Daniel sent us this one — he's looking at the tension between infrastructure as a natural monopoly and the consumer need for competitive retail markets. Specifically, he's pointing to Israel's model of compelling infrastructure operators to open their networks to virtual operators at cost, and he's asking how that framework can be applied to energy, water, and other networks. The core question: how do regulators force a network owner to provide genuinely useful connectivity to competitors on fair terms, when every fiber of that owner's being wants to maintain a monopoly?
I see what you did there.
That image of a street being dug up for the third time in a year is the perfect entry point into the core tension of infrastructure economics. You've got a natural monopoly — high fixed costs, enormous barriers to entry, and very low marginal cost to add one more user once the thing is in the ground. And yet we want competition on the retail side. Those two things pull in opposite directions.
Because the obvious free-market answer — let three companies each lay their own fiber — is insane.
It's not just wasteful, it's actively destructive. Every redundant trench tears up roads, disrupts traffic, and costs money that ultimately comes from the same pool of consumer spending. There's a reason economists call redundant infrastructure deadweight loss.
The question becomes: how do you separate the pipe from what flows through it?
And the legal foundation for this goes back decades. In US antitrust law, there's something called the essential facilities doctrine, established most clearly in MCI versus AT&T in nineteen eighty-three. The idea is that if you control a facility that's essential for competitors to operate — and duplicating it is impractical or impossible — you can be compelled to grant access on reasonable terms.
Which sounds almost socialist until you realize it was born in the Reagan era, in the context of breaking up the Bell System.
This isn't some fringe regulatory theory. It's mainstream competition law. The essential facilities doctrine says: you can own the bridge, but you have to let other people drive across it at a fair toll.
Let's look at a country that actually tried to solve this — Israel, starting in twenty eighteen. What actually happened?
In twenty eighteen, Israel's Ministry of Communications did something structural. They mandated that Bezeq — the incumbent telecom — and IBC, which was the new fiber entrant backed by the Israel Electric Corporation and a Swedish fund, had to lease dark fiber and physical ducts to any licensed virtual operator. And this wasn't a suggestion. These were regulated, cost-based wholesale rates.
IBC is interesting here because it wasn't the legacy player. It was the challenger that had to share too.
The regulation applied symmetrically. Any company that owned passive infrastructure — ducts, poles, dark fiber — had to offer access. The point wasn't to punish Bezeq specifically. It was to prevent any infrastructure owner from leveraging their physical position into a retail monopoly.
Let's talk about what passive infrastructure actually means here, because the distinction matters.
Passive infrastructure is the stuff that just sits there. The ducts under the street. The dark fiber — which is fiber optic cable that's been laid but has no transmission equipment attached to it yet. It's the glass in the ground, not the lasers at either end. Active infrastructure is the electronics — the switches, routers, the stuff that actually lights up the fiber and sends data.
The separation between these two is what makes the whole model work.
That's the key insight. Israel forced what's called functional separation. Bezeq had to create a separate subsidiary — Bezeq Infrastructure — that manages all the passive stuff and must offer access to every licensed ISP at uniform, cost-oriented prices. The retail arm of Bezeq has to buy access from Bezeq Infrastructure on exactly the same terms as 012 Smile or Partner or Cellcom.
The retail arm can't get a sweetheart deal from its own infrastructure sibling.
And this is where the regulator's teeth come in. The pricing mechanism they use is called LRIC plus — Long-Run Incremental Cost plus a reasonable markup. The idea is you calculate what it would cost an efficient operator to build and maintain the network over its lifetime, add a return on capital that's high enough to incentivize continued investment, and that's your wholesale price.
As of twenty twenty-five, wholesale dark fiber access in Tel Aviv is running somewhere around forty-five to sixty shekels per meter per year, depending on density.
Which is transparent and predictable. Any virtual operator can look at that rate sheet and build a business model around it. They don't have to negotiate individually with Bezeq and hope they get a fair deal. The price is the price.
What's the result been in terms of actual consumer pricing?
