Number of employees isn't a straightforward calculation.
- A lot of companies have part time, seasonal and contract workers (e.g. Amazon and Walmart). For some this even makes up the majority of their workforce.
- Some companies operate on a franchise model, and corporate employees are very limited. For example the 414K employees listed for Marriott aren't directly employed by them. But then you don't apply the same logic to McDonalds, where you are only counting the corporate employees.
- Holding companies (like Berkshire Hathaway) show the full employee count of their portfolio companies even through they may only own a small share. And again the logic isn't consistent because you are only counting 714 employees for Welltower, which is also a holding company.
Collecting incomplete or incorrect data and adding a hunded disclaimers at the end of the report is standard LLM behavior. Actually valuable analysis of this sort needs a lot more human effort.
> Employee counts come from company reports via Yahoo Finance, with Wikipedia used where figures weren't available. They're not necessarily from the same date, and companies don't all count employees in exactly the same way.
> As the McDonald's and Welltower examples show, a company's reported workforce doesn't necessarily represent everyone whose labour contributes to its business.
Well, alright. I'm just not sure how useful this is. There's no way to normalize them but the data's presented as though they were. You also sum employees across sectors which compounds the errors.
The blog post OP linked puts more context into it, and I think it also makes it clear that they’re not trying to make an apples to apples comparison. It’s a tool that can aid in visualizing how different companies can be structured.
Berkshire reports employees across its consolidated subsidiaries, while Welltower's 335,000+ operating-partner employees work for separate companies. I'm using reported employee counts consistently, but you're right that different corporate structures make comparisons imperfect.
> Collecting incomplete or incorrect data and adding a hunded disclaimers at the end of the report is standard LLM behavior. Actually valuable analysis of this sort needs a lot more human effort.
Or it's classic modern journalism that gets clicks. LLMs are just mimic'n human behavior.
At a glance, I think the author used S&P sector categorizations. They were set in 2018 and intentionally split the mega caps/FAANG/MAG7 across multiple sectors based on some hand waving about “primary” revenue source at that time.
From a finance perspective I think they’re a _little_ weird but functional (NVDA moves with semis, MSFT with software, etc) whereas from a tech perspective I’d say very weird until you learn about the backstory and what motivated it.
Definitely some strange inclusions and exclusions. "Information technology" has NetApp but not Everpure, even though they have the same market cap and are in the same space. Qnity and Jabil are included with half the market cap.
Curious why the comment by the poster got flagged but not the full post. I like when the poster includes a comment describing the content and reasons why he liked it.
I dislike that. IMO, if the content cannot stand on its own, it did not need to be posted. A comment that adds something to it, to initiate a discussion, is fine. But a comment that just says what that post is? That is neither interesting nor engaging.
Such comments also encourage the idea that people "own" the threads they start. I've seen people get mad when people take the conversation in a direction they did not want, or add instructions telling people what to talk about and putting constraints around discussion. Such things stifle interesting conversation.
So I land heavily the opposite direction. Just post, and leave it at that. If the community finds it interesting, discussion will ensue.
Now, Show HNs are their own thing. An initial explanation is absolutely appropriate for a show HN. But it should be part of the post, not a comment.
I suspect it was one of the content review bots. In my first Show HN post, I handwrote the comment, but using a bulleted list style suggested by an AI agent, and that got flagged. I reached out to the mods, and they told me to switch to prose, which passed muster.
1. The legend uses different icons than the visualization
2. Why would you purposefully make every bubble empty? Change the scale to 1000 employees otherwise it’s deceptive. 10k employees is arbitrary.
1000 or 10k per stick figure is equally arbitrary.
What they actually need to do is scale the size of the stick people correctly.
For example, one meaningful measure might be to scale it so that one stick person fits inside a circle with area corresponding to the amount of money they have to earn in total for that number of people to cover all their living expenses and some amount of savings left over. I am sure some statistical bureau has some meaningful dollar amount of some sort that makes sense to use.
Network hardware, especially high performance fabric is every bit as important as gpus and ram.
People seem to think Nvidia’s gpus are the be all and end all of enabling large models but there isn’t a large model out there that runs on a single gpu.
