Price Per GB of RAM: Normalising Memory Costs Across Budget VPS Plans

Advertised plan prices tell you almost nothing about value because they combine CPU, RAM, storage, and transfer into one number. If RAM is your actual constraint — and for most small web workloads it is — then the only useful metric is price per gigabyte of memory. Normalising that across budget plans exposes a spread that reaches 3x. Use our VPS comparison table to pull the raw plan data before you run the math yourself.

Why RAM Is the Binding Constraint

CPU is elastic on shared hosts — you get time-sliced bursts. Storage is cheap and largely fungible at NVMe speeds. RAM is the resource you cannot borrow: when it runs out, processes get killed or the kernel swaps and throughput collapses. That makes megabytes-per-dollar the metric that most directly predicts whether a plan will actually run your workload.

Computing Price per GB RAM

Take the monthly plan price and divide by provisioned RAM in gigabytes. A $5 plan with 1GB RAM is $5.00/GB. A $10 plan with 4GB is $2.50/GB. The second is twice as efficient per gigabyte, which is why “cheap” plans are usually the worst value.

Plan price	RAM	$/GB RAM	Notes
$3.50	512MB	$6.84	Worst ratio
$5.00	1GB	$5.00	Common entry tier
$7.00	2GB	$3.50	Sweet spot begins
$12.00	4GB	$3.00	Efficient
$22.00	8GB	$2.75	Best ratio
$42.00	16GB	$2.63	Diminishing absolute savings

The curve flattens around 4–8GB. Beyond that you are buying capacity you probably cannot use on a shared node without hitting CPU steal first — see our notes on noisy-neighbour effects for why. The efficient frontier for budget buyers sits at 2–8GB.

Adjust for What Actually Eats RAM

Provisioned RAM is not usable RAM. Subtract the baseline before comparing plans, because the overhead is fixed and hits small plans proportionally harder.

• OS + systemd + sshd: 150–250MB baseline on modern Debian/Ubuntu.
• Docker daemon (idle): 100–200MB before containers.
• Nginx + PHP-FPM (small site): 150–400MB depending on pool sizing.
• MySQL/MariaDB defaults: 300–600MB buffered — often the largest line item.
• A control panel: 200–500MB, which is why panels wreck 1GB plans.

On a 1GB plan with a LEMP stack you have roughly 400–500MB of usable headroom. On a 2GB plan the same stack leaves 1.4GB free. That is a 3x jump in working memory for a $2 price difference — a far larger practical gain than any CPU upgrade in the same bracket.

Price per GB Coupled With Bandwidth

RAM efficiency is necessary but not sufficient. A plan can lead on price-per-GB and still lose on transfer costs. Normalise both by computing a blended score: price divided by (RAM GB × included TB). This penalises plans that pair generous RAM with stingy bandwidth, which is a common budget-host tactic.

Plan	RAM	Transfer	Blended value
A: $7 / 2GB / 1TB	2GB	1TB	3.50
B: $7 / 2GB / 4TB	2GB	4TB	0.88
C: $12 / 4GB / 2TB	4GB	2TB	1.50

Plan B wins decisively despite an identical sticker to Plan A. Lower is better here. Applying this filter usually eliminates a third of the “cheap” plans people gravitate toward.

Practical Buying Rules

• Never buy a sub-1GB plan for anything running a database — the overlay leaves you with almost no headroom.
• Target under $3.50/GB RAM as a healthy budget benchmark in 2026.
• Prefer a 2GB plan at $7 over a 1GB plan at $5; the extra $2 buys three times the working memory.
• Skip RAM upgrades past 8GB unless your monitoring shows sustained pressure — you will hit CPU limits first.
• Recheck blended value at renewal; providers routinely shrink transfer allowances on the second term.

Bottom Line

How Providers Inflate the RAM Number

Not all gigabytes are equal. Some hosts quote “burst RAM” that is only available under light contention, then throttle you to a fraction of it when the node fills. Others advertise shared RAM that is guaranteed to a lower floor. Always read whether the figure is “dedicated” or “burstable” — the difference can be 2x in real terms, and it flips the price-per-GB ranking entirely.

You can detect burstable memory empirically. Allocate and touch most of your quota with a small script, then watch swap usage and latency over an hour. If performance degrades sharply as you approach the advertised ceiling, you are on burstable RAM and should price the plan at its realistic floor, not its headline number.

Reading the Price-Per-GB Curve for Upgrades

The curve itself tells you when to buy more memory. Savings per additional gigabyte peak in the 2–4GB transition and flatten after 8GB. If a provider charges the same per-GB rate across their whole lineup, that is a red flag — it usually means the entry tier is padded rather than the top tier discounted. Competitive hosts price the entry tier at a premium and push savings upward.

• Compute $/GB for every tier before choosing, not just the one you plan to buy.
• A lineup with a flat $/GB rate across tiers is rarely the best value at any size.
• Check whether the higher tier also raises CPU and transfer proportionally.

Price per GB of RAM, adjusted for baseline overhead and blended with transfer allowance, is the fastest way to separate real value from marketing. Entry-tier plans almost always lose; the 2–8GB band almost always wins. Run the numbers on our provider comparison before your next purchase — the spread between best and worst plans at the same price is wider than most buyers assume.