Nvidia CEO expects AI to create more jobs for construction workers, electricians, plumbers, and many others

News
By published

But human-AI collaboration will likely lift certain industries to new levels.

Nvidia GTC Spring 2023 keynote with Jensen Huang
(Image credit: Nvidia)

Artificial intelligence is expected to reshape the labor market by eliminating white-collar positions that do not require a lot of creativity or experience, but at the same time increase demand for blue-collar employees like construction workers, electricians, or plumbers, as AI infrastructure requires a massive built-out, according to Jensen Huang, chief executive of Nvidia. That build-out spans many industries, so while some roles will vanish, others will flourish, reports Bloomberg.

"This is the largest infrastructure buildout in human history that is going to create a lot of jobs," Huang said in a conversation with Laurence D. Fink, the head of BlackRock, at the World Economic Forum 2026. "We are going to have plumbers, electricians, construction and steel workers, network technicians, and people who install and fit out the equipment. In the United States we are seeing quite a significant boom in [these areas]: salaries have gone up, nearly doubled. We are talking about six-figure salaries for people who are building uh chip factories or computer factories or AI factories. We have a great shortage in that."

"Ten years ago, one of the first professions that everybody thought was going to get wiped out was radiology," Huang noted. "The reason for that was that the first thing AI [gpy] superhuman in capability was computer vision, and one of the largest applications of computer vision is studying scans by radiologists. Well, 10 years later, it is true that AI has now completely permeated and diffused into every aspect of radiology. It is true that radiologists use AI to study scans. […] However, […] the number of radiologists have gone up. Is that because a lack of trust of or is that because the human interaction with the results of AI is a better outcome? […] The fact that they are able to study scans now infinitely fast allows them to spend more time with patients diagnosing their disease, interacting with the patients, interacting with other clinicians. […] As a result of that, the number of patients that the hospital can see has gone up, [driving revenues of hospitals]."

Google Preferred Source

Follow 3DTested on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

Anton Shilov
Anton Shilov
Contributing Writer
19 Comments Comment from the forums
  • bit_user
    At the same time, Nvidia is pumping large amounts of resources into robotics. Construction workers are definitely at risk of losing their job to robots. It's already starting to happen.

    Maybe someday, even electricians and plumbers!
    Reply
  • -Fran-
    bit_user said:
    At the same time, Nvidia is pumping large amounts of resources into robotics. Construction workers are definitely at risk of losing their job to robots. It's already starting to happen.

    Maybe someday, even electricians and plumbers!
    Depends on how they go with it in the long term.

    I can see what he's hoping to happen come to fruition if robotics are used to enhance workers and not replace. Warehouses is replacing, yes, but construction work, I'm sure humans are too flexible still vs a more controlled environment like a warehouse, so enhancing workers makes more sense to me than flat out replacing. Enhancing heavy machinery and such is the right play, at least for now, given computing at the current level.

    Other more specialised proffessions, it'll depend, since I doubt robots can also fully replace a human in the next 20 years. Not when current models still guarantee erroneous results at any given time. Some lines of work must have "near zero" error rate.

    Automation is going to grow for sure though. Super-specialised factories are proof of that, and they don't rely on "AI". At least, not yet on a massive scale.

    So, work where the human still needs to be highly adaptable, then robots won't be replacing them any time soon, but I think that has been the conclusion from a long time now, right?

    EDIT: I forgot to add that it'll also come down to price. When you lower a barrier of entry to any job or activity, it's reasonable to expect an uptick in employment. That may also imply lower cost per human (payment) working there to offset some of the "machinery" costs? Hard to predict that too accurately, but overall: lower the complete cost of construction -> more construction -> more positions to fill. At least, under that angle, it can come to pass, but also another requirement. Will it come to pass? Not sure... Most of the "robotics" will be tied to subscriptions, so they're bound to be exploitative, so I'm not too optimistic on that front.

    Regards.
    Reply
  • voyteck
    Robots won't replace humans unless they are cheaper in a long term. And they will need maintaining, and won't last for decades, probably.
    Reply
  • bit_user
    voyteck said:
    Robots won't replace humans unless they are cheaper in a long term.
    Productivity (i.e. The ratio of work output per cost) is what needs to be higher.

    Voyteck said:
    And they will need maintaining, and won't last for decades, probably.
    Cost needs to include ongoing maintenance, as well as the purchase price. For smaller firms, I expect the robots will tend to be leased, as I think is common with lots of heavy construction equipment.

