Whatever you think of AI for your own work, the fact that the entire economy is betting everything on it is really concerning. It’s not just the fact that it may not work well economically speaking and would end up with a market crash, or all the negative externalities from AI development, it’s also the opportunity cost we are paying. With so much human capital and resources dedicated to developing and running LLMs all the other business, research opportunities aren’t being explored and invested in
I prefer that scenario to the one we had before where too many smart people were building out social networks and similar services trying to maximize ad impressions.
We are still dealing with the consequences…
> too many smart people were building out social networks and similar services trying to maximize ad impressions.
Honest question: where should these very smart people work then in your opinion? Nearly all suitable-looking, "smarter" ideas that come to my mind will bring less revenue/profit in expectation than, for example, letting them maximize ad impressions.
I of course wish we lived in a "more intelligent" world where this is not the case. :-(
Call me crazy, but the smartest people I know have rejected maximizing their employer's revenue and profit as their goal. Or even their own company if they go that route. Find a niche, work it, enjoy it, and enjoy the rest of your life, too.
Probably not on systems that rot democracy, freedom, and human expression. Companies like Google and Meta are net negative for humanity and the people that built these companies and continue to work for them are evil in their apathy towards the human cost of their salary.
> where should these very smart people work then in your opinion?
Anywhere else. Ads are a net negative to society and provide no value. ANY activity outside of pushing ads is a better contribution to society as a whole.
I'm not even sure that I do prefer the AI train to the social media / ad train. I certainly would if the direction was "we're throwing ourselves blindly in AI so to gradually free citizens from the need to work", but right now it's "we're throwing ourselves blindly in AI so that we can sacrifice workers for higher corporate profits".
Not to mention how they will be used by terrorists and evil people to create biological weapons and other tools of mass slaughter. Or how it will be/is being used by criminals to target vulnerable people and businesses. Or how coupled with drone warfare the world becomes an unlivable hellhole in the next decade. But Marc Andreesan said only losers are against “tech accelerationiism” on twitter so here we are.
You can, right now, find out how to synthesize VX or sarin, or how to weaponize anthrax. You can and have been able to acquire this knowledge without the use of AI. Knowledge is not the obstacle to proliferation of biological and chemical weapons. It's hard to set up the infrastructure. See Aum Shinrikyo.
Sounds like you're reading into Amodei's propaganda as to why they should control AI. The information to make biological weapons for example is nothing new, it has been on the Internet for decades.
> With so much human capital and resources dedicated to developing and running LLMs all the other business, research opportunities aren’t being explored and invested in
It probably feels this way if you only see headlines like this, but the LLM companies and even extending into their data center buildouts are still only a part of the economy. Investment dollars get spread around far and wide. Investors can’t even get allocations in these companies if they want to because they’re so oversubscribed.
This headline and website are a good example. If you think being a YC founder is rare and reserved for the truly elite, you may not have seen the size of YC batches lately. The number of YC founders is in the tens of thousands. Showing a list of 100 of them who joined a couple big tech companies doesn’t mean anything by itself.
It’s on the order of 1% of YC founders. YC startups don’t work out all the time and their employees and founders go work for other startups. Barely worth noting, if not for a website making it look like a big deal.
I don’t really buy the opportunity cost argument. The bets we are placing look EV-positive to me.
The 90th percentile upside is way, way higher than is priced into the current stocks, the 10th percentile looks to me like 2000 where a bunch of investors loose their shirts and then for the next few decades new companies boom on all the infrastructure that was built.
The “failure mode” for this tech is that we don’t advance much beyond Sol/Fable and the existing models just get cheaper. If there is an AI bust in hindsight it likely will look like the dotcom bust did, which is a small blip on the decades-scale GDP curve.
I fear another related possibility. With *so* much money on the table, if this new industry shrinks or collapses, it is hard to believe that it will collapse same way dot-com did, and instead investors will seek to protect their capital who knows at what cost. Given the influence they have it is not hard to imagine the impact on policies, etc. I fear that tax payers are not the winners in either case, whether this industry grows or not.
This reminds me of the office rental leases during and after lockdowns where it seemed like the idea that owners might loose money was just not even on the table and we needed people to be back at the office otherwise the value of office buildings would fall and that'd be terrible...
> Whatever you think of AI for your own work, the fact that the entire economy is betting everything on it is really concerning. It’s not just the fact that it may not work well economically speaking and would end up with a market crash, or all the negative externalities from AI development, it’s also the opportunity cost we are paying.
It's worth pointing out that "entire economy is betting everything on [AI]" you're talking about here is in the economy in the "numbers go up sense" (e.g. the stock market). It's not the economy in the providing for the general well being sense (e.g. gives people jobs so they can get housing and food).
Investment from VC and wealth funds is going pretty much only to AI projects. The massive capital expenditure we are currently seeing is going to AI infrastructure (HBM has pretty much no use case outside of LLMs and a few niche industries). It’s real money being allocated to a single bet
Researchers are increasingly using LLMs and other transformer based AI. My friend at a well known research institution asked Anthropic for access to their biology model (not sure if it's Mythos or some other sort of uncensored model) and recently made breakthroughs with it and other AI.
Yes, I don't have their permission to dox them, so it will have to be vague for now. Still, there are many news articles outlining the same point I made, except with real examples.
I also think this is a huge problem in government. A lot of government services could be digitized via simple CRUD API/UIs and instead politicians talk about AI...
I know Ed Zitron is a dirty word on HN most days but having recently read some of his (highly editorialized) work… it seems like he’s right and the numbers just aren’t making sense.
Combine that with the fact that “good enough” is a real thing and cheap Chinese models are either there or close depending on your use case and it’s hard to see how this ends well.
Well yeah, but being a CTO at a major org isn't exactly a low-paying gig.
The interesting thing to me is that the belief that Anthropic is going achieve something like AGI appears to have really spread through SV. If not AGI, at least greatly disturb their old cushy SaaS jobs.
Also, I still do believe some people just want to work on "cool" stuff.
edit: not just CTOs and CPOs, I mean John freakin Jumper, who got the 2024 Nobel Prize in Chemistry, left his VP role at DeepMind to have no reports, and be an IC at Anthropic.
> Also, I still do believe some people just want to work on "cool" stuff.
One thing I’ve noticed more and more is all the “cool” jobs require you to already be an established domain expert.
