Sam Altman's argument about RAM utilization is reasonable on a first pass, but misses several key points:
1. people aren't using (GPU) RAM [1] for just running LLMs (in the local model sense), they are using it for other things... games, system displays, other models (Text-to-Speech, Speech Recognition, etc.);
2. being able to run models locally allows them to be used when the internet is down, use them on private data (e.g. medical records, proprietary company data, etc.), including training on that data;
3. the big AI companies are already storing unused GPUs on shelves waiting for their roll-out in data centres, as they can't build the capacity fast enough.
With point (3) that goes against Sam Altman's argument about utilization. The big companies have the hardware, but aren't utilizing it themselves.
[1] And if the argument extends to RAM in general you have all sorts of applications, especially on developer hardware -- editing and compiling software; running local servers; running CAD and other applications; etc.
Better analogy might be that instead of everyone having gas powered generators they run for 14 hours a day, we should put them all in one place and run extension cords to peoples homes.
From a pure efficiency point of view it makes sense to have all our resources pooled and used by whoever needs them at the time. So long as there is always enough to go around, and all the benefits of local computing are kept (though how that works practically if your internet goes offline I don't know) then theoretically, home computers are suboptimal.
But that argument isn't really based on reality. Privacy, offline access, access when you need it, cost... the list goes on. There's many reasons he's wrong, but the core point that it is technically more efficient in the perfect world seems sound.
Altman leaves out of this the fact that OpenAI has basically tried to corner the market on RAM, trying to lock up 40% of the world’s supply without even guaranteeing that it will follow through on the actual purchase orders. I mean, I think that alone would have at least some impact on the current price for RAM, wouldn’t it?
Hes going to lobby the federal government to confiscate all your RAM, like FDR did with the gold. :|
Even putting aside the whole privacy and tuning debates - unless Sam is willing to sell us inference at something near the pre-constrained cost of directly owning the DRAM all he is advocating for is "don't be greedy guys, that's what I'm for. I can be greedy for you!". It's not actually efficient for people to pay huge markups on compute and memory vs letting that same memory site idle some of the time. I mean I guess it's efficient for OpenAIs bank account, but not for anyone else.
Its is inefficient for Sam Altman to have Sam Altman. ChatGPT 31.41592 can do everything he does, but better. He will be the first to understand that and make room.
I have been finding this quiye weird, it seems that the real villany was in what was in Sam Altman's mind when he said something that is basically true.
I have seen many explanations of what that was exactly, they are not all agreed on the bad thoughts were, but they are pretty much in agreement that it was a bad thing.
It's like saying 'public transport is more efficient than driving your own car', and the person hearing you say 'you shouldn't own a car, you should use the public transport system I will own.'
Are LLMs the best use
Of ram? Many other extremely beneficial devices for extremely high value use cases are suffering because of this madman trying to corner the market. Show me the value of everything else suffering.
I think the argument being made is fundamentally wrong. It reminds me of the Socialism vs Capitalism models - this idea that centralizing the resources will somehow unlock efficiency. It's also becoming clearly that merely throwing more data, RAM, compute, etc, is not generating significantly better behaviours.
We are far better off with millions of people running lots of little experiments, than a handful of super large companies running a subset of experiments. We really need to get agentic LLMs onto local machines at a reasonable price to really accelerate.
The likes of qwen3.5-0.8b have been super interesting to me, you have this tiny little model that can run on a CPU that can exhibit agentic behaviours (with a lot of tickling) - in the hands of hobbyists is where I suspect we will see some cool updates.
A problem of both the Altman post and posts that react on him is that they take efficiency/utility in absolute sense instead of in marginal sense.
There is clearly high marginal utility of the first n GB of RAM in your computer, but then its go down probably faster than for RAM in OpenAI's datacenter. And the market just solves for the equilibrium.
> if you have a sufficiently powerful AI tool that is told to “maximize” the creation of paperclips, it might decide the best way to do that is to seize control of all resources
Efficiency and resilience are always in trade-off. Even if there was no other argument it would be worth having local inference as a guard against systemic SPOFs in the big providers.
I don't understand the backlash against this interview.
When listening to the interview it sounded clear that it was just in context of local inference.
To me it came across as stating that setting up local inference, when dram is limited, is indeed limiting from a global perspective.
I think it's a bit disingenuous to state that his opinion is that we shouldn't have any ram being used for any local purposes.
I agree. I mean, his conflict of interest is obvious, but what he was addressing there has been exaggerated. He made no egregious statement. From a utilitarian perspective the distribution of RAM for local inference purposes is indeed inefficient. I still don't want OpenAI or any other company to be able to hoard huge orders of RAM. The knee-jerk reaction from a community that's supposed to be bright was disappointing though.