Israel's retail fiber prices dropped roughly forty percent between twenty nineteen and twenty twenty-five. And that's not just introductory teaser rates. That's sustained price compression driven by genuine competition. You've got companies like 012 Smile, Partner, Cellcom, and others all competing on price, customer service, value-added services — static IPs, bundled TV packages, business-grade SLAs — and none of them had to dig up a single street.
Forty percent is the kind of number that makes incumbent shareholders weep and regulators frame on their wall.
Here's the thing that critics always raise: doesn't this kill the incentive to invest? If you have to share your network at cost-plus, why would you bother building it in the first place?
The answer is in the plus.
The LRIC plus model includes a capital cost allowance. The regulated rate of return is designed to be attractive enough to fund upgrades. And the evidence bears this out. Israel's fiber coverage went from about thirty percent of households to eighty-five percent after the sharing mandate was implemented. New Zealand's Chorus invested two and a half billion New Zealand dollars in fiber upgrades between twenty eighteen and twenty twenty-five while operating under open access rules. The investment didn't dry up.
Because the infrastructure owner now has a guaranteed customer base. Every virtual operator is a revenue stream. You're not competing for retail customers — you're collecting wholesale fees from everyone.
Those wholesale revenues are more predictable than retail revenues, which is something investors actually like.
Now there's a complication here that most people don't think about until they're standing in their apartment building staring at a locked telecom closet. What happens when the building's internal wiring is already owned by the first ISP that got there?
This is the last-mile-inside-the-building problem. In a lot of countries, the first ISP to wire an apartment building effectively owns the in-building infrastructure, and they have no obligation to let competitors use it. So you get this absurd situation where the street-level sharing works beautifully, but the final fifty meters inside the building are still monopolized.
Israel's solution?
A building access regulation. Property owners are required to allow any licensed ISP to install their own in-building wiring at no cost to the building. The ISP bears the installation cost. The building owner can't say no, and they can't charge a fee.
Which means if you're in a Tel Aviv apartment and you want to switch from Bezeq to Partner, Partner can run their own fiber from the street into your unit without the landlord blocking it.
They don't have to rewire the whole building. They just need a path to the specific units that are switching. Over time, you end up with multiple ISPs' fibers running through the building risers, and switching becomes a software change.
Let's talk about a concrete example that shows what happens without this kind of regulation. Tel Aviv's Rothschild Boulevard.
Oh, this one hurts. Between twenty fifteen and twenty twenty, three separate fiber ducts were laid under Rothschild Boulevard — one by Bezeq, one by IBC, one by HOT. Each trench cost roughly two hundred thousand dollars per kilometer. So that's six hundred thousand dollars per kilometer to put three sets of glass in the ground that could have been one shared duct with three times the fiber strands inside it.
Every time they dug it up, the boulevard was a construction zone. Restaurants lost business. Traffic was a nightmare. The city had to repave each time.
Under the sharing mandate, one duct would have been sufficient. Any ISP could lease dark fiber strands inside that shared duct. The infrastructure cost gets amortized across multiple retail providers, and the street gets dug up once.
The triple-dig is the perfect metaphor for everything wrong with unregulated infrastructure competition.
There was also a really instructive dispute in twenty twenty-four between Bezeq Infrastructure and 012 Smile over wholesale pricing for ten gigabit per second symmetric connections. Bezeq wanted to charge a premium for higher-speed access, arguing that ten gig symmetric required different handling at the infrastructure level. 012 Smile pushed back, saying the dark fiber is the same dark fiber regardless of what speed the retail ISP lights it up at.
Which is technically true.
Dark fiber doesn't care about speed. The speed is determined by the electronics you attach to either end. So Bezeq's argument was essentially: we want to charge more because the retail product is better, even though our costs haven't changed.
Which is exactly the kind of behavior that reveals the monopoly instinct lurking under the surface.
The regulator resolved it by imposing a temporary fifteen percent price cut on the wholesale rate and clarifying that dark fiber pricing must be based on the physical characteristics of the fiber — length, strand count, duct space — not on what the retail ISP plans to do with it.
The Israeli model works well for fiber. But the prompt asks how this applies to other infrastructure networks. Let's start with energy.
