“Modular” is the hardest-working, least-supervised word on case spec sheets. A testable definition — removable, replaceable, serviceable — why a one-cent rivet vetoes all three, and what honest modularity costs.
Read ten case product pages and you'll find the word "modular" on nine of them. A removable drive cage: modular. A repositionable SSD bracket: modular. A front panel that pops off for cleaning: modular. The word has become the hardest-working, least-supervised employee in PC case marketing.
Here's the tell: furniture stores don't call a bookshelf "modular" because you assembled it once with the included hex key. Assembly isn't modularity. LEGO is. The difference isn't whether the thing comes apart during manufacturing; it's whether it's designed to come apart and go back together, repeatedly, in your hands, for years.
So before this article argues anything, let's pin the word down. We'd propose "modular" should have to pass three tests, in increasing order of respect for the owner:
Most "modular" cases pass the first test on a few accessories. Very few pass all three on the parts that matter, because of a component that costs about one cent, and that this whole article is really about.
The three-tier test compresses to one question: after something breaks, bends, or bores you, what is the cheapest path back to a perfect case? For a genuinely modular product the answer is "a part and a hex key." For most cases the honest answer is "a new case," which means the modularity was decoration, not architecture.
Sources: the philosophy this test descends from
Flip your case around and look along the edges of the motherboard tray. Those small, round, slightly domed buttons (usually a dozen or more of them) are blind rivets, and they are why your case's structure is effectively welded shut.
First, respect where it's due: rivets aren't a scam. They're honest cost engineering. A blind rivet (the "pop" rivet) is set from one side of the panel in about a second with a pneumatic gun: no nut to hold, no torque to verify, no thread to strip, no fastener to rattle loose in shipping. For a factory assembling thousands of enclosures a day, rivets are fast, cheap, consistent, and permanent. Three of those four properties are exactly what a manufacturer wants.
The fourth is the problem, and it's not a side effect. It's the design intent. A rivet is a one-way fastener. Setting it deforms the metal so the joint can never be undone, only destroyed: to remove a rivet you drill its head off, and to rejoin the panels you need a rivet gun and fresh rivets, tools and parts almost no PC builder owns.
A blind rivet is set by pulling a mandrel through the rivet body until the blind-side flange mushrooms and the mandrel snaps off: permanent by design, installable from one side only, seconds per joint. Removal means drilling out the head, which consumes the fastener and often enlarges or scars the hole. Rivets are excellent engineering for joints that should never open. The question a case buyer should ask is simply: is my case's structure such a joint?
Sources: how the one-way fastener works
Nothing, right up until the first time anything changes.
A riveted case that ships perfect, is used gently, and never gets modified will never remind you of its rivets. The trouble is that "nothing ever changes" is a strange bet to make about a product enthusiasts keep for five to ten years, across two or three complete builds.
Then the changes come. A motherboard standoff strips. A tray flexes under a 2-kilogram graphics card that didn't exist when the case was designed. A shipping drop puts a crease in a structural panel. You want to cut a proper hole for a cable that the designer never planned for, or you simply want the layout to do something new. In a bolted case, each of these is a part swap or an afternoon. In a riveted case, each becomes the same dead end: the damaged part is structurally fused to every other part, there is no spare to buy, and "repair" means drill, improvise, or replace the whole enclosure.
It's the sealed-drawer cabinet: a beautiful chest where one jammed drawer means the carpenter expects you to buy a new chest. Nobody would accept that in furniture. In cases, we've simply stopped noticing, because the rivets have been there for forty years, from the same office-machine assembly lines that gave us the rectangular tower in the first place (that history is question one of this book).
One cent of fastener, quietly vetoing every repair the rest of the design would have allowed. That's the real cost of a rivet, not what it does, but what it forbids.
Swap every structural rivet for a bolt and nut, and here is the honest ledger of what changes, in both directions.
| Property | Blind rivet | Bolt + nut |
|---|---|---|
| Disassembly | Destructive (drill out) | Reversible, unlimited times |
| Assembly speed | ~1 second, one-sided | Slower, often two-handed |
| Fastener cost | ~1 cent | Several cents + hardware |
| Loosening over time | Never | Possible (needs lock hardware) |
| Repair & upgrade path | None | Any part is a spare part |
Read the middle rows first, because they're the manufacturer's side of the argument and they're real: bolts are slower to install, cost more per joint, and a badly designed bolted joint can loosen where a rivet never would. A bolted case is more expensive to make even before anyone uses its advantages.
Then read the first and last rows, because they're yours. A bolted structure means the case is no longer one fused object but a kit of parts that happens to arrive assembled: seconds to unmake what took minutes to make, instead of minutes of drilling to unmake what took seconds to rivet.
And if "consumers can't handle screws" sounds plausible, the counterexample already exists in a far harder product category: Framework builds laptops (the most glued-shut, integrated category in consumer tech) held together with ordinary screws, ships the screwdriver in the box, and sells every internal part as a spare. iFixit scores their machines 10/10 for repairability, and the company is thriving. If a 15-millimetre-thin laptop can be honestly modular, a 25-litre steel-and-aluminum box has no excuse.
The loosening concern is solved engineering, not an open problem: nylon-insert locknuts, split washers, serrated flanges, or thread-locking compound hold bolted joints through vibration, and in thin sheet metal, repeated service cycles are handled with nuts or steel threaded inserts rather than threads cut into the sheet itself. (Steel hardware also sidesteps the stripped-aluminum-thread problem from the materials question.) The engineering cost of serviceable fastening is real but small; the decision is economic, not technical.
