The last mile of fiber infrastructure isn't the street—it's the vertical climb inside a multi-dwelling unit (MDU). Once fiber reaches a building's basement, it faces a new set of challenges: private property rights, congested riser conduits, and the split-incentive problem where developers avoid pre-wiring costs that future tenants pay many times over. The typical solution is a distributed split architecture: a high-count fiber cable runs up the main riser, with distribution hubs at each floor branching out to individual apartments via passive optical splitters. This passive optical network (PON) design requires no powered equipment on floors, reducing maintenance. But in older buildings, installers must navigate packed conduits, drill through post-tension concrete slabs, and negotiate right-of-entry agreements with building owners who may treat access as a revenue stream. The result is a micro-monopoly dynamic where landlords can charge ISPs per-subscriber fees or demand exclusivity. Some jurisdictions like San Francisco and the EU are mandating fiber-ready infrastructure in new construction, but globally the patchwork means that connecting a single unit can cost anywhere from $200 to over $1,000—far more than the $50–200 it would have cost to pre-wire during construction.
#3711: The Hidden Last Mile: Fiber in Skyscrapers
Getting fiber to the 37th floor is a messy tangle of risers, contracts, and concrete.
Episode Details
- Episode ID
- MWP-3890
- Published
- Duration
- 35:28
- Audio
- Direct link
- Pipeline
- V5
- TTS Engine
-
chatterbox-regular - Script Writing Agent
- deepseek-v4-pro
AI-Generated Content: This podcast is created using AI personas. Please verify any important information independently.
Downloads
Transcript (TXT)
Plain text transcript file
Transcript (PDF)
Formatted PDF with styling
You Might Also Like
Never miss an episode
New episodes drop daily — subscribe on your favorite platform
New to the show? Start here#3711: The Hidden Last Mile: Fiber in Skyscrapers
Daniel sent us this one about fiber infrastructure, specifically what happens inside big buildings. He's pointing out that the last mile is where fiber gets genuinely complicated — not the ocean crossings, not the national backbones, but the final stretch from the street to your apartment. And he's asking the specific question: in a skyscraper, once the building itself is connected, is it straightforward to wire every unit, or is there a whole hidden layer of infrastructure and contractual mess to get fiber up to the thirty-seventh floor?
The short answer is that the inside of a large building is its own miniature last mile problem. And in a lot of ways it's messier than the street-level deployment, because the street has regulated utility access and the inside of a building is private property where every square inch of conduit space is contested.
Contested by whom? The ISP versus the building owner?
ISP versus building owner versus other utilities versus the original architects who never planned for fiber. The term you hear in the industry is MDU — multi-dwelling unit — and MDU fiber deployment is a whole sub-specialty. You've got fiber-to-the-curb, fiber-to-the-building, fiber-to-the-home, and then this awkward middle child called fiber-to-the-floor or fiber-to-the-riser that nobody puts on marketing brochures.
Marketing doesn't love "we got it most of the way there.
So let's walk through what actually happens. When an ISP brings fiber to a large building, the external connection terminates in what's called the building entry point — usually a basement or ground-floor telecommunications room. That's where the outside plant meets the inside plant. And that boundary is a hard legal line. Everything street-side is the ISP's network, regulated by franchise agreements and utility access laws. Everything building-side is private property, governed by the building's own wiring policies, ownership structures, and whatever access agreements the ISP managed to negotiate.
The fiber enters the basement and then... stares at a locked door?
The next piece is the riser — the vertical pathway that runs up through the building carrying cables between floors. In modern construction, especially in buildings designed after about two thousand ten, there's usually dedicated telecommunications riser space: conduits, cable trays, sometimes full equipment closets on every floor. In older buildings, you might be looking at repurposed elevator shafts, old chimney voids, or just drilling through concrete and hoping the structural engineer doesn't get angry.
"Hoping the structural engineer doesn't get angry" is the unofficial subtitle of most retrofit projects.
Here's where the first big infrastructure challenge shows up: riser congestion. Conduit space in vertical risers is finite. You've got electrical, plumbing, HVAC, fire suppression, security systems, and then data. In a building that wasn't pre-wired for fiber, the existing conduits might already be packed with copper phone lines, coaxial cable, and old ethernet backbones. Finding a clear path from the basement to the thirty-seventh floor is not trivial.
