A.I. companies are recruiting electricians and carpenters by the thousands
nytimes.com156 points by thm 6 hours ago
156 points by thm 6 hours ago
I would be very careful about basing your career decsions based on this trend. The data center build out will always be very boom and bust, which would really suck to work in. One year you would be making 300k working on dataceters and the next you would make 30k because thousands of electricians are now competing to build a few single family homes.
Industrial electricians have always been in the boom-bust cycle.
IBEW (Union electricians) have been doing really well this decade. Another option if you don't want to be a contractor or an Anthropic employee. But they've also gone years without enough work to go around.
Scroll down to this comment for a better discussion than I could give: https://news.ycombinator.com/item?id=49098874
Also, an electrician just doesn't pop into existence. You have to go through years of apprenticeship and union dues payments.
I was in the IBEW when I was doing basic tech support for HawaiianTel 15 years ago... making $15/hr, union dues were nothing even at that wage and they rarely ever go up.
This cry of "OMG dues suck" is so stupid for what you get in return. But go ahead and buy more into republican propaganda.
Yeah, working people should never sit across each other at a table and discuss mutually beneficial things. For the Epsteinth time, class unity is only for the upper classes.
Very true, it is difficult to even find an opening for the training to start the process, the schools are limited and have short enrollment windows that are fiercely competitive (at least in California).
I applied for an apprenticeship several years ago when I was testing the waters to rotate out of tech after covid. Tested into the highest percentile, had previous high voltage experience with OEMs, a mech eng degree, still ended up like 200 people deep in the queue. That was the end of that interest. You basically have to be friends or family with someone to get into IBEW.
this is a bizarre timeline of reality where an ME degree gets you waitlisted for an electrician apprenticeship
In the end I'm happy it turned out the way it did. I've had some great experiences since and stayed in tech. Plus that freed up a slot for one of the many early 20s kids that were trying to get in.
Sure, but these also aren't professions that will disappear when data center demand goes down or will get automated away in the long term.
I’d go as far as to say that for someone in their teens electrician (or trades in general) is a more stable long term career path than software engineer.
Both have boom and bust cycles. They are different cycles, but they are still cycles. I wouldn't call either more stable than the other.
You'll always have manufacturing and construction. I've never heard of an electrician boom bust cycle. It's infrastructure. It's always in demand.
I know several electricians and they all have been through cycles if they have been around for more than a decade. Things never drop down to zero though, so sometimes they see their buddies without a job while they are still working.
Yeah, but you might still want to enter the field based on the aging of the current workforce. Seems likely to be a shortage of electricians over the next while.
On the other hand, the economy is going to be increasingly electrified. Demand is going to be there in some form.
On a third hand, electricity is just movement of electrons.
While DC is driving electrons down the line, AC is basically just when you vibrate them around.
Anyway my point is the field (no pun intended) may change drastically in the increasingly electrified future. I'm personally excited for more wireless power in areas, and the concept of ambient energy - harvesting power out of the already vibrating electrons in the static in the air.
I feel like we're entering an age of electrification so having training/education in a field related to electrics/electronics seems like a reasonable path for someone to take.
I think that age started over 100 years ago.
The conversion to all-electric homes is more recent, and hasn't even started in some places. Here in Canada, some jurisdictions require newly constructed houses and apartments to have electric heat and cooling (heat pump, with resistive heat as a backup), an electric dryer and stove, and an electric water heater (tank or tankless). Others jurisdictions are still using gas furnaces and gas-fired hot water tanks.
All electric homes existed as far back as the 1960s. Gas heat was generally cheaper back then (heat pump? nobody has air conditioning then) so they were not common where it got cold. Even at that though, the difference in labor for an all electric house today vs the same house with gas appliances is minimal - you get larger wires to the stove, furnace and dryer, everything else is the same. (in the near future I expect to see wires for the EV as well, but not yet and in any case this is the same even if your other appliances are gas)
I live in one of those all electric homes too, made in the 2000s. Doesn't change that a bust will dry up the work and homeowners won't fork over the money for something that has few benefits for their daily life and requires expensive changes to the house.
To add to this there are jurisdictions that will require you to still pay a small monthly fee to support the gas pipes if you cut off gas service.
It is really nice not having a gas bill and not requiring maintenance for gas appliances, but most people won't care if they lived with it for long enough.
The only case I see it working is in a disaster area or blighted areas where the builder decided to bulldoze and rebuild the houses. The southern California fires for example; the city gave priority to permits to houses that went all electric since the gas pipes were destroyed. Permitting and inspection is so much simpler with all electric.