Broadcom sells things like pcie switches. They allow you to plug your in about a hundred gpus into one big switched pcie plane that can all communicate at full pcie bandwidth to each other. Nvidia has nvlink which is similar. You need this technology to enable large models and they sell for even higher margins than the gpus and ram.
Offtopic: It's infuriating how I clicked the site and it wasn't loading. Scrolled the comments looking for a "hug to death" mention but found none, then realized I live in Spain and today is football day...
I'll check the site when a football private company allows me to.
Welltower and Palantir are doing an amazing job! A comparison with NVIDIA in terms of market cap per employee isn't realistic here, but perks to them for the utility that each employee is bringing to the table.
> Welltower and Palantir are doing an amazing job!
It’s funny to read that from this becuase I read it as these companies are so good at exploiting capitalism that they can make this much profit and benefit so few actual people in the country. It’s like they’re a leech on the country, rather than companies to be proud of.
If we have hundreds of hyper-successful companies but we have people going into medical bankruptcy for a cancer diagnosis, I’d say America is failing at taxing these companies enough to fund what actually matters in society.
Number of employees isn't a straightforward calculation.
- A lot of companies have part time, seasonal and contract workers (e.g. Amazon and Walmart). For some this even makes up the majority of their workforce.
- Some companies operate on a franchise model, and corporate employees are very limited. For example the 414K employees listed for Marriott aren't directly employed by them. But then you don't apply the same logic to McDonalds, where you are only counting the corporate employees.
- Holding companies (like Berkshire Hathaway) show the full employee count of their portfolio companies even through they may only own a small share. And again the logic isn't consistent because you are only counting 714 employees for Welltower, which is also a holding company.
Collecting incomplete or incorrect data and adding a hunded disclaimers at the end of the report is standard LLM behavior. Actually valuable analysis of this sort needs a lot more human effort.
Fair point, I was aware of that and mention it in the accompanying blog https://bigwhalehunting.com/blog/workforce/
Towards the bottom:
> Employee counts come from company reports via Yahoo Finance, with Wikipedia used where figures weren't available. They're not necessarily from the same date, and companies don't all count employees in exactly the same way.
> As the McDonald's and Welltower examples show, a company's reported workforce doesn't necessarily represent everyone whose labour contributes to its business.
Well, alright. I'm just not sure how useful this is. There's no way to normalize them but the data's presented as though they were. You also sum employees across sectors which compounds the errors.
> surprised me
> I was aware of that and mention it
I'm confused.
The blog post OP linked puts more context into it, and I think it also makes it clear that they’re not trying to make an apples to apples comparison. It’s a tool that can aid in visualizing how different companies can be structured.
Except this visualization is literally an apples to apples comparison
I imagine Tesla has a vast amount of contract workers with all the construction they do.
Berkshire reports employees across its consolidated subsidiaries, while Welltower's 335,000+ operating-partner employees work for separate companies. I'm using reported employee counts consistently, but you're right that different corporate structures make comparisons imperfect.
> Collecting incomplete or incorrect data and adding a hunded disclaimers at the end of the report is standard LLM behavior. Actually valuable analysis of this sort needs a lot more human effort.
Or it's classic modern journalism that gets clicks. LLMs are just mimic'n human behavior.
No one is calling this journalism especially not me.
Alphabet and Meta are misclassed, they are information technology.
Tesla should (probably) be in industrials, as a car and battery tech company.
Amazon is hard to class, you might check if they break out to subsidiaries for AWS versus consumer retail. Part is definitely competing with Walmart.
At a glance, I think the author used S&P sector categorizations. They were set in 2018 and intentionally split the mega caps/FAANG/MAG7 across multiple sectors based on some hand waving about “primary” revenue source at that time.
From a finance perspective I think they’re a _little_ weird but functional (NVDA moves with semis, MSFT with software, etc) whereas from a tech perspective I’d say very weird until you learn about the backstory and what motivated it.
Definitely some strange inclusions and exclusions. "Information technology" has NetApp but not Everpure, even though they have the same market cap and are in the same space. Qnity and Jabil are included with half the market cap.
