I'm incredibly skeptical of this idea that AI is creating such efficiencies that employers can hire fewer engineers.
Firstly, there's the massive confounding factor of Covid, the stock market went crazy, companies went on crazy hiring sprees etc. and the tail of that bullwhip effect is clearly still putting downward pressure on hiring from organisations that overhired.
But secondly, are we seriously saying that in the last 2 years, relatively slow moving companies adopted AI LLMs to help coders, integrated them into our work flows, and saw the results of those productivity gains in business outcomes?
I think it's unlikely. I think it's much more likely that CEOs love to watch where the crowd is going and then run to the front and shout "follow me". You don't actually need to have productivity gains for shareholders to reward you for saying how this is going to boost your margins and cut your costs. And this is even more true for companies like Salesforce for whom "AI" is a product they're selling. Marc Benioff isn't actually doing "AI is great I'll fire all my engineers", he's saying "AI is great, come buy Salesforce's AI products!". As for Microsoft, they employee almost a quarter of a million people, laying off 6,000 is a drop in the ocean, that scale of layoffs happen frequently at companies the size of Microsoft.
It's very much more just vibes than real data driving this. "<CEO who sells product> says product will cure cancer".
The underlying truth is that the competitive environment hasn't changed, if you can hire fewer engineers to do the same job, great, but your competitors are going to hire more engineers and out-compete you.
Yeah I agree with you. There is also a subtly incorrect belief people have that employees are just costs. Therefore increased efficiency = fire people = increased profits.
The more accurate frame is that employees produce more value than their costs, so each employee is actually a profit producer. If you can increase their efficiency, then they produce more profit. Firing profit producers decreases your profit, not increases it.
"if you can hire fewer engineers to do the same job, great, but your competitors are going to hire more engineers and out-compete you." - this is correct in aggregate across the market. Of course any individual company may have other constraints that makes hiring additional people unviable. I think this is the common mistake, its easy to look at an individual company and believe that the constraint applies to all companies simultaneously.
I agree with you but I think at a certain point with large teams, the marginal benefit of hiring n+1 becomes negligible and then dips negative.
I've never worked for a company with thousands of devs but I imagine at a certain point the cost of simply collaborating and managing those devs becomes pretty high too.
Yeah there are certainly diminishing returns. It's just that if they had N employees today, it's very unlikely they were at the top of their s-curve, such that N+1 employees is now net negative. Unless the business was already very mature. Most companies are expecting to grow revenue.
So it would be the exceptional case that firing employees because they were suddenly more productive makes sense.
What's more likely is they were already overhired and the new found "efficiency" is the excuse to resize.
Edit: btw this comes up IRL for me all the time in my work. Companies default play is to cut expenses to hit desired margin. It's much more beneficial in most cases to keep the team but increase their efficiency and handle more revenue. There are diminishing returns to profit by cutting back.
It isn’t that simple because your profit upside is usually finite. Say you run a lemonade stand in a small town. Number of employees can’t really exceed some threshold set by the amount of lemonade customers a given day in that town. Even if the employee is the profit producer. It isn’t an infinite money machine and money has to be made available to seek.
Yeah that's correct, I mentioned "Of course any individual company may have other constraints that makes hiring additional people unviable". One such constraint would be the total market size.
The rest of us non-AI-whizzes combined literally wrote 100% of the functioning open source code that the AI-whizzes steal and transform to an inferior product using Rube Goldberg agent setups.
> The rest of us non-AI-whizzes combined literally wrote 100% of the functioning open source code that the AI-whizzes steal
Open source is hard to steal. It's intended to be taken for free with some license limitations. Second, most giants are already dead. Their work was used for free and that was/is considered normal. Third, many 'AI-whizzes' contributed to open source for long time.
In any case AI will be used on open source. For example to find bugs and backdoors at scale.
Even if AI-whizzes have contributed to open source for a long time, which I question, it does not give them the right to take everyone else's code. Neither does using "AI" on open source.
Moreover, many "AI" open source code bases are a mess and probably wouldn't even contribute in a positive manner to training.
the top of the pay scale were elite ai labs such as Openai, as well as hedge funds such as Jane Street that are also betting heavily on machine learning. In this tier, median pay exceeded $400,000 a year. Below that were tech giants including Alphabet, Microsoft and, until recently, Meta, where median pay was closer to $300,000. Experienced developers at most other companies earn much less. Their median was around $180,000 (see chart 2).