Another things I’ve noticed, and this is probably mostly just my opinions evolving, but all the “cool” stuff that catches my interest these days is not in software. Software is boring.
I’ve been in a bit of creative/intellectual rut the last few years, but before that there were a handful of things that intrigued me in the last 5-10 years.
Most recently, I became interested in the field of photonics/optics, particularly in the field of novel optical communications research.
I’ve had mild interests in things like
Biology and History since I was a child. The latter particularly hooked me a few years back.
Neuroscience is a pretty cool one as well. The human brain is of course one of the most fascinating pieces of hardware in existence and understanding it (to whatever level we currently can) is incredibly interesting to me.
At one point I thought about going into security research, but now that everyone and their mother is headed that direction, it’s become less feasible.
Unfortunately I don’t even have a bachelors degree, and most of the fields are research heavy and not very accessible to hobbyists, so my actual interaction with these things remains relegated to reading papers/pop sci articles/etc.
> If not AGI, at least greatly disturb their old cushy SaaS jobs.
I mean, this is happening right now.
I doubt there are few people talking to customers and potential customers that haven't run into the "look what I rolled last week with Claude" only for it to be legitimately impressive. The idea of technical moats is evaporating in front of our eyes.
The biggest AI bubble seems to be forming within the minds of the people working in SV itself, judging by the talk about the inevitability of AGI, the strong language around fomo (permanent underclass and all that) and the close proximity of the whole ecosystem there.
So wether it's belief in AGI or a desire for building cool stuff, as an outsider looking in, it seems to be informed by the same echo chamber of thought.
I think it is more than just money. Having a hand in steering the ship that they think will change the world is pretty attractive. I came out of retirement just to watch the spectacle.
I would do that in a heartbeat if I was in that situation.
Having a major leadership role at a big company is kind of a slog. Getting a chance to go back to a fast growing startup as an IC in a way that will only benefit their career (ex-Anthropic or ex-OpenAI makes your resume gold plated) is an easy choice. They can go back to any job they want after this.
I think this might only be shocking to some people who have forgotten that many people in these positions actually enjoy working on hard things and being surrounded by other like minded people. If you’ve only worked in po-dunk startups or other companies where leadership looks like a Game of Thrones style fiefdom building game it’s harder to imagine executives who are smart, capable, and enjoy working hard. It’s pretty common in Silicon Valley despite all of the sneering here.
I disagree. I'd immediately have a ton of questions for most ex-OpenAI and ex-Anthropic people as my default position would be "poor business/economics judgement in role, chose to ride a hype cycle, now sat in front of me for a reason".
I'm not saying it's toxic to have it on your work history, but it's far from gold plated. They'd have work to do for me, and for many, many people in the industry, to prove that I'd want to work with them.
You are not in a position to hire them, let alone at a big interesting company. Forgive my bluntness, but your opinion doesn’t matter to their situation.
When I say their resumes are gold-plated I mean at the companies they want to work at.
I think most of HN can't logically arrive at this conclusion because they don't exactly buy into the premise that
(1) (digital) AGI is a real concept, and will be achieved imminently
(2) AGI will be world (and possibly more) changing.
(3) Anthropic will be the one to do that, or at least capture a large part of it
If you buy into all of these, which I'm assuming a lot of SV does at this point, then its kind of a no brainer to go to OpenAI or Anthropic. And optics wise it seems like Anthropic is "winning" right now, so thats the lab with the talent flow.
If Anthropic actually creates a singleton, contributing to that is way more impactful than being CTO at some random company (though I may disagree that said singleton would treat equity holders any differently).
Do you have examples? Because besides the other reasons mentioned about huge earning potential and being part of an important moment in history, the not-so-hidden secret is that any management position below C-level and above small team manager is usually extremely shitty. You are responsible for cat herding but rarely get to make the decisions in the first place. There are very few people IMO who desire or excel at "mid-high level" management positions (again, levels from director to anything below CTO) - I've seen people who enjoy and are good at leading small teams, but people at the director level are either there because it's lucrative or with the hope to get to CTO level ASAP, but otherwise it's one of the worst jobs in the "stress to moments of enjoyment" ratio.
So I could imagine tons of people in those roles who would love to be ICs again if there was also a chance at outsized wealth.
I'm not surprised, but what is surprising is we're now in a period in which the mask is slipping and no one seems to care enough to try to put it back on.
For all the hubris and arrogance of the tech scene, and especially the Bay Area's, founders at least pretended to pursue a greater calling than money by leaving cushy gigs to pursue a new venture they were passionate about. Sure, that only lasted a handful of years before it became a gold rush to move to the Bay for YC and pursue the startup-to-acquisition pipeline, but they at least pretended to care.
The naked pursuit of wealth at the cost of potentially tearing apart society is mildly alarming because the scale it's happening at is so big.
It seems like different people? The early tech scene was defined by people who got into tech because they loved it, jobs were paying under six figures. Then it became a gold rush, and the tech scene became defined by the newcomers who were after cash first, tech second. It's not that the old timers gave up on their ideals (well, some did), they were simply supplanted by larger numbers of people following the money.
In my opinion there is no "slipping mask", merely a changing of the guard that always happens when massive piles of generational wealth are at stake.
Nah, we knew back in the 90s that the tech industry had open disdain against American workers (pressuring congress to expand immigration to depress wages) and children (supporting charter schools while attacking children (yes when you attack public schools you're attacking the only people that benefit from it: children)). Not too mention all the horrid, extremely hostile business pursuits we saw from companies like Oracle + MSFT + Apple.
SV has always operated on exploitation of their workers, they just hid it well from the public.
Having been there i have to respectfully disagree. Probably because, as far as i can tell, you are warping their intentions to be what your intentions would be if you did what they did.
I just don't understand how to connect this analysis to what's happening. What OpenAI and Anthropic both say, loudly and repeatedly, is that they're working on one of the most impactful technical problems in the history of humanity. The reason they have lots of money is that lots of investors believe they're right.
Perhaps you've seen under their mask, and you know they're scamming the investors. After the NFT bubble it's a plausible hypothesis. But I don't know what it means to say that the mask just isn't there and it's naked pursuit of wealth.
I sometimes think, speeding up the Red Queen hypothesis to accelerate Cambrian Explosion by printing money from the top so that high-agency people would develop new products and services to capture more of the newly created money is a good thing?