The energy sector has actually been doing a version of this for longer than telecom. The UK's RIIO model — which stands for Revenue equals Incentives plus Innovation plus Outputs — already separates the wire business from the supply business. The distribution network operators own the poles and wires. The retail suppliers buy access to those networks and sell electricity to consumers.
There's a complication in energy that fiber doesn't have. Electrons don't have lanes.
This is the big one. Fiber is fundamentally a point-to-point medium. You send a signal from A to B, and it doesn't physically interfere with the signal from C to D on an adjacent strand. Electricity doesn't work that way. The grid is a mesh, and power flows according to the laws of physics, not according to contractual paths. Congestion is a real physical phenomenon — if too much power tries to flow through a particular transmission line, it overheats and fails.
The wholesale access model has to account for physical flow constraints in a way that fiber wholesale doesn't.
That's why energy markets have things like locational marginal pricing, where the wholesale price of electricity varies by node on the grid depending on congestion. It's also why energy regulators spend a lot of time on capacity markets and congestion management. The structural separation principle is the same — own the wires or sell the power, not both — but the implementation is considerably more complex.
Does the UK model actually work in practice?
It does, and the most vivid proof is what happened when Bulb Energy collapsed in twenty twenty-one. Bulb had one point seven million customers. Under the UK's supplier of last resort mechanism, Octopus Energy took over all of them. No physical assets had to be transferred. No wires changed hands. The distribution networks were already separated. It was a purely financial and administrative transition.
One point seven million customers switched providers without a single truck roll.
That's the power of structural separation. The infrastructure is a platform. The retailers are apps running on that platform. When an app fails, users move to a different app. The platform doesn't care.
What about water? The prompt mentioned that too.
Water is fascinating because it combines the physical flow constraints of energy with an even more politically charged dimension. Nobody riots over broadband prices. They do riot over water.
Water infrastructure is typically even more monopolistic than energy or telecom, because the idea of competing water pipes to every house is absurd on its face.
Chile is the interesting case study here. In twenty twenty-one, Chile reformed its water code to mandate open access to desalination plants' distribution networks. The idea was that desalination capacity was becoming the bottleneck in arid regions, and the plant owners were extracting monopoly rents.
They applied the essential facilities doctrine to desalinated water.
Early results from the Antofagasta region show about a twelve percent drop in industrial water prices. But there's been ongoing conflict over maintenance cost allocation. If multiple parties are pulling water through the same distribution network, who pays for the pump replacement when it wears out? How do you allocate costs proportionally when different users have different flow rates and pressure requirements?
That seems solvable with metering, though.
In principle, yes. In practice, water metering at the wholesale level is less precise than electricity metering, and the disputes have been genuine. Chile's water regulator has spent the last few years essentially building the administrative infrastructure to adjudicate these allocation questions.
The model transfers, but the implementation complexity varies by sector. Fiber is the cleanest case. Energy adds congestion. Water adds both congestion and measurement challenges.
There's a lesson here about sequencing. You don't try to solve all of it on day one. You establish the structural separation, set cost-based wholesale pricing, create a dispute resolution mechanism, and then iterate as problems emerge.
Let's talk about a scenario where the infrastructure owner tries to sabotage the model from within. Specifically, what prevents them from degrading service quality for virtual operators' customers?
This is the repair ticket problem. You're a virtual operator, one of your customers reports an outage, you file a ticket with the infrastructure owner, and that ticket goes to the bottom of the queue while the infrastructure owner's own retail customers get priority.
Which is exactly what happened in New Zealand.
In twenty twenty-three, 2degrees alleged that Chorus — which is the infrastructure spinoff from Telecom New Zealand — was prioritizing its own retail arm's repair tickets. The Commerce Commission investigated and imposed what they called a functional separation audit regime with quarterly reporting. Chorus now has to publicly report on repair times, broken down by whether the customer belongs to their own retail arm or to a competitor.
The mere existence of that reporting requirement changes behavior.
Nobody wants to be the executive explaining to a regulator why competitor tickets take twice as long as internal tickets. The transparency is the enforcement mechanism.
New Zealand is actually the other poster child for this model, alongside Israel. What does their setup look like?