Of course he doesn't want anyone 'hoarding' their own RAM, for the same reason institutional landlords usually dislike the idea of the housing supply becoming less constrained and allowing people to escape the rent trap. Altman wants to force us into a landlord-tenant relationship, making alternatives more difficult is in his material interest.
Local inference is like buying a house even if it's more expensive than renting, because you (often quite reasonably) don't want to live under the heel of the landlord-tenant power relationship.
"People shouldn't be allowed to bicycle for fun. Only utilitarian bicycles should be allowed. For there's too much energy spent manufacturing mountain bikes, tires, converting ski station into bike parks for the summer. That energy and money should instead be directed towards creating more RAM factories and more datacenters for the AI companies".
Where does such an "efficient" argument ends?
P.S: as a bonus question: does anyone know how to differentiate a paid shill from a sycophant?
Screw Sam Altman. He's the root cause of this shortage, besides that, how does he justify his car having extra seats that he's not using. Sick and tired of these techbros trying to re-centralize computing all over again whilst stealing left, right and center.
Sam Altman's argument about RAM utilization is reasonable on a first pass, but misses several key points:
1. people aren't using (GPU) RAM [1] for just running LLMs (in the local model sense), they are using it for other things... games, system displays, other models (Text-to-Speech, Speech Recognition, etc.);
2. being able to run models locally allows them to be used when the internet is down, use them on private data (e.g. medical records, proprietary company data, etc.), including training on that data;
3. the big AI companies are already storing unused GPUs on shelves waiting for their roll-out in data centres, as they can't build the capacity fast enough.
With point (3) that goes against Sam Altman's argument about utilization. The big companies have the hardware, but aren't utilizing it themselves.
[1] And if the argument extends to RAM in general you have all sorts of applications, especially on developer hardware -- editing and compiling software; running local servers; running CAD and other applications; etc.
It's really inefficient to let people have access to electricity too.
Running all those lines, people don't even use it all that much.
Much more efficient to put everybody into camps, and simply have the people take turns using electrical appliances operating at 100% duty cycle.
Freight delivery speeds need to be brought down too, 35 MPH maximum.
Food? Give me break, get everyone on a Soylent Green diet.
Better analogy might be that instead of everyone having gas powered generators they run for 14 hours a day, we should put them all in one place and run extension cords to peoples homes.
Is #3 true? This seems very inefficient. Aren't futures markets supposed to fix this type of thing?
I am unsure, I heard this for the first time today, and now I have heard it multiple times, with varying changes to what is actually being stockpiled.
Sources for this would be appreciated.
How would a futures market fix that...?
I sell the GPUs I have now for $2B for the right to buy the same number of GPUs in 10 months for $1.6B when my datacenter is complete
GPUs aren’t liquid like stocks or commodities. There’s no guarantee you’ll be able to get them in 10 months.
Where are these shelves? Asking for a friend.
> Sam Altman's argument about RAM utilization is reasonable on a first pass
How so? It's obviously self-serving.
You can be self-serving and reasonable.
From a pure efficiency point of view it makes sense to have all our resources pooled and used by whoever needs them at the time. So long as there is always enough to go around, and all the benefits of local computing are kept (though how that works practically if your internet goes offline I don't know) then theoretically, home computers are suboptimal.
But that argument isn't really based on reality. Privacy, offline access, access when you need it, cost... the list goes on. There's many reasons he's wrong, but the core point that it is technically more efficient in the perfect world seems sound.
In short sound logical argument absolutely zero sense of optics or relatability
Altman leaves out of this the fact that OpenAI has basically tried to corner the market on RAM, trying to lock up 40% of the world’s supply without even guaranteeing that it will follow through on the actual purchase orders. I mean, I think that alone would have at least some impact on the current price for RAM, wouldn’t it?
Hes going to lobby the federal government to confiscate all your RAM, like FDR did with the gold. :|
Cold.
Dead.
Hands.
Turns out “right to bear arms” was just a typo for “RAMs”
Right to bear ARMs.
Even putting aside the whole privacy and tuning debates - unless Sam is willing to sell us inference at something near the pre-constrained cost of directly owning the DRAM all he is advocating for is "don't be greedy guys, that's what I'm for. I can be greedy for you!". It's not actually efficient for people to pay huge markups on compute and memory vs letting that same memory site idle some of the time. I mean I guess it's efficient for OpenAIs bank account, but not for anyone else.
He's saving the planet, don't you get it? His needs obviously trump ours. /s
I heard someone say it’s inefficient for Sam to have two kidneys
Some say a heart too.
