One thing about these numbers that's absolutely shocking to me is how low the energy use is:
> That model’s usage was well within our budget ($68, about 4kWh of energy use / 365 grams of carbon emissions).
The energy cost is literally 1% of the total cost. For context, 4kWh of energy would drive you about 15 miles in an EV, about half of the average person's daily driving miles. It's boiling 10 gallons of water.
With the talk of AI data centers' impact on the world, you'd think this would be 10x to 100x the amount of energy in order to get the effects they're using here.
My takeaway: the AI data center buildout is an overbuild probably at least as large as the fiber buildout that left us with so much dark fiber. If not even bigger. The only thing that will save the economy is the inability of NVIDIA and chip fabs to produce enough chips to match the buildout planned.
The AI datacenter buildout is in the gigawatts. A single 1.21 GW data center multiplies that single-user 5.5 W average load (4 kWh per month) by as much as ~200 million. Obviously a widely distributed load is much less impactful on the surrounding environment.
BTW, according to the article the full GLM 5.3 model is in the "10x to 100x" ballpark. So this is not outside the realm of plausibility.
As we electrify and decarbonize, we are looking at going to 1000-1500GW average load (clean energy sources are 2x-6x more energy efficient than fossil fuels at delivering energy services , so looking at the primary energy flows here you'll not only see elimination of the "rejected energy" category, but heating by fossil fuels gets counted as ~100% effect when heat pumps are 200%-600% effect on those terms <a href="https://flowcharts.llnl.gov/" rel="nofollow">https://flowcharts.llnl.gov/ )
Going to 50GW of AI would be an amazing jump for which we don't have fab capacity anytime.
epistasis · · focus · HN ↗
> That model’s usage was well within our budget ($68, about 4kWh of energy use / 365 grams of carbon emissions).
The energy cost is literally 1% of the total cost. For context, 4kWh of energy would drive you about 15 miles in an EV, about half of the average person's daily driving miles. It's boiling 10 gallons of water.
With the talk of AI data centers' impact on the world, you'd think this would be 10x to 100x the amount of energy in order to get the effects they're using here.
My takeaway: the AI data center buildout is an overbuild probably at least as large as the fiber buildout that left us with so much dark fiber. If not even bigger. The only thing that will save the economy is the inability of NVIDIA and chip fabs to produce enough chips to match the buildout planned.
zozbot234 · · focus · HN ↗
BTW, according to the article the full GLM 5.3 model is in the "10x to 100x" ballpark. So this is not outside the realm of plausibility.
epistasis · · focus · HN ↗
As we electrify and decarbonize, we are looking at going to 1000-1500GW average load (clean energy sources are 2x-6x more energy efficient than fossil fuels at delivering energy services , so looking at the primary energy flows here you'll not only see elimination of the "rejected energy" category, but heating by fossil fuels gets counted as ~100% effect when heat pumps are 200%-600% effect on those terms <a href="https://flowcharts.llnl.gov/" rel="nofollow">https://flowcharts.llnl.gov/ )
Going to 50GW of AI would be an amazing jump for which we don't have fab capacity anytime.