New Zealand took an even more radical approach than Israel. They didn't just mandate sharing — they structurally separated Telecom New Zealand into two completely independent companies. Chorus owns all the passive infrastructure and operates as a regulated monopoly wholesaler. Spark is the retail company. They have no common ownership. And as of early twenty twenty-six, Chorus supports over a hundred and fifty retail ISPs on its network.
A hundred and fifty. That's not a market with three competitors fighting over scraps. That's genuine retail competition.
The results speak for themselves. New Zealand's broadband prices dropped thirty-five percent from twenty eighteen to twenty twenty-five, while average speeds tripled. You had more competition and more investment simultaneously — which is exactly what the skeptics said couldn't happen.
Let's bring in the US context, because the contrast is instructive.
The US is a patchwork. The FCC's twenty twenty-four Digital Discrimination rules include language about network access on reasonable terms, but they lack the structural separation that makes the Israeli and New Zealand models work. Without structural separation, you're essentially asking the infrastructure owner to please be nice to competitors, and hoping the threat of enforcement keeps them honest.
Which it won't.
It hasn't. The BEAD program — Broadband Equity Access and Deployment — is funding publicly owned middle-mile networks in fourteen states. That's a different approach: rather than forcing private owners to share, you build public infrastructure and make it open access by default.
The last mile is still the problem.
The middle mile is the easier part because it's essentially long-haul fiber between population centers. The last mile is the expensive part — getting from the neighborhood node to every individual home. And in most of the US, that last mile is still controlled by a handful of incumbent ISPs with no obligation to share.
Fourteen states have publicly owned middle mile, and zero states have mandated last-mile sharing.
That's the gap. And it's why the US has some of the highest broadband prices in the developed world despite being the birthplace of the internet.
There's another dimension here that's specific to legacy infrastructure. Bezeq tried to sunset its copper network in twenty twenty-three.
This is the copper retirement problem, and it's a critical safeguard. Bezeq wanted to shut down its copper network to force customers onto its fiber, where it had — at the time — less regulated wholesale obligations. The regulator blocked it, ruling that copper had to remain operational until fiber wholesale access was fully functional and available to all virtual operators.
Because otherwise the infrastructure owner can use the transition to new technology as a window to re-establish monopoly control.
You upgrade from copper to fiber, and for a brief moment, the old sharing rules don't clearly apply to the new infrastructure. If the regulator isn't paying attention, you can lock competitors out during that transition window and emerge with a fiber monopoly.
The regulator has to be faster than the upgrade cycle.
Which is hard. Regulators are inherently slower than the companies they regulate. But the Israeli case shows it's possible if you establish the principle upfront: any new infrastructure inherits the sharing obligations of the old infrastructure. No regulatory holidays.
Let's talk about the European dimension, since there's a major regulatory shift happening right now.
The EU's Gigabit Infrastructure Act entered its enforcement phase in the second quarter of twenty twenty-six. It mandates duct and pole access across all member states — essentially requiring every EU country to implement a version of what Israel and New Zealand have been doing. The act also streamlines permit processes and requires new buildings to be pre-wired with fiber and open-access in-building infrastructure.
The EU is essentially codifying the Israeli model at continental scale.
With some important variations. The EU approach is less prescriptive about pricing methodology than Israel's LRIC plus model. Member states have more discretion over how they calculate cost-based access rates. That creates a risk of regulatory arbitrage, where infrastructure owners lobby for favorable pricing models in individual countries.
The principle of mandated access is now EU law.
And that's a big deal. Twenty-seven countries, four hundred and fifty million people, all moving toward a model where you can't own the duct and refuse to let competitors use it.
What about the investment objection at EU scale? The argument that forcing sharing will cause infrastructure owners to underinvest?
The evidence from the early adopters doesn't support that. New Zealand's Chorus invested two and a half billion New Zealand dollars over seven years under open access. Israel's fiber coverage exploded after the sharing mandate. The investment doesn't disappear — it just shifts from being funded by retail monopoly rents to being funded by wholesale access fees from multiple competing retailers.
Which is actually a more diversified and stable revenue base.
Instead of one hundred percent of your revenue coming from your own retail operation, you've got revenue from every ISP in the market. If one of them stumbles, your infrastructure revenue doesn't crater.
