What is a fair price for a datacenter proxy?

For an exact 100-IP, 30-day datacenter purchase, four conflict-free dedicated or private rows cost from $145 to $170 total, or from $1.45 to $1.70 per IP. Four conflict-free general or shared rows at the same quantity and term cost from $3.50 to $100 total, or from $0.035 to $1 per IP. These eight rows were observed on 2026-07-25. Bright Data datacenter pricing, retrieved 2026-07-25T15:31:32Z · Oxylabs datacenter pricing, retrieved 2026-07-25T15:31:32Z · Decodo datacenter pricing, retrieved 2026-07-25T15:31:32Z · MyPrivateProxy plan evidence, retrieved 2026-07-25T17:42:41.156Z

That is an observed price range, not a declaration that every offer inside it is equally good. A fair comparison has to hold the product class, address quantity and billing period constant. Mixing a dedicated address with a shared address, or a 30-day subscription with a traffic-priced rotating pool, produces a larger range but a weaker answer.

The snapshot contains 710 plan rows across 47 providers. The panel above is deliberately narrower: datacenter only, a single proxy type, exactly 100 IPs, exactly 30 days, a published checkout, and no recorded source conflict. Rows whose price is not published, rows sold across several proxy types and configurable rows without an exact 100-IP checkout are not treated as matching observations. The complete observed plan data keeps those exclusions visible.

The exact 100-IP panel

The unit price below is published checkout / 100 IPs. It is a computed comparison, not an advertised floor borrowed from a larger order.

The dedicated or private range is much tighter than the general or shared range. All four dedicated or private observations sit between $145 and $170 for the checkout and between $1.45 and $1.70 per IP. The general or shared observations span $3.50 to $100 and $0.035 to $1 per IP. Those figures come from the same eight rows and the same retrieval timestamps linked in the table.

Shared and dedicated prices answer different questions

A buyer who needs assigned capacity must first decide whether sharing is acceptable. The dataset labels Bright Data and Oxylabs rows as shared or dedicated, Decodo publishes separate general datacenter and dedicated datacenter rows, and MyPrivateProxy publishes shared and private families. Those labels are why the page reports two ranges instead of averaging all eight checkouts.

An average would hide the most important purchasing choice. It would also make the Decodo general row pull the combined result sharply downward. Keeping the classes visible lets a buyer ask the useful follow-up: does the cheaper row satisfy the same identity and allocation requirement? Price data can show the label and the bill. It cannot prove that two networks will perform equally on a particular target.

Published limits belong beside the price

Capacity disclosure is not uniform even in this narrow panel. Decodo explicitly publishes unrestricted threads and concurrent sessions on both rows. MyPrivateProxy publishes 100 threads per proxy on both rows. Bright Data and Oxylabs do not publish a concurrency value for these exact rows in the snapshot. “Not published” is an unknown state, not permission to assume unrestricted use.

The same discipline applies to bandwidth. MyPrivateProxy marks monthly bandwidth as unrestricted while also publishing a fair-use statement that is not quantified into a usable monthly allowance. The other rows do not provide one common bandwidth field that supports a second normalization. The per-IP range therefore must not be presented as a per-GB range.

Terms can also change the amount paid at checkout. The Decodo evidence says tax is added, so its published checkout is not a tax-inclusive total for every buyer. Tax treatment is not published consistently enough across the full panel to calculate a universal after-tax range. A fair quote should preserve that distinction instead of silently treating an unknown tax field as zero.

A practical fairness test

Start with an exact quantity and term. Keep shared offers separate from dedicated or private offers. Require a real checkout total, then divide by the purchased IP count. Put concurrency, threads, bandwidth state, tax state and refund terms beside the result. Finally, exclude source-conflicting rows until the vendor evidence agrees.

For the exact configuration on this page, a dedicated or private quote can be checked against the observed $145 to $170 checkout range, while a general or shared quote can be checked against the observed $3.50 to $100 range. Those are useful reference bands, not promises about quality or future pricing. A quote outside either band may still be rational, but the buyer should be able to identify the extra capacity, location, support or contract term that makes it a different product.