Hacker News Reader

Comment by elpocko | Hacker News Reader

There are many more millions of GPUs in computers and game consoles around the world burning electricity for your entertainment, for decades. The same class of devices. The environmental impact of having pretty pixels on your screens is at least an order of magnitude higher than what it is for AI, for inference and training. I don't see anyone being up in arms about that.

View on HN

Your calculations must be severely off, because I never heard anyone advocating for the construction of nuclear reactors to power game consoles around the world. However, we hear everyday that we need to build these reactors right now if we want to have AI.

Xboxes, smartphones, and personal computers are geographically distributed and so is their power generation. Data centers are …centralized. Dedicated power plants for large data center installations are not new.

>Xboxes, smartphones, and personal computers are geographically distributed and so is their power generation.

This is irrelevant because most "Xboxes, smartphones, and personal computers" are powered by centralized fossil fuel power plants that could plausibly be replaced with nuclear reactors, just like the power plant for a datacenter can be replaced with nuclear reactors.

My Xbox is powered by solar. I can't say the same for my use of Claude, and I do not have the same agency to change that.

The point is that the centralization of data centers makes them suitable for dedicated power generation.

I hear people advocating for the construction of nuclear reactors every day. They don't mention gaming, just like they don't mention refrigerators or washing machines specifically. A gaming machine consumes the same amount of energy as a machine used for AI, it's the same hardware. AI consumes it for seconds per user, while one gaming machine is used for hours per session. The energy required by one human to play a game for one hour could serve hundreds or thousands of AI users.

The GPU's used for AI have significantly higher utilization rates than gaming GPU's...

Here's some napkin math:

H100: 61% utilization / 700W ~ 3.7MW/year

RTX 3080: 10% utilization / 320W ~ 0.27MW/year

How many AI users are served using a single H100 per time, and how many gamers are served using a single 3080 per time? How many gamers are simultaneously running a 3080 or equivalent for their entertainment?

Yes, AI users are being served by much more than one GPU at a time.

Which is why the power used is so much higher than a single gaming pc

Thousands of users are being served by those GPUs. A gaming PC has one user.

I can't believe I have to point this out on HN of all places.

You can't compare those things very easily.

If you're only talking about the GPU's used for inference, then that's a different story. Not nearly as much hardware is required for inference.

But the number of GPU's needed to train models is in the tens of thousands, and there are rumors that some shops (Meta) are already using 100k+ GPU's, just for training.

Those are likely all/mostly H100s, running at least 60% of the time. Consider that OpenAI, Anthropic, Google, Meta, Tesla, X.com, etc. are all within an order of magnitude of each other in terms of compute.

For arguments sake, that's 6 companies approaching 100K H100's worth of compute for their next gen models.

Now consider that GPT4 used roughly 100x more compute to train, compared to GPT3. And GPT5 is rumored to follow this trend, using 100x more compute than GPT4. Extrapolating, GPT6 might also use 100x more compute than GPT5.

Even if the next generation of AI GPU's are 10x as powerful as the H100 for the same amount of electricity, the next generation of models would need 10x as many GPU's (and thus, 10x as much electric demand).

Extrapolate that to GPT7, 8, 9 etc. And you can see why people are worried about the power usage.

This isn't even theoretical. As mentioned in this thread already, these companies are signing deals to buy all the capacity of power plants in some areas.

>100k+ GPU's, just for training.

That's a tiny drop in the sea of the almost 2 billion PC gamers [0], and hundreds of millions of gaming consoles [1] in the world. Not to mention the energy required to manufacture all that hardware. Plus the datacenters required for online gaming, which must also be considerable.

It's weird to be concerned about power usage of AI, but turning a blind eye to the massive amounts of power required by the gaming industry.

[0] https://www.statista.com/statistics/420621/number-of-pc-game...

[1] https://en.wikipedia.org/wiki/List_of_best-selling_game_cons...

I think it's weird to ignore that every GPU used for AI equates to 10-50 gaming GPUs (10 if you assume every gamer has a 3080, which they don't), and that the number of GPUs used for AI is 10x every 18-24 months.

It's not that AI uses too much power today, it's that at the current trend, it'll be using somewhere between 100x and 1000x as much power by 2030/2035. Which would place it between 2-20% of total power consumption.

Where does this "10-50" figure come from? A quick google search says a H100 draws 700 W, while a 3080 draws 320 W.

AI provides tangible value to businesses and private users beyond mere entertainment. We'll see how much power it consumes in the future, and where that power comes from.

I did the math in previous comment that you already replied too..

"The GPU's used for AI have significantly higher utilization rates than gaming GPU's... Here's some napkin math:

H100: 61% utilization / 700W ~ 3.7MW/year

RTX 3080: 10% utilization / 320W ~ 0.27MW/year"

H100 uses at least 10x as much power as a 3080 over it's life time. And most gamers aren't playing on 3080's.

The difference here is between potential energy use and actual energy use. AI chips as well as other server chips are in use most of the time. But if I sell 1 billion gaming chips, these chips will be in use a small fraction of the time.

