For starters, growing numbers of companies are required to report on their own emissions, which includes software - and therefore having agreed standards on how to measure this would seem a good way forward to me?
> carbon emissions are also proportional to energy
Reducing energy usage reduces carbon emissions, but they are not proportional. Highly dependent on the grid composition, time of day you draw the energy, and any behind-the-meter infrastructure.
> and all large operators do catalog and reduce GHG emissions.
To take AWS as an example, they still do not report scope 3 emissions for customers (due 'early 2024') - without which, these 'catalogued' numbers are essentially meaningless given how understated they are.
Just be careful not to double-count hardware and software emissions.
And what about technology which is controversial, causing a lot of discussions online (all those people spending energy to post comments…)?
I think cost has nothing to do with it.
If the data centre entirely off-grid, the SCI permits using your own grid intensity factors.
I'm not clear on the scenario where the solar is directly connected to the data centre 'behind the meter', or if there are other forms of peak shaving, while still drawing on the grid when needed - does anyone else care to comment/clarify the standards intentions there in terms of incentives?
Well...
I'm strongly against inefficient software for different reasons, but if stuff like this can have an effect beyond the virtue-signaling bureaucracy it's likely to become instead, maybe it's a good thing.