    One advantage of robots is they can work more hours/day. You also don't need to provide them with health insurance or pension plans.
    Reply
  • bit_user
    -Fran- said:
    I can see what he's hoping to happen come to fruition if robotics are used to enhance workers and not replace.
    Increased automation has traditionally replaced many agricultural and factory workers. It seems unlikely that we would use the productivity enhancements provided by robots to simply increase the rate of construction, which is what would have to happen for it not to result in net job losses.

    Also, the types of workers you'd keep on, to oversee and maintain a highly-automated construction site, would be rather different than today's typical construction worker.

    -Fran- said:
    I doubt robots can also fully replace a human in the next 20 years.
    I don't see it being a matter of complete replacement, just like factory jobs didn't all go away when automation ramped up. I see it being a gradual erosion.

    -Fran- said:
    So, work where the human still needs to be highly adaptable, then robots won't be replacing them any time soon, but I think that has been the conclusion from a long time now, right?
    Eh, let's think about LLMs and image generators. You give them a task and they do it. Maybe not elegantly, but I think construction work doesn't need to be elegant. If the task is well-defined, then AI can generation motion plans that accomplish the goal, without needing extensive pre-programming in order to work out exactly how to handle each kind of task. I think that's the power that AI can bring to robotics.

    Also, with robotics, the robot can run a simulation of its motion plan, to check for unintended collisions and to estimate its success rate, before committing to it.

    -Fran- said:
    EDIT: I forgot to add that it'll also come down to price.
    The place to watch is in more specialized tasks that are either difficult or hazardous for humans to do. The next area is probably where robots could offer the greatest productivity gains over human workers.

    BTW, perhaps you've heard of some buildings now being 3D printed? I think this probably counts as an example of the latter.
    Reply
  • -Fran-
    bit_user said:
    Increased automation has traditionally replaced many agricultural and factory workers. It seems unlikely that we would use the productivity enhancements provided by robots to simply increase the rate of construction, which is what would have to happen for it not to result in net job losses.

    Also, the types of workers you'd keep on, to oversee and maintain a highly-automated construction site, would be rather different than today's typical construction worker.
    Not all forms of agriculture. For example: vineyards. You need to be able to traverse rough terrain, be precise and be quick. Plus all the additional work outside of just harvesting. Other crop types, sure. But Agriculture is a massive field in its own right, don't forget.

    And that is my point: the devil is in the details. Square houses can be 100% automated today. Complex architectural works or even repair/expansion work? Not so much, I'd say. Wiring a house from the planning phase vs from an already built one? Etc..

    Bit_user said:

    I don't see it being a matter of complete replacement, just like factory jobs didn't all go away when automation ramped up. I see it being a gradual erosion.
    Yep. Not saying otherwise. Jensen (and me, to a lesser degree) are talking about the potential positive outcome of it and I'm putting as many asterisks as I can think of.

    Bit_user said:
    Eh, let's think about LLMs and image generators. You give them a task and they do it. Maybe not elegantly, but I think construction work doesn't need to be elegant. If the task is well-defined, then AI can generation motion plans that accomplish the goal, without needing extensive pre-programming in order to work out exactly how to handle each kind of task. I think that's the power that AI can bring to robotics.

    Also, with robotics, the robot can run a simulation of its motion plan, to check for unintended collisions and to estimate its success rate, before committing to it.
    I don't think the example you put is quite comprehensive. Building is not the same as designing. There's a lot of steps to a building which falls outside of what an LLM is currently capable of doing.

    Different countries and regions within, have different building regulation and laws to abide. For example: concrete mix tolerances vary somewhat from location to location due to geography, climate and such. In parts of the world you can build just using brick, in others you need metal cages and dampening of movement (Earthquake areas) to certain margins. LLMs can't calculate accurately. Sure they can assist with saying "this location needs these mixes" and you still someone to double check, right? That's what I mean.

    This is not even talking logistics. Mega projects are interesting from that angle. I doubt any "army of robots" by themselves could build such a project with no assistance right now. Not even in the next 20 years. YMMV for sure. Like I said: square box houses can be done right now with zero human assistance under specific conditions.

    Bit_user said:
    The place to watch is in more specialized tasks that are either difficult or hazardous for humans to do. The next area is probably where robots could offer the greatest productivity gains over human workers.