It's accelerating. Electrification of infrastructure and buildings is a huge growth area with big delays in projects due to lack of capacity.
> The data center build out will always be very boom and bust, which would really suck to work in.
HVAC and electrical (e.g. UPSes) need somewhat regular maintenance, so once build-out is done, if you're with one of the OEMs, you'll probably have regular work babysitting the gear.
Doesn't that average out to 165K / year? That's not so bad IMO!
I'd imagine those 300K years they're not just working 40 hours a week, more like 70+ hours. Add higher taxes on those years and it becomes less attractive.
Oooof. I pity the poor fool who has to pay taxes on their 300k annual salary.
If it was a predictable & reliable 300k, 30k, 300k, 30k, 300k, 30k pattern, then yes.
But if not...
I mean, no career is ever perfectly secure against future uncertainty (a few years ago I would've joked "... except prostitution", but even that seems threatened by AI these days). The question is a relative one: how stable is it compared to feasible alternatives? I could certainly see job losses and wage stagnation in software engineering being worse than those of electricians over the next 15 years.
Next, plumbers. The future of data centers seems to be more liquid cooling and less ductwork. Take a look at this new 1 megawatt server rack.[1] It has more pipes than cables. All that coolant flow and all that electrical power in a crowded space is scary.
> All that coolant flow and all that electrical power in a crowded space is scary.
Maybe it's a non-conductive oil? I recall seeing someone who overclocked their PC (circa 2005) to 5GHz, and ran the motherboard completely submerged in an oil bath.
Listen to the video again, he clarifies early on that this is supporting multiple racks. A megawatt is 1000kW. If you took 5 B300 systems (10 rack units each), you'd still only be at 75kW, which if you're doing redundancy properly, means you have wired 150kW to the rack, but you're not generating a megawatt of heat.
It's nice to hear tradespeople getting paid well with lots of work. I'm happy for them :-)
Yes, that's good. It also prices out a lot of other better uses for their labor. Projects I work on have seen prices soar. I don't consider building the things everyone hates because they promise concentration of wealth and an attack on labor at the expense of building literally anything else to be somehow a big W for labor because a few people get a slightly higher temporary paycheck. Ask the Wagyu cows how they feel about the hand they are dealt.
"It also prices out a lot of other better uses for their labor...."
"Projects I work on..."
Considering we are in a gigantic housing shortage, it would be great if a house didn't have to compete for talent with a data center.
resi has been drained pretty hard in the last 10 years, this just adds more fuel to the fire.
tbf, I don't think Wagyu cows were that thrilled about their lot in life before Wagyu became a thing, either.
I am, exactly right this minuit, trying to explain that I am not able to quote competitivly for certain types of specialised metal work as there are specialists for that, but am facing insistent inquirys anyway.I do clean sheet custom one of a kind projects as a rule, for which, due to AI recomending my company through some imaculate process, I am exploring how much I can charge for helping the singuarity along, though all in all I am glad I have other skill sets to deploy should we run the whole thing into the ground.
> Projects I work on have seen prices soar.
Could you be more specific? Things are still competitive in the large commercial electrical market I am bidding in, if you’re bidding to make more than 10-15% you’ll get underbid by someone willing to accept a lower profit margin.
>a lot of other better uses for their labor
Who on Earth are you to decide what is the "better" use for labor? And do you scrutinize every construction project or only AI?
Is it a waste when contractors build a temporary set for a trashy reality TV show? A factory that makes cosmetics for dogs?
Having someone the arbiter of what is the "right" use of labor is communism. In capitalism, people get to decide for themselves what they work on and what they buy.
Uhm, aren't we all allowed to have an opinion on what's best for mankind?
FWIW I am a high voltage test engineer.
I used to be an electrician.
I would prefer to build the substations that power machines for the benefit of all mankind, rather than data centers that feed the already wealthy.
For example, keep the lights on for schools.
You have a warped view of the world if you believe anything should be allowed to be built just because someone has the capital to build it.
If Meta decided to pivot to a brothel and meth lab building spree would you consider that to be a worse, equal, or better use of their capital and the labour in the regions where they are building out this infrastructure? Why?
There's sort of no winning here!
AI unemploys middle-class workers: AI is stealing jobs from the middle class!
AI employs unionized middle-class workers: AI is stealing cheap labor from the rest of us!
The employment is temporary, what follows a net negative, is the idea.
And "AI" as a concept has has no agency. This is all about humans who seek to further concentrate wealth.