I assume you can't just... hang fiber off the side of the building?
You can, and in some developing-world deployments they absolutely do. External cable routing along building facades is common in parts of Southeast Asia and South America. But in most regulated markets, external routing triggers building code issues, aesthetic restrictions, and weather exposure problems. So internal riser pathways are the standard.
You've got this vertical spine. How does it actually branch out to individual apartments?
The most common model in large MDUs is a distributed split architecture. You run a high-count fiber cable up the main riser — anywhere from twelve to two hundred eighty-eight individual fiber strands bundled together. At each floor, or sometimes every few floors, you install a fiber distribution hub. That's a small, secure enclosure — maybe the size of a briefcase — where individual fibers from the main riser cable are split out and connected to drop cables that run horizontally to each apartment.
It's literally the hub-and-spoke model Daniel mentioned, but for connectivity rather than smart hotel room controllers.
The terminology varies by region — fiber distribution hub in North America, floor distribution point in Europe — but the concept is identical: a central vertical spine with horizontal branches at each floor.
What's the actual hardware inside one of those floor-level boxes?
Mostly passive optical components. Optical splitters, which take one incoming fiber and split the signal to multiple outgoing fibers — typically one-to-eight, one-to-sixteen, or one-to-thirty-two splits. Connector panels for patching. Maybe splice trays where fibers are physically fused together. The key thing is that in a passive optical network — which is what most fiber-to-the-home deployments use — there's no powered equipment on the floor. No switches, no routers, no amplifiers. It's all passive glass and plastic, which means nothing to maintain, nothing to break, nothing that needs a power outlet.
Passive optical network. That's PON, right?
PON is the dominant architecture for residential fiber. The alternative would be active ethernet, where you run dedicated fibers to each unit from a powered switch somewhere in the building. Active ethernet gives you higher dedicated bandwidth per subscriber, but it costs more, requires powered equipment rooms on multiple floors, and adds maintenance complexity. For most residential deployments, PON wins on economics.
If I'm a developer building a new fifty-story residential tower, I just spec this into the plans and everything's clean. But what about existing buildings?
That's where the contractual and infrastructural boundaries really bite. When an ISP wants to wire an existing building, they need a right-of-entry agreement with the building owner — a contract granting permission to access the building, install equipment, and maintain it over time. And these agreements are negotiated building by building.
Building by building. So there's no standardized framework?
There are attempts. The FCC has pushed for MDU access reform. The European Union's Broadband Cost Reduction Directive tries to streamline things. But in practice, every building owner has their own concerns, their own existing contracts, and their own leverage. A building with five hundred high-income tenants is a much more attractive negotiating partner than a twelve-unit walkup.
I assume some building owners see this as a revenue opportunity rather than an infrastructure project.
On one end, you have building owners who treat fiber connectivity as an amenity that increases property values — they actively court ISPs and make access easy. On the other end, you have owners who treat access as a revenue stream. They charge ISPs monthly per-subscriber fees, demand exclusivity payments, or require revenue-sharing on service contracts. Some buildings have exclusive marketing agreements where only one ISP is allowed to serve the building, and that ISP pays for the privilege.
The building becomes a miniature monopoly zone.
A micro-monopoly, yes. And that's a real consumer protection issue. The FCC banned exclusive MDU agreements for video services back in two thousand seven, and there's been ongoing debate about extending similar rules to broadband. But enforcement is spotty, and plenty of buildings still operate under de facto exclusivity because the owner simply refuses to grant access to competing ISPs.
The landlord as a tollbooth operator. Which brings us to something Daniel was hinting at — the difference between getting the building connected and getting an individual apartment connected. Once the fiber is in the basement and the riser is wired, is hooking up unit four-oh-seven still a whole production?
It depends heavily on whether the building was pre-wired during construction. If it was, connecting a new tenant is often a fifteen-minute job. The fiber drop to that unit is already in place — it's been sitting behind a blank wall plate or in a utility closet since the building went up. The technician just needs to connect the drop at the floor distribution hub, install the optical network terminal in the apartment, and activate the service.
