[ExI] Is China racing to AGI?

[John Clark]

On Sat, Nov 29, 2025 at 4:28 PM <spike at rainier66.com> wrote:

*>So what if… our EVs have the option of a high-power GPU as a cabin
> heater, replacing the nichrome coil? *


 *The trouble is a high-powered GPU of the sort used in AI, like the NVIDIA
B200, costs between 25 and $50,000, depending on what sort of deal you can
manage to swing. And they consume about 1200 Watts so they need to be
liquid cooled or they'll start to glow like the filament in an incandescent
light bulb. You could use a far less powerful consumer grade GPU but it
still needs to operate the car so I don't know how much compute you'd have
remaining to do other things. *

* John K Clark*









>
>
>
> *And after I've read the following article I'm beginning to think he might
> be right, the US certainly has a huge advantage over China in the amount of
> computing power it has available: *
>
>
>
> *Is China Racing to AGI?*
> <https://www.chinatalk.media/p/is-china-agi-pilled?utm_source=substack&utm_medium=email>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> *So what if… our EVs have the option of a high-power GPU as a cabin
> heater, replacing the nichrome coil?  Then if we had some computing task
> requiring enormous resources, such as Bitcoin mining, then every time we
> turn on the car heater in our EVs, that cranks up and starts calculating.*
>
>
>
> *A Bitcoin miner is “worth” about 11 cents a KWH as I understand it.
> Running that on power that would ordinarily go into resistance heating
> means it is free calculation.  A typical prole in a cold environment would
> accumulate about a 6 to 10% chance he will discover a Bitcoin, which is at
> about 91k this morning.  This means the car will give back about 5 to 9k in
> mathematical expectation over its lifetime.  Most EVs would never discover
> one of course, but about one in 10 to perhaps 1 in 16 will find a Bitcoin
> at some point, which would be an excellent sales feature.*
>
>
>
> *Granted it might add that much cost to the product, but this is what I
> don’t know: an EV heater has an adjustment.  My intuition tells me there
> should be a way to rig a variable clock on the processor, so that the speed
> of the processor produces more heat (almost linearly.)  Processor hipsters,
> it is possible and practical to have a variable clock to adjust the amount
> of heat pouring off of a processor?*
>
>
>
> *spike*
>
>
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20251130/67427fba/attachment.htm>