The question I have is why are these companies hiring these people and what does it say about their hiring practices and the amount of capital they’re poorly spending?
It’s exceedingly unlikely that any of the people who were working on YC startups previously have any real professional experience with any of the following: slurm, collectives, NUMA systems, RDMA, compilers, systems programming, general HPC performance estimation or measurement, CUDA or ROCM or any kind of GPGPU/accelerated computing. But that is the core business of both of these companies.
I’m not surprised that these companies are well funded and hiring a lot of people. I’m surprised that they chose to hire the people who were previously making “Uber but for dogs” gimmick apps and not just hollowing out the HPC specialists from national labs.
The top % of YC founders are legitimately quite talented, and high agency/entrepreneurial talent goes a long way in these companies, even if its not in deep infra or research
People ask the same question of why YC funds yet another Uber for dogs or a button on the Touch Bar that cost $10/mo to help join a meeting faster.
They invest in people more than ideas, so you’ve got, at least in many cases, people with good pedigree and skills (age adjusted anyway) building on stupid ideas, but that are eminently employable.
Obviously there are other factors, I’m not really trying to defend anything but just point out that there are legit reasons why someone impressive enough to get into YC would also be impressive enough to get a good job. It’s not like it’s random founders off the street.
This is what always stood out to me. Founders talk about the risks they take as their major legitimizing force. But what risk? They’re the types who can (mostly) skip an interview and go direct into a six figure job - because of the sales and founder networks they develop.
Startup founders with no risk doesnt sound like a recipe for great companies.
YC founders aren't the ones taking risks. The risk comes from pouring your life savings into a company and not getting into any incubators and never getting any VC funding and then having no savings and nothing to show for it.
The actual AI part of AI startups are at constant risk of commoditization. Where the companies will survive or fail is how they manage to attract customers and lock them into their offerings so they can't move away to the next competitor.
Someone with experience running a startup that has no moat probably does have some relevant experience in that area.
I actually have experience with that stuff, “old school” deep learning and ML as well. Is that something worth joining the fray with I wonder? Or is it as you point out really only the richly connected “locals” that are recruited?
Not to try to grab any pity, but I have pretty deep experience in the GPU optimization space and have not been able to even get an engineering screen even with referrals.
I have a sinking feeling that AI models have gotten so good internally, and/or the GPU guys in their labs have found a good enough processes, that AI is doing the research behind the doors right now (it should be at least capable, at heuristic optimization problems that are benchmarkable), while the hard part of their business is the political game: convincing other companies to buy into AI, integrations, trying to gather more data where they can.
> I’m not surprised that these companies are well funded and hiring a lot of people. I’m surprised that they chose to hire the people who were previously making “Uber but for dogs” gimmick apps and not just hollowing out the HPC specialists from national labs.
Culture. The Uber for Dogs idiots will gulp down the kool-aid like they’re dying in the dessert.
> The question I have is why are these companies hiring these people and what does it say about their hiring practices and the amount of capital they’re poorly spending?
I mean, what are you suggesting “someone” should do about it? The companies can spend their money however they want, and if they want to hire a bunch of product people from YC startups nobody can stop them.
I’m just surprised they need this many of them. What do they do all day? And who is left to manage their hundreds of thousands of servers?
A lot of the AI game is still integrations and getting entrenched in processes. It has the double whammy that you get to train on their data (not talking about ALL enterprise contracts - but cursor pretty explicitly has this business model. So I'd imagine some enterprise contracts have this.)
I’m just surprised that the people actually keeping the business operating and the tokens flowing don’t resent these product people for sitting around all day coming up with ideas for the same TC package as the people putting out ops fires and bringing new capacity online all day every day. I feel like the actual engineers are probably getting shortchanged.
Effectively, industry roles are a tiered system which determines access to the cashflow. Going the YC founder route is a much faster and more efficient way to secure a high tier than climbing the SWE ladder. The whole thing resembles a kind of mini class system, with high tiers granting access to generational wealth, and low tiers being better off than the average non-industry Joe. Once you're in the high tier, there's no way to fall anymore, you just move wherever the cash is being pumped in, collecting it.
I’ve seen many people climb the SWE ladder or build a YC company (having done both myself), and trust me the founder route is not the most efficient. You only believe this because you didn’t see the 2 failed startups and 10 year journey grinding 80h weeks almost running out of cash a few times.
Not that I disagree, but it doesn't refute the original argument. It could even support it, class protecting their own from more people figuring out there's no clothes on the emperor.
What part of "start a company" requires "grinding 80h weeks"?
Sounds to me as inability to delegate. Or implicit in the [acquire VC, aim for high growth] mindset as if that were the only path to success.
Companies can grow organically. With a positive cash-flow early on & reasonable work/life balance for founders & employees alike. Heck, there's even non-profits.
Yes they can, but unless you strike luck, they take many years, often decades of work to grow to a level where you can retire and enjoy the money.
With FIRE and similar concepts being pushed around everywhere, and the plight of the masses to keep their jobs and homes, I see where that desire for "make money fast" comes from. And from a game theory mindset - it's not wrong. (That is also a bit related to the explosion of gambling and gambling-alike stuff such as "prediction markets"...)
> Heck, there's even non-profits.
Non-profits often enough run on conditions bordering on (self-)exploitation.
Nobody in their right mind should start a company up to some equilibrium. As a society we should fund the appetite of experimental founders to put more pressure on quality than what natural conditions might allow.
This. I saw way more people build a few millions of retirement nest egg reliably in 10-20 years, by working on a boring good swe job and manage their investment a bit better. It's a much safer route.
> Going the YC founder route is a much faster and more efficient way to secure a high tier than climbing the SWE ladder.
YC has funded over 5,000 companies. If you assume 2-3 co-founders per company, that's more than 10,000 to 15,000 people. The vast majority of these founders aren't producing "generational wealth" outcomes. There's no glamorization of the companies that shut down, the ones that are scraping by, and the ones that get their founders a normal job, but those are the far more likely outcomes, especially in the more recent spray-and-pray batches.
This is not even close to true. I work at a big tech company and we reject applications from YC founders every week. Most of them are broke with failing companies and would love a cushy mid-level engineering or management job at a faang. People are biased by the success of the top few founders but 99% of them have worse outcomes than the average corporate engineer.