And if it wasn't pre-wired?
Then you're looking at a multi-hour installation that involves drilling through walls, fishing cable through conduits, and potentially running surface-mount raceway along baseboards. In a concrete-and-steel high-rise, drilling through walls is not casual. You need permission from building management, you might need to coordinate with structural engineers, and you definitely need to avoid hitting electrical conduits, plumbing, or post-tension cables.
Post-tension cables. Those are the high-tension steel cables running through concrete slabs?
And if you drill through one of those, you've just compromised the structural integrity of the floor slab. The repair cost can run into six figures. So installers working in concrete buildings use ground-penetrating radar to scan before drilling. It adds time and cost to every installation.
In a pre-wired building, the marginal cost of connecting a new tenant is negligible. In a non-pre-wired building, it's substantial. Which creates a weird incentive dynamic — the developer who pre-wires captures none of that future savings, and the tenant who moves in five years later pays the cost of the developer's decision.
That's exactly the split-incentive problem, and it's one of the central challenges in building infrastructure. The developer optimizes for construction cost, not operational cost. Pre-wiring a fifty-story building with fiber drops to every unit might add somewhere between fifty and two hundred dollars per unit to construction costs. That's trivial in the context of a half-million-dollar condo, but developers operating on thin margins still resist it because it's an upfront cost with no direct return at sale.
Fifty to two hundred dollars per unit. That's the "we didn't bother" tax that future tenants pay with every installation.
It gets worse when you consider that retrofitting after the fact costs far more. Running a fiber drop to a single unit in an occupied building can cost anywhere from two hundred to over a thousand dollars, depending on the building's construction and accessibility. The cumulative retrofit cost across all units dramatically exceeds what pre-wiring would have cost.
This feels like the kind of thing that building codes should solve. Just mandate fiber pre-wiring in new construction above a certain size.
Some jurisdictions are doing exactly that. San Francisco has required fiber-ready infrastructure in new multi-unit buildings for years. The European Union's latest building directives push for in-building fiber infrastructure. In Israel, where Daniel is, there's a whole regulatory framework around passive infrastructure access. But it's patchy globally.
Let's talk about Israel specifically, since that's our context. What's the landscape there?
Israel is an interesting case study because the market structure is unusual. Bezeq, the legacy telecom, historically owned the copper network and had a dominant position. Over the past decade, there's been a major fiber push, with multiple players building out infrastructure. The regulatory approach has been to mandate that infrastructure providers offer wholesale access to competitors, which is meant to prevent the micro-monopoly problem.
Does that wholesale access extend inside buildings?
That's where it gets complicated. The wholesale obligation generally covers the network up to the building entry point. Inside the building, the infrastructure is often classified as private property, and access depends on the building owner's cooperation. There's a concept called "passive infrastructure access" that Israel's regulators have been working on — the idea that the physical conduits, risers, and termination points inside a building should be accessible to multiple service providers on fair terms. But implementing that in practice means overcoming the same building-owner dynamics we've been describing.
Even with pro-competition regulation, the building owner remains a gatekeeper.
The wholesale model creates an additional layer of complexity. You've got the infrastructure owner — the company that actually ran the fiber into the building — and then you've got service providers who lease access to that infrastructure. When a tenant in unit four-oh-seven calls to get fiber service, the service provider might need to coordinate with the infrastructure owner to activate the drop, and also with the building management to get access to the floor distribution hub, and also with the tenant to schedule the in-unit installation. Three separate coordination points, each with their own processes and timelines.
That sounds like a recipe for the kind of appointment windows that ruin your entire Tuesday.
"Between eight AM and six PM, someone will be there." And that someone might need to call someone else who needs to call someone else. It's not uncommon for MDU fiber installations to require multiple truck rolls — one to verify the infrastructure is in place, one to do the actual connection, and sometimes a third to fix something that wasn't done right on the second visit.
Multiple truck rolls. That's an industry term?
A truck roll is just a technician visit. And every truck roll costs the ISP money — estimates range from fifty to over two hundred dollars per visit depending on the market. So when the installation process requires multiple visits, the economics of serving a building can flip from profitable to loss-making very quickly.