Sources: the existence proof and the hardware
Because the case is the component you'll own the longest, and the industry treats it as the most disposable.
Think about the lifecycle honestly. CPUs and GPUs turn over every few years. Drives fill up and get replaced. But a good enclosure outlives all of them: it's just metal, and metal doesn't obsolete. A case bought today could plausibly host three complete builds across a decade, if it can adapt and be repaired along the way. A riveted case with no parts supply gets one accident or one incompatible future standard, and becomes seven kilograms of scrap with tempered glass.
The wider world has stopped accepting this. The EU's Right to Repair directive (adopted in 2024, applying across member states from the end of this very month, 31 July 2026) now obliges manufacturers of covered products to repair them at reasonable cost, supply spare parts at reasonable prices, and stop using "hardware techniques that impede repair." PC cases aren't on the covered-products list (yet), but the direction of travel is unambiguous: designed-in unrepairability is becoming a defect, not a business model.
A brand can wait to be dragged there by regulation, or design that way because it always made sense. Either way, the era of the one-cent fastener that writes off the whole product is ending.
Directive (EU) 2024/1799, adopted 13 June 2024, applies from 31 July 2026: repair obligations within reasonable time and price for products listed in its Annex II, spare-parts access at reasonable prices, and a ban on contractual, hardware, or software techniques that impede repair, with a European repair platform to follow. Cases aren't in Annex II today; the norm it sets (parts, openness, no repair-blocking design) is exactly the three-tier test from section 1.
Sources: the regulation, primary
They're related, but one is the foundation and the other is the house.
Customization without modularity is paint. You can wrap, spray, or sleeve a sealed case, cosmetics on a fixed object, undone the day your taste changes. It's fine; it's also all the industry usually means by "make it yours."
Customization with modularity is different in kind: when panels mount on standard, reversible fasteners, a panel isn't a surface, it's a slot. A slot can accept the stock part, a different-material part, a community-designed part, or the one you printed at 2 a.m. The case stops being a finished sculpture and becomes a platform that accumulates possibilities over its life instead of losing them.
That's why the fastener decision from section 4 quietly decides the customization question too, and why "why did PC cases become boring, and how do we fix it" gets its own question later in this book. Modularity is the load-bearing wall; everything fun is built on it.
Time to hold ourselves to the standard this article just set.
The Octa V1 contains no structural rivets. The internal steel skeleton is joined with hex screws and nuts. Every structural member can be unbolted, replaced, or upgraded with one ordinary tool. The outer 3 mm aluminum panels don't even need the tool: they mount magnetically and lift off by hand, which means the most damage-prone, most style-defining parts of the case are also the most trivially swappable. A modular bay accepts optional components (including the optional touchscreen module), and because panels are discrete parts on standard mounts, we're exploring publishing 3D-printable panel files so the community can make parts we never imagined. Removable, replaceable, serviceable: the three tests from section 1 were design requirements, not marketing copy.
And the honest cost accounting, because this book doesn't do free lunches: bolted assembly is slower, hex hardware costs real money at case scale, magnetic mounting costs more than a bent-steel clip, and some of the Octa V1's price tag is exactly this choice. We think that's the correct trade for a product designed to be owned for a decade, but it is a trade, and you should know you're paying for it. (What material those bolts hold together, and why the skeleton is steel while the skin is aluminum, is the previous question.)
The checklist, portable to any spec sheet or showroom floor:
A case that passes all five is modular in the sense that should count. A case that passes none can still be a great case, so buy it knowing it's a sealed product with a warranty, not a platform.
No, for a joint that never needs to open, a properly set rivet is excellent: light, vibration-proof, and consistent. The argument against structural rivets in cases isn't strength; it's that they make repair, replacement, and upgrade of the structure impossible without a drill.
Not if it's engineered rather than just assembled: nylon-insert locknuts, serrated flanges, and correct torque hold joints through years of fan vibration. It's the same solved problem as your car, which vibrates far harder than your PC and is held together almost entirely by reversible fasteners.
Hardware cost (bolts, nuts, inserts, magnets cost multiples of rivets), slower assembly, and the obligation to actually stock spare parts. You're paying for the second and third life of the product at the first sale. Whether that's worth it depends entirely on how long you keep hardware.
Not at all. At budget prices, rivets are part of why the price is possible, and that's an honest trade. It becomes dishonest only when a riveted, parts-less product wears the word "modular" because a drive cage slides out.
One common tool for the structure (a single hex size, ideally), zero tools for the parts you'll touch monthly: panels, filters, the parts that get cleaned or swapped. Tool diversity is friction, and friction is how "repairable in theory" becomes "never repaired in practice."
"Modular" should mean something testable: parts that come out with a common tool, go back in without drama, and can still be bought, or printed, years later. Most of what wears the word today fails the test, and the failure usually traces back to a one-cent, one-way fastener holding the whole structure hostage.
Rivets are honest engineering for products meant to be sealed. A PC case (the longest-lived component you own, host to a decade of builds) is the last product that should be sealed. Bolts, nuts, magnets, and a real parts supply cost more to build and slightly more to buy, and in exchange the case stops being a purchase and becomes a platform.
We built the Octa V1 that way, we've told you exactly what it costs, and next in this book we'll show you the messy road there: seventeen prototypes' worth of mistakes, dead ends, and lessons, because a repairable case, it turns out, is built by a very error-tolerant process.
Explore the Unknown.
// Explore the Unknown
© 2026 Olicus Technology & Gaming AB. All rights reserved.
Explorers of the Unknown