This is a very hot take. Being a founder might confer you some career advantage [1] if you're one of the few to make it to series A and you grow a large team, but if you're like the vast majority of founders and never make it past seed (or realistically, to seed), your network is the alpha and omega of any subsequent job search. It's a bit like saying that being an NBA starter is a good way to get a job as a basketball coach.
Point being: don't start a startup if your goal is to get a job. Just get a job.
[1] Note that I'm not arguing about experience -- you can gain a lot of experience as a startup founder, but that experience is rarely directly marketable. Also, most startup founders are completely clueless when they start, so "a lot of experience" is a relative term.
This analysis doesn't actually prove the title, as it only considers founders that have gone to OpenAI or Anthropic, however even if it did it isn't particularly suprising.
Sam Altman was the president of YC, so it's not particularly suprising that he has hired lots of talent from YC - these are people that are pre-qualified because he has seen their work before outside of a job interview (better the devil you know than the devil you don't).
This happens everywhere - person leaves company A to company B and then poaches the best talent from company A. Obviously in this case the talent are founders rather than staff members, but still.
YC has tens of thousands of founders, more are probably at Google and Facebook than Anthropic and OpenAI. I’m not sure a sample of 105 founders shows… anything?
A successful unicorn founder once said a billion dollar company is just made up of a bunch of people who could each have started a $10m company by themselves.
I guess you can extrapolate that and say a trillion dollar company is made of a bunch of people who could have started a billion dollar company.
The site infers that 105 people are working at the two companies. What’s the denominator though?
After 20 or so years of YC, with multiple catches per year, and the insanely high failure rate of start ups, there must be a lot of former or floundering founders.
IMO we are in the era of “Final Companies” and the appeal of being a founder these days just isn’t great compared to the entrepreneurial boom of the 2010s and 2000s. The risk to reward ratio has never been worse.
If Product-Market fit for small companies take 3-4 years and AI can streamroll you by coming from behind, it makes sense to join the big boys rather than trying to finetune your creation in the hope of making something big.
The sepia-tinted color scheme and the way LLMs overuse rounded corner tiles to the absolute death is part of it. Really what it comes down to, is that these LLMs have a design vocabulary of like 15 traits. None of the websites they generate have every single trait, but they are composed of some subset of 10 such traits, so it's still overwhelmingly obvious the moment you open the page, even though each one is slightly different. Humans might use 1~5 of those traits, but they'll also be mixed in with a much wider set of traits with more individuality to them.
- Orange/beige-ish colors
- Rounded corners
- Cards with a thin border
- A thicker colored border on the left of cards
- Serif font for headings
- Monospace fonts for small text
- Headings that often have an unnecessary subheading / pre-heading
- Little badges, often with a "status indicator" dot on the left
- Obviously LLM-generated text / language
I'm sure I'm missing many but the above are dead giveaways!
The copy is usually a bit of a giveaway too, ”explore your future path to greatness and experience the defining divider between those who can and those who can’t” or as most humans would write the header ”plans”.
Most AI-generated sites have coopted the Inter font, and usually they go for a dark colour scheme, the rounded tiles, a container representing a macOS-styled terminal with code, and, the thing that annoys me the most, divs that fade is as you scroll.
Whatever you think of AI for your own work, the fact that the entire economy is betting everything on it is really concerning. It’s not just the fact that it may not work well economically speaking and would end up with a market crash, or all the negative externalities from AI development, it’s also the opportunity cost we are paying. With so much human capital and resources dedicated to developing and running LLMs all the other business, research opportunities aren’t being explored and invested in
I prefer that scenario to the one we had before where too many smart people were building out social networks and similar services trying to maximize ad impressions. We are still dealing with the consequences…
1000000% this. I remember lamenting all through the 2010s that all the high paying jobs were basically ad revenue supported.
Now its AI.
Maybe AI doesn't materialize the way we want, but I'll take that bet over business as usual before every time.
I think those people are the ones in charge now and will be directing their ai resources to make even better ads.
> too many smart people were building out social networks and similar services trying to maximize ad impressions.
Honest question: where should these very smart people work then in your opinion? Nearly all suitable-looking, "smarter" ideas that come to my mind will bring less revenue/profit in expectation than, for example, letting them maximize ad impressions.
I of course wish we lived in a "more intelligent" world where this is not the case. :-(
> will bring less revenue/profit
Call me crazy, but the smartest people I know have rejected maximizing their employer's revenue and profit as their goal. Or even their own company if they go that route. Find a niche, work it, enjoy it, and enjoy the rest of your life, too.
Probably not on systems that rot democracy, freedom, and human expression. Companies like Google and Meta are net negative for humanity and the people that built these companies and continue to work for them are evil in their apathy towards the human cost of their salary.
> where should these very smart people work then in your opinion?
Anywhere else. Ads are a net negative to society and provide no value. ANY activity outside of pushing ads is a better contribution to society as a whole.
I'm not even sure that I do prefer the AI train to the social media / ad train. I certainly would if the direction was "we're throwing ourselves blindly in AI so to gradually free citizens from the need to work", but right now it's "we're throwing ourselves blindly in AI so that we can sacrifice workers for higher corporate profits".
Not to mention how they will be used by terrorists and evil people to create biological weapons and other tools of mass slaughter. Or how it will be/is being used by criminals to target vulnerable people and businesses. Or how coupled with drone warfare the world becomes an unlivable hellhole in the next decade. But Marc Andreesan said only losers are against “tech accelerationiism” on twitter so here we are.
Edit: all that and spellcheck still sucks too.
You can, right now, find out how to synthesize VX or sarin, or how to weaponize anthrax. You can and have been able to acquire this knowledge without the use of AI. Knowledge is not the obstacle to proliferation of biological and chemical weapons. It's hard to set up the infrastructure. See Aum Shinrikyo.
Sounds like you're reading into Amodei's propaganda as to why they should control AI. The information to make biological weapons for example is nothing new, it has been on the Internet for decades.
Terrorists and evil people already use the internet for the same thing.
Same people, different context.
During the social media boom: Attention is all you need
During the AI boom: Attention is all you need
> With so much human capital and resources dedicated to developing and running LLMs all the other business, research opportunities aren’t being explored and invested in
It probably feels this way if you only see headlines like this, but the LLM companies and even extending into their data center buildouts are still only a part of the economy. Investment dollars get spread around far and wide. Investors can’t even get allocations in these companies if they want to because they’re so oversubscribed.