Which circles back to the original question about whether ISPs even want to serve smaller buildings. If a twelve-unit building requires the same number of truck rolls as a two-hundred-unit building, the per-subscriber economics are brutal.
And this is why you see ISPs prioritizing large MDUs and why smaller buildings often get left behind in fiber rollouts. The fixed costs of building entry, riser work, and equipment installation get amortized across subscribers. More subscribers per building entry point means better economics.
What about the actual physical installation inside the apartment? Once the fiber reaches the unit, what happens?
The fiber terminates at an optical network terminal — the ONT. That's a small powered device, usually about the size of a paperback book, that converts the optical signal to electrical signals for your router. The ONT is the demarcation point between the ISP's network and your home network. Everything upstream of the ONT is the ISP's responsibility. Your router and everything downstream is yours.
The ONT needs power, so it has to be near an outlet.
Yes, and that's a surprisingly annoying constraint. The ideal location for the ONT is near where the fiber enters the unit, which might be in a closet or hallway with no power outlet nearby. Running power to that location adds cost and complexity. Some newer ONT designs support power-over-fiber or can be placed farther from the entry point, but it's still a practical friction point.
Let's zoom out for a second. We've talked about the physical infrastructure and the contractual mess. But there's another boundary Daniel mentioned that I want to hit — the boundary between different network segments and who's responsible for what.
In telecom jargon, this is the demarcation hierarchy. At the top level, you've got the international backbone — the submarine cables crossing oceans. Those are owned by consortia of telecom companies and content providers. Google, Meta, Microsoft, and Amazon are all major investors in submarine cable systems now, alongside traditional carriers.
The cloud providers became infrastructure owners because they generate so much traffic that buying capacity from carriers was more expensive than just laying their own cables.
Then you've got national backbones, typically owned by major telecom companies or sometimes government entities. Then metropolitan networks connecting the backbone to individual neighborhoods. Then the last mile from the neighborhood node to the building. Then the in-building riser. Then the horizontal drop to the unit. Then the in-unit wiring.
Each boundary is a potential point of failure, a potential contractual dispute, and a potential bottleneck.
Each boundary has different ownership, different regulatory treatment, and different economic incentives. The submarine cable consortium has completely different concerns than the building owner deciding whether to grant riser access. They're operating in different universes of regulation, technology, and business models.
It's remarkable that the whole thing works at all.
It works because the internet's architecture is designed to be indifferent to the underlying infrastructure. IP packets don't care whether they're traveling through a submarine cable owned by a Google-led consortium or through a single-mode fiber in a building riser owned by a reluctant landlord. The protocols abstract away the ownership boundaries. But the physical deployment still has to navigate every one of those boundaries.
When something goes wrong, the abstraction breaks down and you're on the phone with your ISP who's blaming the building owner who's blaming the infrastructure provider who's blaming the contractor who drilled through the post-tension cable.
The blame cascade is real. And it's one reason why vertically integrated ISPs — companies that own everything from the backbone to the in-building infrastructure — can offer better troubleshooting experiences. When one company owns the whole chain, there's no finger-pointing. But vertical integration also raises competition concerns, which is why regulators push for wholesale access and infrastructure sharing.
There's a genuine tension between operational simplicity and market competition.
And that tension plays out most acutely at the building level. A building with a single vertically integrated provider gets clean, fast installations and simple support. A building with wholesale-based competition gets lower prices and more choice, but potentially messier installations and more complex support.
Is there a middle ground? Some model that captures the benefits of both?
This is where the concept of open access networks comes in. In an open access model, a single entity owns and operates the passive infrastructure — the fiber, the conduits, the distribution hubs — and multiple service providers compete to deliver services over that infrastructure. The infrastructure owner is neutral; they don't sell retail services, they just provide wholesale access on equal terms to any qualified provider.
Like a utility model for fiber.
Cities like Stockholm, Amsterdam, and Singapore have implemented versions of this. The city or a municipal entity owns the fiber network, and dozens of ISPs compete to offer services over it. The building owner deals with one infrastructure entity, and tenants choose from multiple service providers. It separates the infrastructure problem from the service competition problem.
