When Google ended unlimited free storage for Google Photos in June 2021, a lot of us did a double-take. Storage had felt free forever. So I got curious: what does it genuinely cost Google to host everyone’s photos — and is the service actually profitable?
This is an estimation exercise, not an internal Google financial report. I’ll be upfront about which numbers are verified facts and which are reasoned guesses. But even rough math at this scale is revealing.
A note on why I refreshed this: I originally ran these numbers back in 2021. Since then the inputs have changed dramatically — Google Photos more than doubled its photo count, and storage hardware got far denser and cheaper. So this isn’t a reprint; the conclusions genuinely shifted.
What’s Changed Since 2021
Here’s how the key inputs have moved — and these are verified figures, not estimates:
| Metric | 2021 | 2025–2026 |
|---|---|---|
| Monthly active users | ~1 billion | 1.5 billion |
| Total photos & videos stored | ~4 trillion | 9+ trillion |
| Typical high-capacity data-center HDD | ~10 TB | 30 TB+ |
| Street price per TB (high-capacity HDD) | ~$20/TB | ~$10–17/TB |
Sources: Google’s 10th-anniversary announcement (May 2025) for user and photo counts; 2026 enterprise storage market reports for drive capacity and pricing.
The takeaway before we even do the math: Google is storing more than twice as many photos, but each modern drive holds 3x more data at a lower price per terabyte. Density and cost-efficiency outran the growth in volume.
How Much Data Are We Talking About?
Google offers two upload options: Original quality and Storage saver (formerly “High quality”), which caps images at 16 megapixels. For most people, 16 MP is plenty — that’s a 4000×4000 square image.
To estimate average file size, I downloaded my own complete Google Photos archive (all years, including videos) — about 164 GB — and wrote a small script to compute the mean photo size: 0.63 MB. Since I shoot at higher resolutions than the average user, I’ll round down to 0.5 MB as a conservative per-photo average.
The math:
- 9 trillion items × 0.5 MB ≈ 4.5 Exabytes of photo data.
(An exabyte is a billion gigabytes. “Exa” is the unit you reach for when “tera” and “peta” stop being big enough.)
That’s the headline number. Now let’s turn it into hardware.
How Many Drives Does That Take?
Modern hyperscale data centers use 30 TB-class drives. At 4.5 Exabytes:
- 4.5 EB ÷ 30 TB ≈ 150,000 drives for a single copy of the data.
But you never store one copy. Google replicates data for durability and availability — a standard safety factor is roughly 3x:
- 150,000 × 3 ≈ 450,000 drives.
Compare that to my 2021 estimate of ~600,000 smaller 10 TB drives for less than half the data. Denser drives mean Google needs fewer physical units today despite storing far more — a vivid illustration of why storage costs don’t scale linearly with data growth.
The Hardware Bill
At current high-capacity HDD pricing of roughly $10–17 per TB, and 13.5 Exabytes of raw drive capacity (4.5 EB of data × 3 for replication):
- 13,500,000 TB × ~$12/TB ≈ $162 million in raw drive cost.
Of course, drives don’t run on their own. You need server racks, CPUs, RAM, network gear, and cooling. Factoring in full server infrastructure rather than bare drives, a realistic estimate lands in the range of:
- $300 million to $500 million in total storage hardware.
Estimate, not fact. This is back-of-the-envelope. Google buys at a scale and with custom hardware that ordinary pricing doesn’t capture, almost certainly lower per unit than retail. Treat this as an order-of-magnitude figure.
The Electricity Cost
A single enterprise HDD draws roughly 8–9 watts. For 450,000 drives:
- 450,000 × ~8.6 W ≈ 3.9 megawatts just for drives.
Add CPUs, memory, networking, and cooling — typically doubling total draw:
- ≈ 7.8 megawatts continuous.
Google reports paying below-market rates and buying renewable energy, but using a conservative blended rate:
- Roughly $9 million to $18 million per year in electricity.
Summary
Let me remind you that “Google Photos” stores also videos and the file size might be much bigger than photos. Also “Google Photos” needs to analyze each photo, video, to check all the faces and label all the objects. So I would multiply infrastructure costs by x2 or x3. And here are the final calculations:
- Infrastructure costs between $400 mln and $600 mln.
- The electricity between $9 mln and $18 mln per year.
- Expenses for the Google Photos engineering team is around $17.5 mln per year
Is Google Photos a profitable service? If they have 1 billion active users, and let’s assume 3% of them pay for the service, i.e. 30 mln users paying for the service, the minimum paid plan is $20 per year. 30 mln x $20 = $600 mln per year of profit. I guess it should be profitable 🙂