This headline and website are a good example. If you think being a YC founder is rare and reserved for the truly elite, you may not have seen the size of YC batches lately. The number of YC founders is in the tens of thousands. Showing a list of 100 of them who joined a couple big tech companies doesn’t mean anything by itself.
It’s on the order of 1% of YC founders. YC startups don’t work out all the time and their employees and founders go work for other startups. Barely worth noting, if not for a website making it look like a big deal.
I don’t really buy the opportunity cost argument. The bets we are placing look EV-positive to me.
The 90th percentile upside is way, way higher than is priced into the current stocks, the 10th percentile looks to me like 2000 where a bunch of investors loose their shirts and then for the next few decades new companies boom on all the infrastructure that was built.
The “failure mode” for this tech is that we don’t advance much beyond Sol/Fable and the existing models just get cheaper. If there is an AI bust in hindsight it likely will look like the dotcom bust did, which is a small blip on the decades-scale GDP curve.
I fear another related possibility. With *so* much money on the table, if this new industry shrinks or collapses, it is hard to believe that it will collapse same way dot-com did, and instead investors will seek to protect their capital who knows at what cost. Given the influence they have it is not hard to imagine the impact on policies, etc. I fear that tax payers are not the winners in either case, whether this industry grows or not.
This reminds me of the office rental leases during and after lockdowns where it seemed like the idea that owners might loose money was just not even on the table and we needed people to be back at the office otherwise the value of office buildings would fall and that'd be terrible...
> Whatever you think of AI for your own work, the fact that the entire economy is betting everything on it is really concerning. It’s not just the fact that it may not work well economically speaking and would end up with a market crash, or all the negative externalities from AI development, it’s also the opportunity cost we are paying.
It's worth pointing out that "entire economy is betting everything on [AI]" you're talking about here is in the economy in the "numbers go up sense" (e.g. the stock market). It's not the economy in the providing for the general well being sense (e.g. gives people jobs so they can get housing and food).
Well estimates claim AI may contribute 50-75% of US GDP growth. Which is confusing because we're spending 2% of US GDP on AI.
I wasted like a half hour trying to find decent sources but I guess I don't know how to search the internet anymore.
I think we're betting more than the stock market on AI, even if I can't say for sure how much.
Investment from VC and wealth funds is going pretty much only to AI projects. The massive capital expenditure we are currently seeing is going to AI infrastructure (HBM has pretty much no use case outside of LLMs and a few niche industries). It’s real money being allocated to a single bet
at the current pace, LLMs are the researchers
the end goal isnt to serve a chatbot to replace google or a coding agent to build crud
Tech != the entire economy
Moonshots are concerning until you consider the alternative.
AI is the one and only possibility for saving millions (likely billions) of lives from “old age”, cancers, Alzheimer’s, ALS… you name it.
It’s a big bet. The alternative is we all die anyway: sadly after a gradual decline at best or horribly after long undignified suffering at worst.
I’ll take the economic risk.
> AI is the one and only possibility for saving millions (likely billions) of lives from “old age”, cancers, Alzheimer’s, ALS… you name it.
Says you and everyone else with a financial incentive to trick people into believing this.
If those billions were poured in research it could bring results too.
Researchers are increasingly using LLMs and other transformer based AI. My friend at a well known research institution asked Anthropic for access to their biology model (not sure if it's Mythos or some other sort of uncensored model) and recently made breakthroughs with it and other AI.
what a delightfully vague story!
Yes, I don't have their permission to dox them, so it will have to be vague for now. Still, there are many news articles outlining the same point I made, except with real examples.
I also think this is a huge problem in government. A lot of government services could be digitized via simple CRUD API/UIs and instead politicians talk about AI...
I know Ed Zitron is a dirty word on HN most days but having recently read some of his (highly editorialized) work… it seems like he’s right and the numbers just aren’t making sense.
Combine that with the fact that “good enough” is a real thing and cheap Chinese models are either there or close depending on your use case and it’s hard to see how this ends well.
Well it could end well for the general public in the sense that we get cheap, good enough AI.
Sunk cost fallacy is already too established for anything else.
Based on YC's directory, there are approximately 13,000 YC founders since YC started.
105/13000 is a very small number to focus on. This data doesn't really mean anything.
What I find more interesting is how many people have left big orgs where they had major leadership roles, to become ICs at Anthropic.
Is it interesting? They want to be rich.
Well yeah, but being a CTO at a major org isn't exactly a low-paying gig.
The interesting thing to me is that the belief that Anthropic is going achieve something like AGI appears to have really spread through SV. If not AGI, at least greatly disturb their old cushy SaaS jobs.
Also, I still do believe some people just want to work on "cool" stuff.
edit: not just CTOs and CPOs, I mean John freakin Jumper, who got the 2024 Nobel Prize in Chemistry, left his VP role at DeepMind to have no reports, and be an IC at Anthropic.
I find this whole trend extremely interesting.
> Also, I still do believe some people just want to work on "cool" stuff.
One thing I’ve noticed more and more is all the “cool” jobs require you to already be an established domain expert.
Another things I’ve noticed, and this is probably mostly just my opinions evolving, but all the “cool” stuff that catches my interest these days is not in software. Software is boring.
What catches your interest these days?
I’ve been in a bit of creative/intellectual rut the last few years, but before that there were a handful of things that intrigued me in the last 5-10 years.
Most recently, I became interested in the field of photonics/optics, particularly in the field of novel optical communications research.
I’ve had mild interests in things like Biology and History since I was a child. The latter particularly hooked me a few years back.
Neuroscience is a pretty cool one as well. The human brain is of course one of the most fascinating pieces of hardware in existence and understanding it (to whatever level we currently can) is incredibly interesting to me.
At one point I thought about going into security research, but now that everyone and their mother is headed that direction, it’s become less feasible.
Unfortunately I don’t even have a bachelors degree, and most of the fields are research heavy and not very accessible to hobbyists, so my actual interaction with these things remains relegated to reading papers/pop sci articles/etc.
I really don't think it's AGI. From living in the valley, it looks a whole lot more like giant piles of cash and "prestige".