Agreed, Datadog is missing despite being in the S&P 500 as well
Curious why the comment by the poster got flagged but not the full post. I like when the poster includes a comment describing the content and reasons why he liked it.
I dislike that. IMO, if the content cannot stand on its own, it did not need to be posted. A comment that adds something to it, to initiate a discussion, is fine. But a comment that just says what that post is? That is neither interesting nor engaging.
Such comments also encourage the idea that people "own" the threads they start. I've seen people get mad when people take the conversation in a direction they did not want, or add instructions telling people what to talk about and putting constraints around discussion. Such things stifle interesting conversation.
So I land heavily the opposite direction. Just post, and leave it at that. If the community finds it interesting, discussion will ensue.
Now, Show HNs are their own thing. An initial explanation is absolutely appropriate for a show HN. But it should be part of the post, not a comment.
Even so, not sure it qualifies for flagging
Probably because posting summaries of the submission is discouraged by the HN moderators, e.g. https://news.ycombinator.com/item?id=44946681 or https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
I suspect it was one of the content review bots. In my first Show HN post, I handwrote the comment, but using a bulleted list style suggested by an AI agent, and that got flagged. I reached out to the mods, and they told me to switch to prose, which passed muster.
1. The legend uses different icons than the visualization 2. Why would you purposefully make every bubble empty? Change the scale to 1000 employees otherwise it’s deceptive. 10k employees is arbitrary.
1000 or 10k per stick figure is equally arbitrary.
What they actually need to do is scale the size of the stick people correctly.
For example, one meaningful measure might be to scale it so that one stick person fits inside a circle with area corresponding to the amount of money they have to earn in total for that number of people to cover all their living expenses and some amount of savings left over. I am sure some statistical bureau has some meaningful dollar amount of some sort that makes sense to use.
Just make it so that the companies that employ the most per market cap are filled, so we can see how empty the rest are.
That rather appears to be what they have done, given that the company with lowest market cap per employee has a circle full of people.
There's a linear relationship between revenue and employees across sectors. The desired visualization is the linear fit where outliers can be seen.
Market cap is a function of earnings and earnings growth.
TIL that Broadcom is a 1.6T company.
Network hardware, especially high performance fabric is every bit as important as gpus and ram.
People seem to think Nvidia’s gpus are the be all and end all of enabling large models but there isn’t a large model out there that runs on a single gpu.
Broadcom sells things like pcie switches. They allow you to plug your in about a hundred gpus into one big switched pcie plane that can all communicate at full pcie bandwidth to each other. Nvidia has nvlink which is similar. You need this technology to enable large models and they sell for even higher margins than the gpus and ram.
Very cool! It would be nice to have a constant dollars option. I always like to be able to correct for inflation.
Offtopic: It's infuriating how I clicked the site and it wasn't loading. Scrolled the comments looking for a "hug to death" mention but found none, then realized I live in Spain and today is football day...
I'll check the site when a football private company allows me to.
how do I zoom? there's no + and - buttons and I don't see people icons
I think revenue per employee is more interesting. By this measure, tech companies are minuscule compared to energy and real estate companies.
https://i.imgur.com/4DPs1Io.png
https://i.imgur.com/LgFx831.png
What will this look like after the coming AI investment crash/growth stagnation inflection point?
It’s gonna be a wild ride to 2030.
We have less than 4 years left to prepare before 2030 [0]
[0] https://news.ycombinator.com/item?id=47558516
Counterpoint: fortunately we still have more than 5 years to prepare for 2032.
Welltower and Palantir are doing an amazing job! A comparison with NVIDIA in terms of market cap per employee isn't realistic here, but perks to them for the utility that each employee is bringing to the table.
> Welltower and Palantir are doing an amazing job!
It’s funny to read that from this becuase I read it as these companies are so good at exploiting capitalism that they can make this much profit and benefit so few actual people in the country. It’s like they’re a leech on the country, rather than companies to be proud of.
If we have hundreds of hyper-successful companies but we have people going into medical bankruptcy for a cancer diagnosis, I’d say America is failing at taxing these companies enough to fund what actually matters in society.