Why “Unlimited” Hosting Plans Aren’t Really Unlimited
Every hosting company's pricing page has one, usually in the largest font on the slide: unlimited storage, unlimited bandwidth, unlimited websites. It's one of the most successful pieces of marketing language the hosting industry has ever produced, and it is also, almost without exception, not literally true. Understanding why requires understanding what shared hosting actually is: a physical server, with a fixed amount of CPU, RAM, disk I/O, and network throughput, divided among hundreds or sometimes thousands of accounts.
The Math That Makes "Unlimited" Possible
Unlimited plans work on the same principle as an all-you-can-eat buffet or a gym membership: the business is betting that the average customer will use far less than the theoretical maximum, and that the aggregate cost across all customers stays predictable. A typical shared hosting node might host 300–800 accounts. If even a small fraction of those accounts genuinely tried to use unlimited storage or unlimited CPU cycles simultaneously, the server would fall over in minutes. So hosts build in invisible ceilings — not because they're being dishonest exactly, but because the entire pricing model depends on nobody actually testing the word "unlimited" at face value.
These ceilings show up in a few concrete forms. The most common is a process and CPU-second limit enforced at the account level, often through tools like CloudLinux's CageFS and LVE (Lightweight Virtual Environment), which silently throttle an account's PHP processes once it crosses a threshold — frequently somewhere in the range of 25–100% of a single CPU core, and a fixed number of concurrent processes, commonly 20–40. Cross that line consistently, and pages start timing out or returning 508/509 errors, even though your storage usage looks nowhere near "full."
Inodes: The Limit Nobody Explains
The second hidden ceiling is inodes — the filesystem structures that track every single file and folder on your account, including every cached thumbnail, every log file, every email in every folder, and every file inside every WordPress plugin. A host might advertise 50GB of "unlimited" disk space but cap you at 200,000–250,000 inodes. A WordPress site with a large media library, an aggressive caching plugin, and a few years of Woocommerce order history can hit six figures in inode count while using only a few gigabytes of actual storage. When that ceiling is hit, uploads start failing with cryptic "disk quota exceeded" errors that have nothing to do with disk space.
Bandwidth Isn't Really Unlimited Either
"Unlimited bandwidth" typically means the host doesn't meter and bill overages the way a cloud provider does — but every network port on a server has a physical cap, usually shared across every account on that node. If your site suddenly gets featured somewhere and traffic spikes 50x, most hosts reserve the right to throttle your connection speed, move you to a different server tier, or ask you to upgrade, citing an "acceptable use policy" that's usually linked in eight-point font at the bottom of the terms of service.
Where to Find the Real Numbers
The honest version of a hosting plan's limits rarely lives on the pricing page — it lives in the Acceptable Use Policy (AUP) or Terms of Service, and sometimes only becomes visible once you're logged into the control panel and can see resource usage graphs. Some things worth checking before signing up:
- Search the host's knowledge base for "LVE," "CageFS," or "resource limits" — most CloudLinux-based hosts document their default tier limits somewhere, even if not on the pricing page.
- Ask support directly what the CPU/process limits are for the plan you're considering. A host that answers precisely is usually more trustworthy than one that repeats "unlimited" back at you.
- Check whether the control panel shows a resource usage widget (cPanel's "Resource Usage" tab, for instance) — this is where the real ceiling becomes visible after the fact.
Why This Isn't Necessarily a Scam
It's worth separating "unlimited is imprecise marketing" from "unlimited is a scam." For the overwhelming majority of small sites — brochure sites, low-traffic blogs, small stores — the fair-use ceilings are never actually reached, and the value proposition of unlimited plans holds up fine. The problem only shows up at the edges: sites with huge media libraries, high-traffic blogs, or anything running background cron jobs and API calls at scale. If that describes your project, the right move isn't to avoid unlimited plans altogether — it's to ask pointed questions about actual resource ceilings before you commit, and to treat "unlimited" as a starting point for a conversation rather than a literal spec.
A Concrete Illustration
Picture two accounts on the same "unlimited" plan, on the same physical server. The first is a five-page consulting firm's brochure site, updated twice a year, receiving a few hundred visits a month. The second is a busy regional news blog publishing daily, running several plugins, embedding video, and pulling in tens of thousands of monthly visitors. Both pay the identical monthly fee, both were sold the identical word "unlimited," and yet only one of them is realistically anywhere near the invisible ceiling the plan's pricing model was built around. This is precisely why unlimited plans can be simultaneously a completely fair deal and a source of legitimate frustration — the word is doing two very different jobs depending on which of these two accounts is reading it.
It also explains a pattern many long-time hosting customers eventually notice: a site that ran fine for years on an entry-level "unlimited" plan suddenly starts hitting resource warnings, not because the host changed anything, but because the site itself crossed from the first category into the second through completely organic growth — more content, more plugins, more visitors. The plan's marketing never changed; the account's actual resource footprint did.
What Reading the Fine Print Actually Reveals
When hosts do disclose fair-use ceilings, they're usually phrased defensively rather than as hard numbers — language like "accounts must not use a disproportionate share of shared server resources" or "excessive resource consumption may result in service limitations," without a specific CPU percentage or process count attached. This vagueness is itself informative: a host that can't or won't state a number typically applies limits reactively, on a case-by-case basis, which means two accounts with identical usage patterns could theoretically receive different treatment depending on overall server load at the time. Hosts running CloudLinux-based infrastructure are the exception — because the limits are enforced programmatically per account, many are willing to state the exact numeric ceilings for each plan tier if asked directly, since the number already exists in their provisioning system rather than being an after-the-fact judgment call.
The Takeaway
"Unlimited" in hosting is closer to a pricing philosophy than a technical fact. It reflects a bet that most customers won't approach the real limits of the shared infrastructure underneath the plan. That bet is usually a safe one — until your site grows past what the fine print was written to accommodate, at which point the invisible ceiling becomes very visible, very fast.
Tags: fair use policy, hosting plans, shared hosting