—-
A median of 180K is mostly definitely raking it in compared to the median of all US employees.
I’m well over double the household median income in my metro area and while I don’t feel like I’m raking it in, I guess I am when compared to others.
It's also worth noting the take-home pay is closer than that, due to progressive taxation. E.g. in California[0]:
$180k/year gives you $118,970 after tax
$300k/year gives you $186,880 after tax
I.e. the gross jump from $180k -> $300k is 67%, but the net jump is 57%. You have to increase the gross more to attract people, because take-home pay is what matters to them when switching job.
Microsoft really getting lucky with being grouped in with FAANG in this article, but everything I've read says they pay far lower than most top tier tech companies, no?
It's certainly true by comparing "senior" at G/Amazon/Apple to Microsoft, but is there level skew that compensates for this?
Combine the level skew with stock appreciation and high employee count, and there are a lot of Microsoft engineers raking in that kind of money. So lots of people think of it as competitive even if there are also many working there for less.
I’ve been making over $500k a year since 2020 working fully remote; everyone I know with 7+ YOE at big tech or unicorns is also making 500k TC which often appreciates to ~700k+ with the current market.
We’re just writing dumb gRPC services that use Postgres. I work probably 30 hours a week and still get awards, bonuses etc.
Those are all definitely not asking about writing little services in postgres on their job apps. That tier job posting today looks more like a post for a tenure track ML faculty position.
I’ve worked at half of these places and they’re all doing yaml massaging and grpc sculpting. Not much more than basic KTLO with some regularly scheduled feature work and reliability heroics.
Well that is not what that tier job description is describing today at least. Maybe in the past or even still today if you got in early. Hard to imagine so much weight being put on ML knowledge and skills in the job app if it isn’t getting used in the role at all.
I'm incredibly skeptical of this idea that AI is creating such efficiencies that employers can hire fewer engineers.
Firstly, there's the massive confounding factor of Covid, the stock market went crazy, companies went on crazy hiring sprees etc. and the tail of that bullwhip effect is clearly still putting downward pressure on hiring from organisations that overhired.
But secondly, are we seriously saying that in the last 2 years, relatively slow moving companies adopted AI LLMs to help coders, integrated them into our work flows, and saw the results of those productivity gains in business outcomes?
I think it's unlikely. I think it's much more likely that CEOs love to watch where the crowd is going and then run to the front and shout "follow me". You don't actually need to have productivity gains for shareholders to reward you for saying how this is going to boost your margins and cut your costs. And this is even more true for companies like Salesforce for whom "AI" is a product they're selling. Marc Benioff isn't actually doing "AI is great I'll fire all my engineers", he's saying "AI is great, come buy Salesforce's AI products!". As for Microsoft, they employee almost a quarter of a million people, laying off 6,000 is a drop in the ocean, that scale of layoffs happen frequently at companies the size of Microsoft.
It's very much more just vibes than real data driving this. "<CEO who sells product> says product will cure cancer".
The underlying truth is that the competitive environment hasn't changed, if you can hire fewer engineers to do the same job, great, but your competitors are going to hire more engineers and out-compete you.
Yeah I agree with you. There is also a subtly incorrect belief people have that employees are just costs. Therefore increased efficiency = fire people = increased profits.
The more accurate frame is that employees produce more value than their costs, so each employee is actually a profit producer. If you can increase their efficiency, then they produce more profit. Firing profit producers decreases your profit, not increases it.
"if you can hire fewer engineers to do the same job, great, but your competitors are going to hire more engineers and out-compete you." - this is correct in aggregate across the market. Of course any individual company may have other constraints that makes hiring additional people unviable. I think this is the common mistake, its easy to look at an individual company and believe that the constraint applies to all companies simultaneously.
I agree with you but I think at a certain point with large teams, the marginal benefit of hiring n+1 becomes negligible and then dips negative.
I've never worked for a company with thousands of devs but I imagine at a certain point the cost of simply collaborating and managing those devs becomes pretty high too.
Yeah there are certainly diminishing returns. It's just that if they had N employees today, it's very unlikely they were at the top of their s-curve, such that N+1 employees is now net negative. Unless the business was already very mature. Most companies are expecting to grow revenue.
So it would be the exceptional case that firing employees because they were suddenly more productive makes sense.