    BTW, perhaps you've heard of some buildings now being 3D printed? I think this probably counts as an example of the latter.
    For sure. That's precisely my angle: assist and/or complement. Unless the machinery has a "precise" component to it (like gas sniffers or metal checkers) that a robot can use to "make a decision", they'll never be able to replace or complement a human.

    I've addressed the "3D printing" angle above, indirectly. The easier to build it is, the more likely humans won't be needed. It's the same with current car assistances: they work under very specific conditions or not at all. We're getting there, but not quick enough to worry me in the big majority of potential scenarios they need to cover and I'm not even an expert on the matter.

    Regards.
    Reply
  • bit_user
    -Fran- said:
    Not all forms of agriculture. For example: vineyards. You need to be able to traverse rough terrain, be precise and be quick. Plus all the additional work outside of just harvesting. Other crop types, sure. But Agriculture is a massive field in its own right, don't forget.
    I'm talking about the overall workforce size. If you look at the proportion of the population that worked in agriculture, prior to mechanization, or in factory work, prior to modern automation, it was much larger than it is today. This is not debatable. It's that total workforce size that's at issue, here.

    -Fran- said:
    And that is my point: the devil is in the details. Square houses can be 100% automated today. Complex architectural works or even repair/expansion work? Not so much, I'd say. Wiring a house from the planning phase vs from an already built one? Etc..
    I think it was already clear that I didn't expect 100% of construction jobs to be automated, in the foreseeable future.

    -Fran- said:
    I don't think the example you put is quite comprehensive. Building is not the same as designing. There's a lot of steps to a building which falls outside of what an LLM is currently capable of doing.
    Whoever said anything about architecting? That's another topic, in its own right.

    I was strictly talking about motion planning, which is a very low-level task fundamental to robotics.
    Https://en.wikipedia.org/wiki/Motion_planning
    -Fran- said:
    Unless the machinery has a "precise" component to it (like gas sniffers or metal checkers) that a robot can use to "make a decision", they'll never be able to replace or complement a human.
    This is not as hard a problem as it seems. The blueprint says what needs to go where. The robot's AI can work out how to accomplish the task, and then the robot can judge whether it was successfully accomplished. Machine vision has long been used for inspection in many contexts, ranging from factory automation to infrastructure maintenance, agriculture, and even construction. Verifying the task was accomplished within tolerances is an easier problem than working out how to accomplish it.

    The goal isn't perfection, as humans aren't infallible, either. Of course you'd have an appropriate degree of human oversight, in order to establish that robots are working as intended.

    The key thing about AI is that it frees a human from having to explicitly program a robot to do every kind of task in every circumstance. AI requires only that it be trained how to do a variety of tasks in a variety of circumstances. Then, it can work out the particulars of whatever is called for, in a specific situation.

    -Fran- said:
    I've addressed the "3D printing" angle above, indirectly. The easier to build it is, the more likely humans won't be needed. It's the same with current car assistances: they work under very specific conditions or not at all. We're getting there, but not quick enough to worry me in the big majority of potential scenarios they need to cover and I'm not even an expert on the matter.
    3D printing is capable of handling a lot of the "bread and butter" of the construction industry. It's already happening at scale, and the designs can be a lot less "vanilla" than you might think.
    Https://www.google.com/search?q=3d+printed+buildings
    Reply
  • ravewulf
    It should be noted that those job increases (construction workers, electricians, plumbers) are likely temporary and highly localized to the construction of AI data centers.

    The severe reduction of entry-level white-collar jobs isn't just bad in the short term for people in or seeking those positions, it'll be a long-term disaster when trying to fill mid and higher-level positions because the talent pool has dried up.
    Reply
  • thestryker
    For me the most laughable thing here is him calling this infrastructure. Sure it's compute infrastructure, but this is not what most people would refer to as just infrastructure. This is enabling companies to keep the current wealth train running, and not improving anything else. Eventually it might be of some more broad benefit, but there's no sign that this is coming any time soon.
    Reply
  • SSGBryan
    ravewulf said:
    It should be noted that those job increases (construction workers, electricians, plumbers) are likely temporary and highly localized to the construction of AI data centers.

    The severe reduction of entry-level white-collar jobs isn't just bad in the short term for people in or seeking those positions, it'll be a long-term disaster when trying to fill mid and higher-level positions because the talent pool has dried up.
    Yep, that is all he is talking about, not permanent jobs.
    Reply