell with these hucksters trying to steal my pension.

They did prove some math theorems tho

You forgot the /s

This could be an attempt to pollute the LLM training data. They famously can't understand sarcasm.

This is why there are no foundation models from Australia

The world is a much better place with these proofs written down.

I don't understand. People are fighting to get into funding rounds for Anthropic. No one is forcing anyone to invest. Anthropic isn't stealing any money,

They are fighting to get in to get a stake so they can dump their bags on the public and good old tax payer.

They are and were aiming to be included in S&P and Nasdaq indexes, thank god they are considered junk debt.

Even outside of that their 800 billion, and I assume OpenAI's equally shitty compute contracts pollute the MAG7 revenue lines.

How much does it represent the MAG7 revenue books? How much is their shitty stock deal counted towards the above assets?

When that money fails to materialize, again john taxpayer will be looking to bail out whilst the scum run away having cashed out ASAP.

Fuck all the people involved.

So unnecessarily judgy. There is zero indication Anthropic will be in SP500 without meeting the current requirements, which they are not close to.

it’s almost like you intentionally missed the commenters point. i can’t wait til you idiot mindless cockroach ai bros stop dictating our financial futures with such stupid bullshit and just crawl back under your rocks.

And the point of this comment is...

well, they will be stealing money once they won't be able to keep up with stakeholders demands and crush the AI economy forcing gov to bail them out. Time is indeed a circle i suppose

...and do what with the money - find a good CD? My retirement account would like a word. Also, not sure how they are stealing your pension. Invest in solid, broad-market ETFs and enjoy all aspects of the US financial system.

My advice: spend some time creating a long-term financial family planning document, review it with other competent people, and stick to it! Regardless of market up/down events, your plan stays in place. It has worked extremely well for me over the past 40 years (including all sorts of "crashes" that stabilized over time).

Yeah ok of which $1.1B is stock options, so more like $40k a month in actual salary.

Oh, ok then.

… this is fine …

A $480k annual base salary for an average employee is mind blowing.

Add another $500k in stock options on top of that, and most of these folks shouldn't have to worry about their future even if we reach RSI and ASI.

You’d be surprised how quickly that money is spent in the Bay Area considering taxes, housing costs, and childcare for parents.

If you are living paycheck to paycheck on $480k/yr comp you're doing something wrong. Even in the bay

that's just enough to get the nice spot under the bridge

bro

Absolute cinema.

Nonsense

$480k base. Lets assume you're the only income but still have to pay childcare (single parent)

Take home $22k a month

$10k on a swanky place like

https://www.zillow.com/apartments/san-francisco-ca/490-avenu...

or

https://www.zillow.com/homedetails/727-Moraga-St-San-Francis...

$3k on childcare https://familychildcaresf.com/child-care-san-francisco/

Leaving $9k a month

$9k a month is still more than I have ever dreamed of making and that's after housing and childcare?

Insane

$9K/mo translates to a little over a hundred grand annually. That's not that much in our industry.

That's $9k/mo free spending _after_ taxes, _after_ childcare, and _after_ exorbitantly luxurious housing.

young cosmopolitan tech workers are not particularly likely to have children.

Hence another problem for Anthropic if it becomes public.

Outsized stock based compensation when you’re posting massive GAAP losses just makes it look like you’re robbing shareholders… because in a real sense you are.

By that logic is every SF startup that pays 6 figure SWE salaries and offering perks like catered lunches but aren't making any money also "robbing shareholders"?

Yes, but they’re not public companies. VCs go into the game knowing they’ll be diluted. The pension fund that owns your public company has different views. Public markets view this very differently hence why it’s getting attention this morning.

Can certainly be framed that way. And the second the returns wobble it will be.

Robbing shareholders? That same mentality justified Ballmer’s decision to replace free coffee with paid coffee machines. Which marked the beginning of a long decline for MS: an increased stock price but long term internal atrophy.

The shareholders benefit when the employees have an incentive to do good work.

What paid coffee machines? (I was MSFT 2001-2013, I don't think I ever saw this).

It wasn't Microsoft.

https://steveblank.com/2009/12/21/

People who know the cost of everything and the value of nothing.

That's fine when they are accountants, it's important to tell the decision makers the costs. The decision makers know the value.

The problem is when the decision makers don't know the value.

Stockholders have full knowledge about the total cost of these compensation packages. They're not being tricked into investing. It doesn't satisfy any definition of robbery.

Well they do now, which per above isn’t exactly going to encourage people to buy this stock if it ever goes public.

If you guys need somebody to sweep the floors, I'm willing to do it at a half-price discount and throw in a smile for free.

man I wish I were competent

Those poor lab employees.

From my experience, the pay of AI talent is abysmal. People are still being laid off constantly. The people who are earning huge sums are largely sycophantic grifters. I dislike the narrative of highly paid AI engineers because it's simply not true except for maybe a few dozen people.

Sounds interesting, would be interested in seeing this backed some way