It's inefficient for the world to have a Sam Altman
Its is inefficient for Sam Altman to have Sam Altman. ChatGPT 31.41592 can do everything he does, but better. He will be the first to understand that and make room.
I have been finding this quiye weird, it seems that the real villany was in what was in Sam Altman's mind when he said something that is basically true.
I have seen many explanations of what that was exactly, they are not all agreed on the bad thoughts were, but they are pretty much in agreement that it was a bad thing.
It's like saying 'public transport is more efficient than driving your own car', and the person hearing you say 'you shouldn't own a car, you should use the public transport system I will own.'
It's different when the person that says it bought all the cars up and is is actively stripping them down to build a public transport system.
They have no credibility as an impartial analyst.
Are LLMs the best use Of ram? Many other extremely beneficial devices for extremely high value use cases are suffering because of this madman trying to corner the market. Show me the value of everything else suffering.
What kind of specs would you need on a local machine to be able to run fairly decent (Opus 5.5-level) models on it?
How many houses owns Sam? How many rooms in those. Maybe he should let homeless people live in them when he's not there.
Or maybe he should live in a one bedroom apartment.
That would be way more efficient.
I think the argument being made is fundamentally wrong. It reminds me of the Socialism vs Capitalism models - this idea that centralizing the resources will somehow unlock efficiency. It's also becoming clearly that merely throwing more data, RAM, compute, etc, is not generating significantly better behaviours.
We are far better off with millions of people running lots of little experiments, than a handful of super large companies running a subset of experiments. We really need to get agentic LLMs onto local machines at a reasonable price to really accelerate.
The likes of qwen3.5-0.8b have been super interesting to me, you have this tiny little model that can run on a CPU that can exhibit agentic behaviours (with a lot of tickling) - in the hands of hobbyists is where I suspect we will see some cool updates.
A problem of both the Altman post and posts that react on him is that they take efficiency/utility in absolute sense instead of in marginal sense.
There is clearly high marginal utility of the first n GB of RAM in your computer, but then its go down probably faster than for RAM in OpenAI's datacenter. And the market just solves for the equilibrium.
And if we cared even a little bit about efficiency the world would look very different, so we can just stop listening to him anyways
> if you have a sufficiently powerful AI tool that is told to “maximize” the creation of paperclips, it might decide the best way to do that is to seize control of all resources
Hey don't talk about Sam like that!
Efficiency and resilience are always in trade-off. Even if there was no other argument it would be worth having local inference as a guard against systemic SPOFs in the big providers.
We don't have to hold up paperclips as theoretical anymore. AI needs RAM. All of it. The real life paperclip maximization is RAM chips.
I don't understand the backlash against this interview. When listening to the interview it sounded clear that it was just in context of local inference. To me it came across as stating that setting up local inference, when dram is limited, is indeed limiting from a global perspective.
I think it's a bit disingenuous to state that his opinion is that we shouldn't have any ram being used for any local purposes.
I agree. I mean, his conflict of interest is obvious, but what he was addressing there has been exaggerated. He made no egregious statement. From a utilitarian perspective the distribution of RAM for local inference purposes is indeed inefficient. I still don't want OpenAI or any other company to be able to hoard huge orders of RAM. The knee-jerk reaction from a community that's supposed to be bright was disappointing though.
Of course he doesn't want anyone 'hoarding' their own RAM, for the same reason institutional landlords usually dislike the idea of the housing supply becoming less constrained and allowing people to escape the rent trap. Altman wants to force us into a landlord-tenant relationship, making alternatives more difficult is in his material interest.
Local inference is like buying a house even if it's more expensive than renting, because you (often quite reasonably) don't want to live under the heel of the landlord-tenant power relationship.
His case his "efficient"? What about this one:
"People shouldn't be allowed to bicycle for fun. Only utilitarian bicycles should be allowed. For there's too much energy spent manufacturing mountain bikes, tires, converting ski station into bike parks for the summer. That energy and money should instead be directed towards creating more RAM factories and more datacenters for the AI companies".
Where does such an "efficient" argument ends?
P.S: as a bonus question: does anyone know how to differentiate a paid shill from a sycophant?
"It's inefficient for people to have leisure time. They would be more productive if they were chained to their desks/tractors/whatever."
I think that's about where the argument ends.
Sam just wants to make paperclips.
You should expand your paragraph into a 30,000 word blog post. Only then is the message efficient.
A surpisingly good analogy
Screw Sam Altman. He's the root cause of this shortage, besides that, how does he justify his car having extra seats that he's not using. Sick and tired of these techbros trying to re-centralize computing all over again whilst stealing left, right and center.
Better: how does he justify having a car at all, considering it's probably stationary 95% of the time?