An honest, fully-metered accounting of what a local-AI and 4K-video studio actually uses — built from real Louisiana utility bills, with nothing hidden. My generation runs local. My cloud use is small, and I count every gram of it.

Power bill
~0%
My AI + 4K studio's share of the bill
Carbon
0.0 t
My digital footprint per year
Water
~0 L
On-site AI cooling water (sealed loop)
Sovereignty
0
Data centers my local AI uses
The bill

My real electric bill

Whole-house usage for one billing month, straight off the meter at a 76°F average with the AC running. The air conditioner runs this bill — my studio is the green sliver.

0 kWh
This month
$0
Total bill
$0
Effective rate / kWh
0 kWh
Average per day
AI ~8%
Air conditioning, appliances & everything else ~92%

AI + 4K studio: ~160 kWh  ·  AC + appliances + Dad's TV: ~1,739 kWh  ·  metered on 3× Panamax M5300-PM

Where the power goes

Where the 1,899 kWh actually goes

Two people, one Louisiana house. AC is the giant — my studio and Dad's home theater are small, similar slices.

My studio (metered) Dad's home theater (est.) House loads (est.)
Air conditioning
850 kWh · 45%
Lighting + plug loads
280 kWh · 15%
Home theater (Dad, sports)
190 kWh · 10%
Water heating
190 kWh · 10%
AI + 4K video studio
160 kWh · 8%
Refrigerator + freezer
95 kWh · 5%
Washer + dryer
90 kWh · 5%
Stove / oven
44 kWh · 2%
Honest note: AC and appliances are estimated from EIA end-use shares (summer-weighted) — they aren't individually metered. Only the high-value electronics (my studio and the home theater) sit on Panamax conditioners. The point: my dad's sports streaming (~10%) now edges out my AI studio (~8%), and the air conditioner still dwarfs them both.
The comparison

How my house compares

Louisiana is the #1 state for home power use — because of heat and AC, not AI. My peak-summer month sits right where a Louisiana household sits.

My house (peak summer month) State / national averages
My house — summer
1,899
Louisiana avg (#1)
1,231
Tennessee avg (#2)
1,185
Mississippi avg (#3)
1,172
U.S. average
886
Hawaii avg (lowest)
503

kWh / month  ·  state figures are annual averages; like every Southern home, my summer runs above my own yearly average — and only ~8% of it is my studio.

Carbon

My digital footprint, line by line

My entire digital life at Louisiana grid intensity (≈0.4 kg CO₂/kWh): ≈ 1.2 tonnes a year. My own power is local; my cloud services run in data centers — counted, not hidden.

My own power (local) Cloud / data-center services
AI + 4K video studio
770 kg
Home media server (24/7)
140 kg
Google Drive — 15 TB work
~150 kg
Claude — coding, few hr/day
~100 kg
Social media — 2 hr/day
22 kg
My streaming — 3 hr/week
9 kg
For perspective, 1.2 tonnes is about one and a half round-trip flights across the country, or ~8 miles of driving a day — roughly 7–8% of an average American's yearly footprint (≈15 tonnes).
Water

My bidet vs my AI

The comment that started this whole thing — checked against the real numbers. There are two kinds of water: the kind a machine evaporates, and the kind a power plant does.

AI cooling (on-site) My bidet AI grid water (indirect)
AI on-site cooling
~0 L/yr
My bidet (2 people)
~2,000 L/yr
AI grid water
~4,500 L/yr
The verdict
Right — on the water that actually counts.
My cooling loop is sealed (a one-time ~1–2 L fill that recirculates for ~7 years), so it evaporates ~0. My bidet beats it thousands of times over. The only way I'm "wrong" is counting the indirect grid water behind my electricity — a different thing, and even that flips with heavy bidet use.
AI power

My local AI power vs the cloud

Image and video generation on an RTX 5090 — same ballpark as cloud models, and well under the heaviest ones. And it never leaves the building.

My RTX 5090 (local) Cloud models
My rig — quick image
0.8 Wh
My rig — detailed flyer
~4 Wh
Cloud — average model
2.9 Wh
Cloud — heaviest model
11.5 Wh
The comparison the critics skip
5 hours gaming on this card ≈ 2.6 kWh.
That's the same energy as generating 1,000+ AI images. The footprint people worry about is one-time training and the scale of billions of daily queries — not one person making flyers at home.
Owned vs streamed

What I own vs what they stream

Every stream re-downloads through a data center. My discs play from the shelf — and my own dad is the real-world heavy streamer.

0
Vinyl records
0
LaserDiscs
0
Blu-rays
0
Data centers for my media
My dad (same house) Typical streamers My owned library
My dad (8–12 hr/day sports)
~15 TB
Typical 4K streamer (3 hr/day)
7.7 TB
Typical HD streamer (3 hr/day)
3.3 TB
My owned library
0 streamed

terabytes pulled from the internet each year  ·  buy once, play forever — every replay pulls zero new data.

The household

My dad's footprint, broken down

He uses a lot — so here's all of it, honestly. Sports streaming 8–12 hours a day on a big home theater adds up.

8–12 hr
Sports streaming / day
0 TB
Data pulled / year
0 kWh
Home theater power / mo
0 t
His carbon / year
His home theater power Data-center streaming
Home theater electricity
~910 kg
Sports streaming (data centers)
~200 kg
The honest comparison
Dad's ~1.1 t ≈ my entire ~1.2 t digital footprint.
A heavy TV-and-sports habit rivals my whole AI, 4K, cloud, and coding workflow — in one house. His streaming leans on data centers; my AI doesn't. AI isn't the uniquely bad guy.
The honest ledger

Let's be honest about my data-center use

I'm not pretending I never touch a data center. My generation is sovereign — local, zero data centers. But I code with Claude and store work on Drive, and those run in the cloud. So I count them.

Sovereign — local hardware
0 data centers
AI image & video generation (ComfyUI, RTX 5090)0 DC
4K video editing & rendering0 DC
Home media server & owned library0 DC
On-site cooling (sealed loop)~0 L
Cloud — counted, not hidden
~280 kg/yr
Google Drive — 15 TB of work content~150 kg
Claude — coding assistant, a few hr/day~100 kg
Social media — 2 hr/day22 kg
My streaming — 3 hr/week9 kg
The bottom line
My whole cloud footprint is ~280 kg a year — and it isn't even the biggest data-center load in this house.
That title goes to my dad streaming sports 8–12 hours a day (~1.1 t). My AI generation is the one thing in my life that never touches a data center. Everything else, I own up to — measured, sourced, and on the page.