> If not AGI, at least greatly disturb their old cushy SaaS jobs.
I mean, this is happening right now.
I doubt there are few people talking to customers and potential customers that haven't run into the "look what I rolled last week with Claude" only for it to be legitimately impressive. The idea of technical moats is evaporating in front of our eyes.
The biggest AI bubble seems to be forming within the minds of the people working in SV itself, judging by the talk about the inevitability of AGI, the strong language around fomo (permanent underclass and all that) and the close proximity of the whole ecosystem there.
So wether it's belief in AGI or a desire for building cool stuff, as an outsider looking in, it seems to be informed by the same echo chamber of thought.
I think it is more than just money. Having a hand in steering the ship that they think will change the world is pretty attractive. I came out of retirement just to watch the spectacle.
I would do that in a heartbeat if I was in that situation.
Having a major leadership role at a big company is kind of a slog. Getting a chance to go back to a fast growing startup as an IC in a way that will only benefit their career (ex-Anthropic or ex-OpenAI makes your resume gold plated) is an easy choice. They can go back to any job they want after this.
I think this might only be shocking to some people who have forgotten that many people in these positions actually enjoy working on hard things and being surrounded by other like minded people. If you’ve only worked in po-dunk startups or other companies where leadership looks like a Game of Thrones style fiefdom building game it’s harder to imagine executives who are smart, capable, and enjoy working hard. It’s pretty common in Silicon Valley despite all of the sneering here.
I disagree. I'd immediately have a ton of questions for most ex-OpenAI and ex-Anthropic people as my default position would be "poor business/economics judgement in role, chose to ride a hype cycle, now sat in front of me for a reason".
I'm not saying it's toxic to have it on your work history, but it's far from gold plated. They'd have work to do for me, and for many, many people in the industry, to prove that I'd want to work with them.
You are not in a position to hire them, let alone at a big interesting company. Forgive my bluntness, but your opinion doesn’t matter to their situation.
When I say their resumes are gold-plated I mean at the companies they want to work at.
Are you a hiring manager?
I think most of HN can't logically arrive at this conclusion because they don't exactly buy into the premise that
(1) (digital) AGI is a real concept, and will be achieved imminently
(2) AGI will be world (and possibly more) changing.
(3) Anthropic will be the one to do that, or at least capture a large part of it
If you buy into all of these, which I'm assuming a lot of SV does at this point, then its kind of a no brainer to go to OpenAI or Anthropic. And optics wise it seems like Anthropic is "winning" right now, so thats the lab with the talent flow.
If Anthropic actually creates a singleton, contributing to that is way more impactful than being CTO at some random company (though I may disagree that said singleton would treat equity holders any differently).
Do you have examples? Because besides the other reasons mentioned about huge earning potential and being part of an important moment in history, the not-so-hidden secret is that any management position below C-level and above small team manager is usually extremely shitty. You are responsible for cat herding but rarely get to make the decisions in the first place. There are very few people IMO who desire or excel at "mid-high level" management positions (again, levels from director to anything below CTO) - I've seen people who enjoy and are good at leading small teams, but people at the director level are either there because it's lucrative or with the hope to get to CTO level ASAP, but otherwise it's one of the worst jobs in the "stress to moments of enjoyment" ratio.
So I could imagine tons of people in those roles who would love to be ICs again if there was also a chance at outsized wealth.
Participating in the creation of the Singleton is going down in the history books. Major leadership role at BigOrg is not.
You're surprised that people are trying to capture a bag that could possibly lead to multigenerational wealth in the tune of tens to 100s of millions?
I'm not surprised, but what is surprising is we're now in a period in which the mask is slipping and no one seems to care enough to try to put it back on.
For all the hubris and arrogance of the tech scene, and especially the Bay Area's, founders at least pretended to pursue a greater calling than money by leaving cushy gigs to pursue a new venture they were passionate about. Sure, that only lasted a handful of years before it became a gold rush to move to the Bay for YC and pursue the startup-to-acquisition pipeline, but they at least pretended to care.
The naked pursuit of wealth at the cost of potentially tearing apart society is mildly alarming because the scale it's happening at is so big.
It seems like different people? The early tech scene was defined by people who got into tech because they loved it, jobs were paying under six figures. Then it became a gold rush, and the tech scene became defined by the newcomers who were after cash first, tech second. It's not that the old timers gave up on their ideals (well, some did), they were simply supplanted by larger numbers of people following the money.
In my opinion there is no "slipping mask", merely a changing of the guard that always happens when massive piles of generational wealth are at stake.
Nah, we knew back in the 90s that the tech industry had open disdain against American workers (pressuring congress to expand immigration to depress wages) and children (supporting charter schools while attacking children (yes when you attack public schools you're attacking the only people that benefit from it: children)). Not too mention all the horrid, extremely hostile business pursuits we saw from companies like Oracle + MSFT + Apple.
SV has always operated on exploitation of their workers, they just hid it well from the public.
Having been there i have to respectfully disagree. Probably because, as far as i can tell, you are warping their intentions to be what your intentions would be if you did what they did.
I just don't understand how to connect this analysis to what's happening. What OpenAI and Anthropic both say, loudly and repeatedly, is that they're working on one of the most impactful technical problems in the history of humanity. The reason they have lots of money is that lots of investors believe they're right.
Perhaps you've seen under their mask, and you know they're scamming the investors. After the NFT bubble it's a plausible hypothesis. But I don't know what it means to say that the mask just isn't there and it's naked pursuit of wealth.
I sometimes think, speeding up the Red Queen hypothesis to accelerate Cambrian Explosion by printing money from the top so that high-agency people would develop new products and services to capture more of the newly created money is a good thing?
Probably because of stock options. Assuming the bubble bursts after the vesting period, they stand to make a looooooot of money on paper.
Right, at this point its ~150 person cohorts, 4x a year. 105 is a drop in the bucket.
I wonder what the income distribution is like
90% of the wealth at the top 1%.
The question I have is why are these companies hiring these people and what does it say about their hiring practices and the amount of capital they’re poorly spending?
It’s exceedingly unlikely that any of the people who were working on YC startups previously have any real professional experience with any of the following: slurm, collectives, NUMA systems, RDMA, compilers, systems programming, general HPC performance estimation or measurement, CUDA or ROCM or any kind of GPGPU/accelerated computing. But that is the core business of both of these companies.