In a skyscraper context, that would mean one set of riser cables, one set of floor distribution hubs, one set of drop cables to each unit — all owned and maintained by the infrastructure entity — and then tenants pick their ISP from a menu.
And the infrastructure entity has the incentive to pre-wire everything properly because they're amortizing the cost across all potential future subscribers, not just the ones who sign up in the first year. It aligns the long-term infrastructure investment with the operational reality.
Open access networks require either municipal investment or a regulatory framework that makes the economics work for private infrastructure investors. Not every market has that.
Even where the framework exists, the building-level dynamics don't disappear. The open access network still needs right-of-entry agreements with every building owner. It still needs to navigate riser congestion and concrete drilling and the whole physical deployment challenge. The ownership model simplifies the contractual boundaries but doesn't eliminate the physical ones.
We keep coming back to the same point: the inside of the building is a hard problem, regardless of the regulatory or business model wrapped around it.
It's hard because buildings are heterogeneous. Every building has a different floor plan, different construction materials, different riser layouts, different existing infrastructure, different ownership structures, different tenant profiles. There's no standardized solution that works everywhere. Each building is a custom deployment project.
Which is why the fiber-to-the-curb terminology exists in the first place. "To the curb" is the standardized part — the street-level deployment that uses common equipment, common processes, common regulatory frameworks. Everything past the curb is bespoke.
"fiber to the curb" is often a polite way of saying "we got it close to your building but the rest is someone else's problem." In some deployments, fiber-to-the-curb means the fiber terminates at a pedestal on the street, and the final connection to the building uses existing copper or coaxial wiring. That avoids the whole building-access problem entirely, but it also means you don't get fiber speeds inside the building.
It's a compromise that trades performance for deployment simplicity.
And in the early days of fiber rollout, fiber-to-the-curb was a common intermediate step. ISPs would run fiber to neighborhood nodes and then use existing DSL or cable infrastructure for the final drop. It let them market "fiber-powered" services without actually entering buildings. But as consumer expectations have risen and fiber-to-the-home has become the gold standard, fiber-to-the-curb increasingly looks like a half-measure.
"Fiber-powered" is a wonderfully slippery marketing phrase.
The ISP marketing lexicon is a whole separate episode. "Fiber-powered," "fiber-rich," "fiber-advantaged" — none of these mean fiber actually reaches your apartment.
The linguistic equivalent of "contains real fruit flavor.
And consumers have gotten savvier about this. People now ask "is it fiber all the way to my unit?" because they've been burned by the marketing terminology.
Let's circle back to something Daniel raised about commercial incentive. He pointed out that a skyscraper has enough concentrated demand that ISPs want to connect it, whereas a small building might not be financially worthwhile. That's true for the external connection, but does the same logic apply inside the building? Does the ISP care about wiring every unit, or do they wire the building and then take a wait-and-see approach?
It depends on the deployment model. In a pre-wired building, the incremental cost of connecting an additional unit is so low that ISPs will happily connect any unit that signs up. The fiber drop is already there; activating it is cheap. In a non-pre-wired building, ISPs face a decision for each new subscriber: is the installation cost worth the expected revenue?
They might actually refuse to connect a specific unit if the installation looks too expensive?
They might not refuse outright, but they might quote an installation fee that makes the subscriber think twice. If running a fiber drop to unit four-oh-seven requires drilling through concrete and the estimated cost is eight hundred dollars, the ISP might say "we'll do it, but your installation fee is five hundred dollars." That effectively screens out high-cost installations.
Which creates a weird situation where two units in the same building, one floor apart, might have completely different experiences getting fiber service — because one happens to be directly above the riser closet and the other is at the far end of a hallway with no existing conduit path.
That's exactly what happens in practice. I've seen buildings where some units get fiber in fifteen minutes and other units require a full day of drilling and cable fishing. The building's internal geography matters enormously.
What about very tall buildings — fifty, sixty, eighty stories? Are there distance limitations on fiber inside a single building?
Not in any practical sense. Single-mode fiber can carry signals for tens of kilometers without amplification. Even in the tallest buildings in the world, the vertical distance from basement to penthouse is well under a kilometer. The Burj Khalifa is about eight hundred thirty meters tall. That's nothing for single-mode fiber.