That's pretty disingenuous.
The people losing jobs to AI can't just pivot into the trades. When you think about it, companies announcing they're pivoting into some other area where they have no prior background is usually a bit of a desperation move, aided by the fact that companies don't need to be licensed in any way to offer particular services, other than maybe in the military/national security sector.
As for the second point, it's true that this is just old-fashioned supply and demand. I predict the outcome is that there's a rush into the construction/contracting sector followed by a bust due to either the tech bubble popping and construction slowing way down, or the replacement of construction workers by humanoid robots, or perhaps some better form factor like robot spiders that can squeeze into small places - all the benefits of child labor but with double the appendages and double the productivity!
However, I'm not worried about the possibility of a robot takeover. The American people's final safeguard against a robot uprising is the irrationality of imperial weights and measures. Robots are digitals and can't do fractions like us higher beings. That's why I'm building my data centers measuring everything in cubits.
As an engineer in the US - we already do all the math in metric units. There are conversion functions in the UI layer to transform our units to whatever you want to see. (they also convert metric units if you don't like the units were are using internally)
> AI employs unionized middle-class workers: AI is stealing cheap labor from the rest of us!
What cheap labour? Tradespeople are not cheap (especially if you factor in the quality of the work). This just makes them even more expensive. Meaning a big impact on homes – you know, the things people and families live in, in order to survive. Pretty important
>Meaning a big impact on homes – you know, the things people and families live in, in order to survive.
Maybe those people should pay their tradespeople better then. Everybody's all "workers should be paid more" until they're the one paying the worker.
"Those people" who? Homebuilders? You're saying you think the cost of homebuilding should be much higher? When I listen to some of you guys it's not hard for me to see why these DCs are so unpopular
Everyone should just compete with the trillions of dollars of capital floating around for AI? And it's their fault for suddenly being 'poor'?
Sounds like you're assuming people are paying pennies for labour. I don't think that's true at all
While we're at it, how much should small bathroom, 2 week renovation cost? $100,000 in labour?
That small bathroom job is literally the worst case scenario for a tradesman. They have to work nearly as hard to line that up as better paying longer term commercial work.
The residential buyer is the source of much less repeat work, is harder to deal with, more price conscious and harder to collect from than the commercial buyer they could be working for.
Anyone that can competently market, manage, and execute that small job can do the same for better customers.
The residential buyer has one advantage though: they do not demand as much from those working. You can start this business today, and likely find enough friends and family willing to take a chance on you that you can get started and from there you have references. Commercial projects tend to be big enough that you cannot get into them without experience (which in turn you cannot get).
The crew that can build a driveway cannot build a parking lot, and they will never be able to break into that market because it requires experience that a driveway cannot give. From a parking lot there is then road building - the previous companies cannot break into them. Again there are just not enough projects of the size that sits in between to build experience.
And even in residential, building million-dollar houses is easier (or at least not harder) and more profitable than building "affordable" starter homes.
> While we're at it, how much should small bathroom, 2 week renovation cost? $100,000 in labour?
Obviously much more than 2 weeks worth of work at a commercial job, since that's the worst possible kind of work you could ever offer any tradesman. Maybe 8x the hourly rate would be fair since it's intermittent work with generally horrible customers?
It's been interesting to watch the IT/engineering classes all of a sudden pretend to care about labor costs and outsourcing/job loss. As an industry they have been by far the most impactful in reducing wages, enabling outsourcing, and "gaining efficiency" via automation while as a class being paid top-tier wages for doing so. There is not another industry more directly culpable for blue collar job loss except perhaps finance and banking.
Only now when the shoe is on the other foot do such folks seem to care at all.
Learn to code indeed.
For any electricians/plumbers/etc. out there? Go get yours. When the shoe was on the other foot you were mocked and told to switch careers, lowballed on horrible residential projects, told to skill up and get better if you wanted to get paid more, there was no future in the trades, etc. Make hay while the sun shines and ignore the bleating. At least this time around the bubble is sharing a bit of that tech wealth.
And in time AI robots will reduce the need for these blue collar workers who are presently helping replace the white collar workers.
Just learned of a new acronym today "NILF," not in the labor force with over one million supposed NILFs out there -- dropped out and gave up.
Good luck with robots traing to lay up bricks, or doing plumbing in a home.