Indeed. But how many gaming GPUs are out there in the world?

That's his point? Greenpace wants AI datacenters to be built with clean energy. elpocko points out that plenty of other pointless electricity consumers also aren't being built with clean energy today. That's not an argument against green energy, but is pointing out that greenpace isn't very rigorous with their pleas. They're seemingly picking whatever is the most topical. We should be against this, because latching on to the latest thing basically guarantees that the next thing rolls around, all the momentum will be lost. Remember when everyone was up in arms about crypto mining? How it's barely brought up because everyone's focused on AI.

The GPUs in PCs, consoles and phones aren't running full tilt 24/7. They run very bursty workloads for a couple hours a day at most.

Those in AI data centers never stop running and completely utilize their capacity. The difference in power usage is astronomical.

Can you please at least do the back-of-the-envelope math behind the “astronomical”?

I don’t claim to know, but we ought to be able to have a rational debate on this.

You don't need concrete numbers or even napkin math to realize that a gaming computer running a GPU for a couple of hours in the evening is going to use much less energy than a GPU running maxed out 24/7 for AI

There's nothing irrational about suggesting AI GPUs are consuming far more power

Yes, one AI GPU uses more energy than one GPU used for gaming. The one AI GPU however is shared between a large number of users, while the one gaming GPU is used by one player.

Apparently a single gaming GPU can be used to run an LLM that serves hundreds of concurrent requests.

> Benchmarking Llama 3.1 8B (fp16) on our 1x RTX 3090 instance suggests that it can support apps with thousands of users by achieving reasonable tokens per second at 100+ concurrent requests.

https://backprop.co/environments/vllm

But that's a tiny model; it's the smallest version of Llama 3.1. The commercially marketed models are way bigger - e.g. GPT-4 has been estimated to use about 1.76 trillion parameters, 220 times more than the Llama build you mentioned. Their resource and performance requirements are vastly different.

You're essentially arguing that shipping naval diesel aggregates must be trivial because you can fit a dozen moped motors on the bed of your pickup truck just fine.

Okay but these tiny models are being used by people and businesses instead of GPT-4. My point was that they consume less energy per user than a rig used for gaming.

I have no insight into how many GPT-4 users are served per GPU, but I would assume OpenAI heavily optimizes for that, considering the cost to run that thing. It's probably in the same ballpark: hundreds-thousands of concurrent user requests per GPU. Still better than one GPU per gamer, even if it requires 10x the energy.

There are a lot of folks who vastly underestimate the carbon output of current AI training and work, and you're among them. The number of data centers being raised right now with increased power planning around data centers around the globe points to a reality of energy consumption that's probably an order of magnitude higher than you imagine. At a time when the costs of carbon poisoning the oceans is getting really ugly and driving extinctions, melting polar ice, and driving global warming, writing off a major new generator of atmospheric carbon is dangerously irresponsible.

https://www.technologyreview.com/2025/05/20/1116327/ai-energ...

>writing off a major new generator of atmospheric carbon is dangerously irresponsible.

I agree: ignoring the carbon footprint of the gaming industry is irresponsible.

I wouldn't recommend ignoring any of the sources of pollution.

It’s not the same class of device.

A typical NVIDIA server GPU consumes 700W, and a server might have eight of them, so 5.6kW.

A PlayStation 5 consumes 200W total.

A PS5 serves a single person, maybe two. A datacenter GPU might be shared by dozens or hundreds of people, depending on how you count occupancy.

A typical server is serving hundreds or thousands of user sessions, while a PlayStation 5 is serving only one.

However, each consumer is paying for his own gaming console power. With AI (but also other "free" Internet services), there is a tragedy of commons (a coordination) problem, where you don't know the direct cost to you and so you (and others) will not make rational choices.

Show your math.

Besides common sense, I can tell you about my kW h counter going brrr when playing games (400 W continuously, sometimes for hours on end) vs. running Stable Diffusion or Llama-whatever (400 W for 15 seconds every 3 minutes for an hour or two). Extrapolate from that.

Training runs basically 24/7 at full throttle. Inference isn’t the main source of energy consumption.

Show your math.

The most common GPU per the steam survey is the RTX 3060 at 170W TDP. A huge % of users have cards near or below this TDP. The SXM H100 has a 700W TDP, and will spend far more of its life at or near that value.

Given an average ~8 hours of work/school and ~8 hours of sleep, gaming GPUs likely don't use anywhere near as much power. Plus, even when they are on, they will probably idle near 30W-60W for a lot of time spent browsing the web or watching videos.

There are more gaming GPUs in existence right now, but the number of AI chips is likely closing that gap rapidly.

And of course, what is that energy being used for? People playing games are typically having fun, bonding with friends, or engaging in social behavior. A huge amount of AI is illegally trained on copyrighted works without license to use them, causing significant harm to various fields. Plus the deluge of AI slop bogging down the internet, social media, forums, image/art-hosting sites, search, and more.

I think it will be a while before modern generative AI is even close to providing value in aggregate.

Reply on news.ycombinator.com