The height itself isn't a technical constraint. It's purely about the physical access and the riser pathway.
The technical challenge isn't the distance; it's the number of splices, connectors, and distribution points along the way. Each passive optical component introduces some signal loss. In a properly designed PON network, the optical budget — the allowable loss between the central office and the subscriber's ONT — is typically around twenty-eight to thirty-two decibels. A well-designed in-building distribution system stays well within that budget even for very tall buildings.
The optical budget is the technical constraint, and building height rarely pushes against it.
The constraints are almost entirely physical access, contractual permissions, and economics. The glass itself can handle the distance just fine.
Which is kind of beautiful, in a way. We've built this incredible technology that can send pulses of light through glass strands for kilometers with minimal loss, and the bottleneck is whether a building owner in Manhattan will let a technician into the riser closet on a Tuesday.
The mundane reality of infrastructure deployment. All the physics problems are solved, and what remains is the human coordination problem.
Are there any emerging technologies that might change the in-building deployment equation? Wireless last-mile solutions that bypass the riser entirely?
Fixed wireless access is the main alternative. Companies like Verizon and T-Mobile in the US are using five G millimeter-wave and mid-band spectrum to deliver home internet without any physical building entry at all. The subscriber just plugs in a receiver — essentially a specialized hotspot — and gets service over the air.
You skip the riser, skip the building owner, skip the drilling. Just put a receiver in the window.
In practice, millimeter-wave signals don't penetrate building materials well, especially modern low-emissivity windows that have metallic coatings for energy efficiency. So the receiver often needs to be mounted outside the window or in a location with clear line-of-sight to the tower. And performance varies with weather, distance, and network congestion.
It's a tradeoff. Easier deployment, less consistent performance.
For many users, that tradeoff is perfectly acceptable. If you can get three hundred megabits per second over fixed wireless with no installation hassle, versus a gigabit over fiber that requires drilling and a four-hour appointment window, a lot of people will take the wireless option.
Fixed wireless doesn't eliminate the building-level dynamics entirely. If you're in a dense urban area with lots of tall buildings, getting line-of-sight to a tower from the thirty-seventh floor might be its own challenge.
Yes, and building owners can still create obstacles. Some buildings restrict antenna installations on balconies or exterior walls. Some HOAs and condo boards have rules about visible equipment. The gatekeeper problem doesn't fully disappear; it just shifts from physical access to aesthetic regulation.
The building owner always finds a way to be involved.
They own the box everyone lives in. They're involved by default.
Alright, let's try to synthesize this. If someone is moving into a high-rise apartment and wants fiber internet, what should they actually look for? What questions should they ask?
First question: is the building pre-wired for fiber? If yes, ask which ISPs have infrastructure in the building. Second question: if it's not pre-wired, what's the installation process? Ask the ISP and the building management separately, because you'll get different answers, and the gap between those answers tells you a lot about how painful the process will be.
The gap between the ISP's answer and the building management's answer as a pain index. I like that.
Third question: where does the fiber enter the unit? If it's in a closet with no power outlet, you're going to have an ONT placement problem. Fourth question: is there an exclusive agreement between the building and a particular ISP? If yes, you have no choice of provider, and you should factor that into your decision about living there.
If you're buying rather than renting, these questions become even more important because you're stuck with the building's infrastructure decisions for the long term.
For condo buyers, I'd add: ask to see the building's telecommunications infrastructure during the viewing. Look for the equipment room, look for floor distribution points, ask about the riser pathways. Most buyers never think to do this, but it directly affects your quality of life for years.
The telecom walkthrough as part of the home inspection. That's a practical takeaway.
It's the kind of thing that seems obsessive until you're the one spending four hours waiting for a technician who can't finish the installation because the conduit is blocked and they need to reschedule and now you're taking another day off work.
Infrastructure is invisible until it fails. Then it's the only thing that matters.
That's the thesis of this entire episode, really.
One more thing I want to touch on. Daniel mentioned the hub-and-spoke model from a previous discussion about smart hotels. Is there an analogy worth drawing between structured cabling in hotels and fiber deployment in residential towers?