Making a brick laying robot is easy. I'm happy to do that if you want to fund my mid-life crisis... Note that I did not promise my brick laying robot can effectively compete with human break layers on cost. I fully expect my robot to be at least double the cost of humans doing the same work, and it will be less flexible than humans as well (meaning you need a different robot for different buildings)
Ummmm in time it's inevitable as in 2026 you do not need any skill to be a computer programmer. AI has replaced that once valuable skill and will continue it's way through ever white collar job til the robots (5 to 10 years) start taking the blue collar jobs away.
The blue collar job boom is great present day, but what they are helping build is their machine driven replacement.
Maybe this all sounds hyperbolic, but it's what Musk and all the AI overlords have said as far back as a 2013 or prior. Musk and Altman have noted the need for everyone to be on welfare (Universal Basic Income) since then. And no doubt we see this taking shape now within white collar jobs that once we held in high regard like programming being replaced by AI. I just created a whole startup web design, front-end, back-end development, native iPhone app, Android app and more using Codex (previously i hired others to help me a UI Design Front End guy).
Vibe coding by people with no programming skills is not going to happen. What will happen is moving to higher and higher code representations. A compiler is already doing 99% of the actual coding when you realize that the only language computers understand is assembly.
That's correct! AI is a negative for society no matter how you look at it. That's the entire idea!
> There's sort of no winning here!
> AI unemploys middle-class workers: AI is stealing jobs from the middle class!
> AI employs unionized middle-class workers: AI is stealing cheap labor from the rest of us!
Come on, you've got to know it's not going to employ those "unionized middle-class workers" for long. Their jobs are sort of like these jobs: https://www.nytimes.com/2026/07/10/business/ai-white-collar-... ("people participat[ing] in feeding their own careers into the A.I. wood chipper...Many people who contract for these companies understand that this is a short-term opportunity...before they themselves are automated out of the job").
And yeah, there's "sort of no winning" when you're gobbling up resources to build stuff people don't like.
I don't think that ANY project employs electricians and carpenters for long, because they are there to build something that will ideally last decades or more...
I'm not even sure it's as simple as "stuff people don't like." Many of the anti-AI crowd use ChatGPT which kind of complicates the question of "do they REALLY dislike it?"
I guess you could complain that robots will replace electricians and carpenters, but it's still going to be a decade before anyone trusts an LLM to wire a house.
> Many of the anti-AI crowd use ChatGPT which kind of complicates the question of "do they REALLY dislike it?"
Not more "revealed choice" BS. There's no free choice: even if someone doesn't want to use ChatGPT, they may not be allowed to avoid it. I'm literally forced to use it at work, for instance. Others may be forced to use it due to market or other pressures.
Personally, I wouldn't mind of someone dropped a couple cluster bombs on each of these AI data centers, after construction is complete, and LLMs just disappeared as a practical technology.
> I don't think that ANY project employs electricians and carpenters for long, because they are there to build something that will ideally last decades or more...
See my reply to the other guy who made the same point before you did.
> Not more "revealed choice" BS. There's no free choice
I'm not even talking about at work, I'm saying that a large majority of Americans are using AI of their own free will in situations where it's not required but is genuinely useful. The sheer number of people means that a lot of them must also say they're anti-AI, and my point is that whether or not they say they like it they're undeniably choosing to use it.
> Personally, I wouldn't mind of someone dropped a couple cluster bombs on each of these AI data centers, after construction is complete, and LLMs just disappeared as a practical technology.
This would be great if your goal was limiting AI access to the rich, but I'm pretty sure that would be much worse than democratizing access.
> See my reply to the other guy who made the same point before you did.
Yeah, and I still don't get it. Do you think that qualified electricians struggle to find work? They generally don't, these skills are in high demand and have been for decades. You really haven't made the case for "after they gain more experience on a difficult and high-paying job, they're gonna be unemployed." Everyone in these comments is complaining about waiting lists and backlogs and you're acting like the opposite is true.
>I'm not even talking about at work, I'm saying that a large majority of Americans are using AI of their own free will in situations where it's not required but is genuinely useful. The sheer number of people means that a lot of them must also say they're anti-AI, and my point is that whether or not they say they like it they're undeniably choosing to use it.
I don't think there's really a contradiction there. You can recognize the utility of something while acknowledging that, on the whole, it'd be better if it didn't exist.
If people are using it presumably they're getting some value out of it which means that most objections to datacenters are NIMBYism and not some kind of principled stance.
Datacenters are fine. They're better than most kinds of industrial buildings, by and large they don't illegally set up fields of gas turbines, and their water usage is really not that bad compared to other large users. They make for a few nice, high-paying jobs. Sure, they're ugly and large, but this is America and we have no shortage of ugly, large buildings.