I’m not surprised that these companies are well funded and hiring a lot of people. I’m surprised that they chose to hire the people who were previously making “Uber but for dogs” gimmick apps and not just hollowing out the HPC specialists from national labs.
The top % of YC founders are legitimately quite talented, and high agency/entrepreneurial talent goes a long way in these companies, even if its not in deep infra or research
People ask the same question of why YC funds yet another Uber for dogs or a button on the Touch Bar that cost $10/mo to help join a meeting faster.
They invest in people more than ideas, so you’ve got, at least in many cases, people with good pedigree and skills (age adjusted anyway) building on stupid ideas, but that are eminently employable.
Obviously there are other factors, I’m not really trying to defend anything but just point out that there are legit reasons why someone impressive enough to get into YC would also be impressive enough to get a good job. It’s not like it’s random founders off the street.
> but that are eminently employable
This is what always stood out to me. Founders talk about the risks they take as their major legitimizing force. But what risk? They’re the types who can (mostly) skip an interview and go direct into a six figure job - because of the sales and founder networks they develop.
Startup founders with no risk doesnt sound like a recipe for great companies.
YC founders aren't the ones taking risks. The risk comes from pouring your life savings into a company and not getting into any incubators and never getting any VC funding and then having no savings and nothing to show for it.
The actual AI part of AI startups are at constant risk of commoditization. Where the companies will survive or fail is how they manage to attract customers and lock them into their offerings so they can't move away to the next competitor.
Someone with experience running a startup that has no moat probably does have some relevant experience in that area.
It isn’t that long ago that AI was heavily gatekept and had some serious barriers of entry, due to its origins from academia.
I figure most of these are working on products, rather that deep frontier research.
I actually have experience with that stuff, “old school” deep learning and ML as well. Is that something worth joining the fray with I wonder? Or is it as you point out really only the richly connected “locals” that are recruited?
Not to try to grab any pity, but I have pretty deep experience in the GPU optimization space and have not been able to even get an engineering screen even with referrals.
I have a sinking feeling that AI models have gotten so good internally, and/or the GPU guys in their labs have found a good enough processes, that AI is doing the research behind the doors right now (it should be at least capable, at heuristic optimization problems that are benchmarkable), while the hard part of their business is the political game: convincing other companies to buy into AI, integrations, trying to gather more data where they can.
> I’m not surprised that these companies are well funded and hiring a lot of people. I’m surprised that they chose to hire the people who were previously making “Uber but for dogs” gimmick apps and not just hollowing out the HPC specialists from national labs.
Culture. The Uber for Dogs idiots will gulp down the kool-aid like they’re dying in the dessert.
> The question I have is why are these companies hiring these people and what does it say about their hiring practices and the amount of capital they’re poorly spending?
They are hiring for sales.
Silicon Valley is not about having specialists or knowledge. It's about having connections and charisma.
Is this causing massive problems for us as a society? Of course, but no one seems to want to do anything about it, so here we are.
I mean, what are you suggesting “someone” should do about it? The companies can spend their money however they want, and if they want to hire a bunch of product people from YC startups nobody can stop them.
I’m just surprised they need this many of them. What do they do all day? And who is left to manage their hundreds of thousands of servers?
A lot of the AI game is still integrations and getting entrenched in processes. It has the double whammy that you get to train on their data (not talking about ALL enterprise contracts - but cursor pretty explicitly has this business model. So I'd imagine some enterprise contracts have this.)
> What do they do all day?
Breeding product ideas? ChatGPT, but for dogs?
> And who is left to manage their hundreds of thousands of servers?
They learned from Elon how to get by with a skeleton crew?
I’m just surprised that the people actually keeping the business operating and the tokens flowing don’t resent these product people for sitting around all day coming up with ideas for the same TC package as the people putting out ops fires and bringing new capacity online all day every day. I feel like the actual engineers are probably getting shortchanged.
As with anything else, they may resent, but the check still cashes at the end of the week.
> What do they do all day?
Same as they were doing in their startups: burn mon..., eh, tokens.
> And who is left to manage their hundreds of thousands of servers?
AI, of course.
Probably because they want snake oil salesman
Why spend to resources to invent AGI when you can just gaslight the whole world by telling everyone that LLMs are already better than people?
Effectively, industry roles are a tiered system which determines access to the cashflow. Going the YC founder route is a much faster and more efficient way to secure a high tier than climbing the SWE ladder. The whole thing resembles a kind of mini class system, with high tiers granting access to generational wealth, and low tiers being better off than the average non-industry Joe. Once you're in the high tier, there's no way to fall anymore, you just move wherever the cash is being pumped in, collecting it.
I’ve seen many people climb the SWE ladder or build a YC company (having done both myself), and trust me the founder route is not the most efficient. You only believe this because you didn’t see the 2 failed startups and 10 year journey grinding 80h weeks almost running out of cash a few times.
Jensen Huang himself says nobody in their right mind should start a company https://www.cnbc.com/amp/2024/05/11/jensen-huang-i-didnt-kno...
> Jensen Huang himself says
Not that I disagree, but it doesn't refute the original argument. It could even support it, class protecting their own from more people figuring out there's no clothes on the emperor.
What part of "start a company" requires "grinding 80h weeks"?
Sounds to me as inability to delegate. Or implicit in the [acquire VC, aim for high growth] mindset as if that were the only path to success.
Companies can grow organically. With a positive cash-flow early on & reasonable work/life balance for founders & employees alike. Heck, there's even non-profits.
> Companies can grow organically.
Yes they can, but unless you strike luck, they take many years, often decades of work to grow to a level where you can retire and enjoy the money.
With FIRE and similar concepts being pushed around everywhere, and the plight of the masses to keep their jobs and homes, I see where that desire for "make money fast" comes from. And from a game theory mindset - it's not wrong. (That is also a bit related to the explosion of gambling and gambling-alike stuff such as "prediction markets"...)
> Heck, there's even non-profits.
Non-profits often enough run on conditions bordering on (self-)exploitation.
Except that he did start a company so his comment is really saying you have to be crazy enough to start a company
"Lottery winner says you should not gamble"
Nobody in their right mind should start a company up to some equilibrium. As a society we should fund the appetite of experimental founders to put more pressure on quality than what natural conditions might allow.