What's more likely is they were already overhired and the new found "efficiency" is the excuse to resize.
Edit: btw this comes up IRL for me all the time in my work. Companies default play is to cut expenses to hit desired margin. It's much more beneficial in most cases to keep the team but increase their efficiency and handle more revenue. There are diminishing returns to profit by cutting back.
It isn’t that simple because your profit upside is usually finite. Say you run a lemonade stand in a small town. Number of employees can’t really exceed some threshold set by the amount of lemonade customers a given day in that town. Even if the employee is the profit producer. It isn’t an infinite money machine and money has to be made available to seek.
Yeah that's correct, I mentioned "Of course any individual company may have other constraints that makes hiring additional people unviable". One such constraint would be the total market size.
The rest of us non-AI-whizzes combined literally wrote 100% of the functioning open source code that the AI-whizzes steal and transform to an inferior product using Rube Goldberg agent setups.
> The rest of us non-AI-whizzes combined literally wrote 100% of the functioning open source code that the AI-whizzes steal
Open source is hard to steal. It's intended to be taken for free with some license limitations. Second, most giants are already dead. Their work was used for free and that was/is considered normal. Third, many 'AI-whizzes' contributed to open source for long time.
In any case AI will be used on open source. For example to find bugs and backdoors at scale.
Even if AI-whizzes have contributed to open source for a long time, which I question, it does not give them the right to take everyone else's code. Neither does using "AI" on open source.
Moreover, many "AI" open source code bases are a mess and probably wouldn't even contribute in a positive manner to training.
Us? Exactly how much functioning open source code did you contribute? Something tells me is less than the majority of the AI whizzes
that's completely irrelevant to the point they're making.
[dead]
From the article:
the top of the pay scale were elite ai labs such as Openai, as well as hedge funds such as Jane Street that are also betting heavily on machine learning. In this tier, median pay exceeded $400,000 a year. Below that were tech giants including Alphabet, Microsoft and, until recently, Meta, where median pay was closer to $300,000. Experienced developers at most other companies earn much less. Their median was around $180,000 (see chart 2).
—-
A median of 180K is mostly definitely raking it in compared to the median of all US employees.
I’m well over double the household median income in my metro area and while I don’t feel like I’m raking it in, I guess I am when compared to others.
This just seems like a silly article.
It's also worth noting the take-home pay is closer than that, due to progressive taxation. E.g. in California[0]:
$180k/year gives you $118,970 after tax
$300k/year gives you $186,880 after tax
I.e. the gross jump from $180k -> $300k is 67%, but the net jump is 57%. You have to increase the gross more to attract people, because take-home pay is what matters to them when switching job.
[0] https://www.mypaycalculator.net/us-paycheck-calculator/calif...
Microsoft really getting lucky with being grouped in with FAANG in this article, but everything I've read says they pay far lower than most top tier tech companies, no?
It's certainly true by comparing "senior" at G/Amazon/Apple to Microsoft, but is there level skew that compensates for this?
Combine the level skew with stock appreciation and high employee count, and there are a lot of Microsoft engineers raking in that kind of money. So lots of people think of it as competitive even if there are also many working there for less.
I’ve been making over $500k a year since 2020 working fully remote; everyone I know with 7+ YOE at big tech or unicorns is also making 500k TC which often appreciates to ~700k+ with the current market.
We’re just writing dumb gRPC services that use Postgres. I work probably 30 hours a week and still get awards, bonuses etc.
Do tell, I’m very good at my job and can’t get an interview. :-D
Senior+ eng roles at Netflix, Stripe, Discord, Figma, Reddit
Now all I need is an interview and offer.
Offtopic - what kind of unicorns now hire fully remote people with that experience?
Senior+ eng roles at Netflix, Stripe, Discord, Figma, Reddit
Those are all definitely not asking about writing little services in postgres on their job apps. That tier job posting today looks more like a post for a tenure track ML faculty position.
I’ve worked at half of these places and they’re all doing yaml massaging and grpc sculpting. Not much more than basic KTLO with some regularly scheduled feature work and reliability heroics.
Well that is not what that tier job description is describing today at least. Maybe in the past or even still today if you got in early. Hard to imagine so much weight being put on ML knowledge and skills in the job app if it isn’t getting used in the role at all.
https://archive.ph/oV3L3