I guess, if you run a big tech company, you think about big tech company things. It would be easy to forget that there is a world outside of your big tech company, full of people who think about different things – like watching Netflix, texting their kids, writing research papers, and contemplating what they did wrong to be stuck making video calls in Microsoft Teams.
I don’t know that Sam Altman was thinking about any of that when he was a guest on the Cheeky Pint podcast with John Collison, the co-founder of Stripe. On a recent episode, Altman said that it was a shame that people were hoarding so much RAM that could be much better put to use in one of OpenAI’s data centers.
Well, specifically, what he said was that “from a global allocation perspective, actually, it’s quite terrible to have everybody have a bunch of the world’s limited supply of DRAM sitting in their own computer and being used 5% of the time.”
I’m not sure where he got the 5% figure, but that’s not really the point. A much more important question is, “Hey Sam, why do you think there’s a RAM shortage in the first place?”
To be clear, I don’t think Altman was suggesting that people should literally give up the RAM in their computers. He’s basically arguing that RAM is scarce and letting us commoners have it in our millions of personal devices is an inefficient way to divvy it up. Instead, we just put it all in OpenAI data centres where it can be shared by thousands of users. You know, users who pay a monthly fee for ChatGPT.
Honestly, he probably has a point. From a purely big tech company perspective, a limited resource like RAM should be put to its most efficient use. But there’s also an obvious problem with Altman’s comment.
Right now, the technology industry is experiencing a serious memory shortage. That’s largely because companies building AI models and infrastructure are buying up basically all of it to power their data centres.
The companies that make memory are happy to shift to making the higher-margin type used in AI servers, which means there’s less capacity for them to make the stuff used in our iPhones and laptops. That pushes prices up on everything because, as it turns out, everything is a computer and every computer needs memory.
OpenAI is one of the companies building out massive compute resources in order to train and operate its increasingly powerful models. In other words, the companies building AI data centres are the primary reason there’s a shortage, and the CEO of one of those companies just told you that the RAM in your computer is part of the problem.
Huh.
I actually think there’s a far more interesting lesson here. Altman’s comments reveal exactly how easy it is for people to convince themselves that whatever is best for their business must also be best for everyone else. For Altman, all the memory sitting in personal computers around the world is a terrible way to waste something he desperately needs.
Of course, that’s only true if you think that getting the maximum efficiency out of a stick of memory is the most important thing. But none of the rest of us think that way. If you want to use a laptop, you need memory. You probably don’t care if Altman’s world-conquering ambition is slowed down because you’re running 47 Chrome tabs on your MacBook Air.
Or, maybe you’re someone who wants to run AI models locally precisely because you don’t want to send everything you do to a company like OpenAI. Maybe you want a Mac mini in a rack with 64GB of memory, or a MacBook Pro with 128GB, so you can do LLM things without paying a few hundred dollars a month for ChatGPT or Claude.
That’s the real source of Altman’s annoyance, and it says a lot about his bias. Of course, he’d rather no one run local AI models on their own computers. Of course he’d rather you pay for ChatGPT. And, of course, he needs all the memory he can get.
I don’t think Altman is necessarily being dishonest. I’m sure he genuinely believes that letting him have all the RAM is the best way to allocate a scarce resource. That’s what makes it so easy to fall into this kind of thinking.
When you spend all your time looking at the world through a big tech company perspective, you forget that normal people care about normal things. They have very different priorities. Mostly, they just want to live their life without being told they are an inconvenience because they aren’t using RAM the way some AI billionaire says they should.
Ultimately, Altman sees unused RAM and thinks about all the ChatGPT requests it could process. The rest of us – well, we don’t see RAM at all. The rest of us just see the computer we already paid for, doing exactly what we bought it to do. And we’d like to keep it, thank you. – Inc./Tribune News Service