This. I saw way more people build a few millions of retirement nest egg reliably in 10-20 years, by working on a boring good swe job and manage their investment a bit better. It's a much safer route.
survivorship bias
> Going the YC founder route is a much faster and more efficient way to secure a high tier than climbing the SWE ladder.
YC has funded over 5,000 companies. If you assume 2-3 co-founders per company, that's more than 10,000 to 15,000 people. The vast majority of these founders aren't producing "generational wealth" outcomes. There's no glamorization of the companies that shut down, the ones that are scraping by, and the ones that get their founders a normal job, but those are the far more likely outcomes, especially in the more recent spray-and-pray batches.
This is not even close to true. I work at a big tech company and we reject applications from YC founders every week. Most of them are broke with failing companies and would love a cushy mid-level engineering or management job at a faang. People are biased by the success of the top few founders but 99% of them have worse outcomes than the average corporate engineer.
This is a very hot take. Being a founder might confer you some career advantage [1] if you're one of the few to make it to series A and you grow a large team, but if you're like the vast majority of founders and never make it past seed (or realistically, to seed), your network is the alpha and omega of any subsequent job search. It's a bit like saying that being an NBA starter is a good way to get a job as a basketball coach.
Point being: don't start a startup if your goal is to get a job. Just get a job.
[1] Note that I'm not arguing about experience -- you can gain a lot of experience as a startup founder, but that experience is rarely directly marketable. Also, most startup founders are completely clueless when they start, so "a lot of experience" is a relative term.
This analysis doesn't actually prove the title, as it only considers founders that have gone to OpenAI or Anthropic, however even if it did it isn't particularly suprising.
Sam Altman was the president of YC, so it's not particularly suprising that he has hired lots of talent from YC - these are people that are pre-qualified because he has seen their work before outside of a job interview (better the devil you know than the devil you don't).
This happens everywhere - person leaves company A to company B and then poaches the best talent from company A. Obviously in this case the talent are founders rather than staff members, but still.
YC has tens of thousands of founders, more are probably at Google and Facebook than Anthropic and OpenAI. I’m not sure a sample of 105 founders shows… anything?
This. the comments here talk as if YC is this secret cabal of upper elites when they're probably one of the least selective good accelerators
Is there a selection bias?
SELECT * FROM yc_founders WHERE employer IN ('OpenAI', 'Anthropic');
If there are 7k founders [1], the graph only shows 1.5% of the people from YC.
[1]: https://www.ycombinator.com/investors
I hope some people remain to build the services that we still use more every day than LLMs.
Edit : and money. I guess most founders follow the money here.
Based on typical announcements, they only get acquired to serve their customers even better. The selfless unsung heroes of our time...
The word "mostly" doesn't belong in this title, because YC has funded over 5,000 companies, and this page accounts for 100 founders.
A successful unicorn founder once said a billion dollar company is just made up of a bunch of people who could each have started a $10m company by themselves.
I guess you can extrapolate that and say a trillion dollar company is made of a bunch of people who could have started a billion dollar company.
I wonder what they do, are they joining as engineers or leadership? because Anthropic has Member of Technical Staff role only.
Would be curious what some of the VP+ people are building inside Anthropic if they joined as engineers
The site infers that 105 people are working at the two companies. What’s the denominator though?
After 20 or so years of YC, with multiple catches per year, and the insanely high failure rate of start ups, there must be a lot of former or floundering founders.
This graph biases outcomes (A| O) it would be interessting to have an others.. so we can view the whole picture
According to the stats in this article https://www.ycombinator.com/investors, we can estimate 98.5% of the founders are not in OpenAI or Anthropic.
“Emmet Shear, Twitch, OpenAI - former”. I would not be surprised if the whole site was made for this punchline.
this has gotta be one of the most unclear and misleading sankey diagrams i've ever seen
this has got to be one of the most unclear and misleading sankeys i've ever seen
The wheels on the bus go round and round, round and round, round and round...
they are totally focusing on making those their internal products so they dont need sell ai to customers anymore. that's the future
It’s a big club and you ain’t in it
I bet this easy cash coke binge will last forever!
IMO we are in the era of “Final Companies” and the appeal of being a founder these days just isn’t great compared to the entrepreneurial boom of the 2010s and 2000s. The risk to reward ratio has never been worse.
If Product-Market fit for small companies take 3-4 years and AI can streamroll you by coming from behind, it makes sense to join the big boys rather than trying to finetune your creation in the hope of making something big.
They abandoned the dream of democratizing sectors of the market by ignoring regulations to democratizing knowledge by ignoring copyright.
alternate title:
Where are YC founders now who went to openai and anthropic ? OpenAI and Anthropic, mostly
The fonts the layouts it all screams claude at me.
I still can’t put a finger on it. I’ve seen real people use these fonts and layouts yet theirs look original.
Whoever finds an explanation for this solves AGI (/s)
It's a set of properties/attributes commonly referred to as "style" (generic, not the CSS one). Or call it look & feel.
So you're saying the site screams "Claude's style!" to you. Not too different from "I know a WordPress site when I see one".
The sepia-tinted color scheme and the way LLMs overuse rounded corner tiles to the absolute death is part of it. Really what it comes down to, is that these LLMs have a design vocabulary of like 15 traits. None of the websites they generate have every single trait, but they are composed of some subset of 10 such traits, so it's still overwhelmingly obvious the moment you open the page, even though each one is slightly different. Humans might use 1~5 of those traits, but they'll also be mixed in with a much wider set of traits with more individuality to them.
I tried to list all the traits I could think of:
- Orange/beige-ish colors - Rounded corners - Cards with a thin border - A thicker colored border on the left of cards - Serif font for headings - Monospace fonts for small text - Headings that often have an unnecessary subheading / pre-heading - Little badges, often with a "status indicator" dot on the left - Obviously LLM-generated text / language
I'm sure I'm missing many but the above are dead giveaways!
That left side border is a plague…
The copy is usually a bit of a giveaway too, ”explore your future path to greatness and experience the defining divider between those who can and those who can’t” or as most humans would write the header ”plans”.
Most AI-generated sites have coopted the Inter font, and usually they go for a dark colour scheme, the rounded tiles, a container representing a macOS-styled terminal with code, and, the thing that annoys me the most, divs that fade is as you scroll.