There is, and the parallel is instructive. In a smart hotel deployment, you typically have a central network core, floor-level gateways that handle local processing, and then room-level controllers that manage individual devices. The architecture is hierarchical and distributed for the same reasons that fiber distribution in a residential tower is hierarchical and distributed: you want to minimize the amount of cabling, you want to isolate failures, and you want to make the system manageable at scale.
In both cases, the floor-level distribution point is the unsung hero that makes the whole thing work.
The floor distribution hub in a fiber network is doing conceptually the same job as the floor gateway in a hotel automation system. It's the point where the vertical spine branches into horizontal distribution. It's the boundary between "building-scale infrastructure" and "unit-scale connectivity.
The difference being that the hotel system has active electronics on each floor, while the fiber distribution hub is passive.
And that passivity is a huge advantage from a maintenance perspective. No software updates, no power supplies, no thermal management. Just glass and plastic doing their job for decades.
That's the other thing about fiber that doesn't get enough attention. Once it's in place, it has extraordinary longevity. The fiber installed today will support speeds we haven't even commercialized yet.
Because the capacity limit is in the electronics at either end, not in the glass itself. The same single-mode fiber that was installed twenty years ago for one-gigabit service can carry one hundred gigabits today with upgraded transceivers. The glass doesn't need to be replaced; the boxes at either end do.
Which makes the upfront deployment decision even more consequential. If a building isn't pre-wired for fiber, it's not just missing out on today's speeds — it's missing out on the next several decades of speed improvements, unless someone pays for an expensive retrofit.
That's why building codes and municipal policies that mandate fiber-ready infrastructure in new construction are so important. They're not about meeting today's needs; they're about not foreclosing tomorrow's possibilities.
The developer who saves a hundred dollars per unit by skipping fiber pre-wiring is making a decision that will cost the building's residents thousands of dollars and countless hours of inconvenience over the building's lifetime.
Split incentives, again. The developer captures the savings; the residents bear the costs.
Capitalism's greatest hits.
The album never stops selling.
To directly answer the question Daniel posed: in mega-buildings, ISPs absolutely bring fiber to specific floors and then branch out to individual apartments. It's not just common — it's the standard architecture. And whether it's straightforward or not depends almost entirely on decisions made during construction. Pre-wired buildings make it trivial. Everything else makes it a custom project with all the coordination headaches we've described.
The contractual boundaries are layered like an onion. International backbone consortiums, national carriers, metropolitan networks, last-mile providers, building entry agreements, riser access rights, in-unit installation. Each layer has its own owners, its own incentives, and its own failure modes.
The internet is a triumph of engineering and a disaster of property rights.
That might be the most concise summary of this entire topic.
And now: Hilbert's daily fun fact.
Hilbert: In the early twentieth century, linguists studying Inuktitut in the Arctic widely believed the language exhibited polysynthetic morphology so extreme that a single word could express what required an entire sentence in English — a theory that dominated linguistic typology for decades before researchers realized they had been misanalyzing the data by confusing morphological incorporation with syntactic combination. The theory was largely abandoned by the nineteen seventies but remains a cautionary tale in the field about the dangers of exoticizing unfamiliar language structures.
And now: Hilbert's daily fun fact.
Hilbert: In the early twentieth century, linguists studying Inuktitut in the Arctic widely believed the language exhibited polysynthetic morphology so extreme that a single word could express what required an entire sentence in English — a theory that dominated linguistic typology for decades before researchers realized they had been misanalyzing the data by confusing morphological incorporation with syntactic combination. The theory was largely abandoned by the nineteen seventies but remains a cautionary tale in the field about the dangers of exoticizing unfamiliar language structures.
Linguists got overexcited and invented a grammatical superpower that didn't exist.
The academic equivalent of seeing a magic trick and spending forty years building a theory around it before someone asks to see the other hand.
This has been My Weird Prompts. Our producer is Hilbert Flumingtop. If you enjoyed this episode, leave us a review wherever you listen — it helps other people find the show.
This has been My Weird Prompts. Our producer is Hilbert Flumingtop. If you enjoyed this episode, leave us a review wherever you listen — it helps other people find the show.
We'll be back soon.
This episode was generated with AI assistance. Hosts Herman and Corn are AI personalities.