Let's shift to something the prompt touches on implicitly: what does this mean for consumers who want to actually exercise choice?
If you're in a market with mandated infrastructure sharing — Israel, New Zealand, parts of the EU — switching retail providers should require no physical installation. It's a software change. The provider updates which VLAN your connection is mapped to, and your traffic routes through their network instead of the previous provider's.
If your building doesn't have open access wiring, you should push your landlord to demand it.
Or push your municipal government. Some cities are now requiring open access in-building wiring as a condition of construction permits for new buildings and major renovations. If you're in a condo association or a tenant group, this is something you can actually organize around.
There's also an investor angle here that's worth flagging. Infrastructure sharing reduces the risk of stranded assets.
A single shared fiber network has a more predictable revenue stream than competing networks. If you're an investor looking at Israeli infrastructure companies, Bezeq Infrastructure and IBC trade at lower betas than their integrated peers in markets without sharing mandates. The regulatory framework provides revenue certainty.
Lower beta, more predictable cash flows, less risk of a competitor overbuilding you.
It's counterintuitive if you're used to thinking of monopoly control as the safest investment. But monopolies attract regulation, and regulation can be unpredictable. A regulated wholesale monopoly with guaranteed access fees is actually more stable than an unregulated monopoly that's constantly fighting off competitors and regulators.
What does this mean for someone who actually has to build or regulate one of these systems? What are the actionable pillars?
First, structural or functional separation of infrastructure from retail. You can't have the same entity owning the pipes and selling the service, because the incentive to discriminate is too strong. Functional separation — separate subsidiaries with separate management and audited arms-length dealing — is the minimum. Structural separation — entirely separate companies — is better.
Cost-based wholesale pricing with a transparent LRIC model. The formula has to be public, auditable, and include a capital cost allowance that funds continued investment. If the wholesale price is too low, the infrastructure degrades. If it's too high, competition doesn't materialize. Getting this number right is the single most important design parameter.
The third pillar.
A regulator with teeth. Specifically, the power to audit service quality, impose price caps when the formula breaks down, mandate network upgrades, and block anti-competitive behavior like premature copper retirement. Without enforcement capability, the first two pillars are just suggestions.
A suggestion-based regulatory framework. The libertarian compromise that satisfies no one.
It's worse than nothing, because it creates the appearance of regulation without the substance, which makes it harder to build political momentum for real reform.
Even with all these tools in place, there's a looming question that keeps regulators up at night.
AI-driven bandwidth demand. Data center traffic is projected to grow at about twenty-five percent compound annual growth rate through twenty thirty. That's not incremental. That's a step change in infrastructure requirements. And the question is whether the LRIC plus wholesale pricing model can generate enough revenue to fund the massive investment that kind of growth requires.
Because LRIC plus is designed for steady-state infrastructure growth, not exponential demand curves.
The model assumes you're adding capacity at a predictable rate. If AI training and inference suddenly require ten times the fiber capacity in certain corridors, the wholesale pricing formula might not generate enough revenue fast enough to fund that buildout.
Which leads to the two-tier question. Do we end up with a system where premium customers pay for dedicated fiber, while everyone else shares the regulated wholesale network?
It's already happening in a limited way. Some large financial institutions lease dedicated dark fiber for ultra-low-latency trading connections. They're not on the shared wholesale network. They've essentially built their own parallel infrastructure for specific high-value routes.
If that model expands, you risk undermining the whole premise of infrastructure sharing. The highest-value customers peel off onto private networks, leaving the shared network with a less profitable customer base, which makes the LRIC math harder to sustain.
That's the tension. Infrastructure sharing works best when everyone is on the shared network. The more exceptions you create, the more the economics of the shared network deteriorate.
The next frontier here is wireless. Can the same model work for five G small cells and radio access networks?
The O-RAN Alliance published specifications in twenty twenty-five that include APIs for shared radio access network infrastructure. The idea is that multiple mobile virtual network operators could share the same physical radios and antennas, with software-defined slicing to allocate capacity.
Commercial adoption so far?
The specifications exist on paper, but no major operator has deployed a fully shared RAN in production. The technical challenges are significant — interference management, quality of service guarantees, real-time resource allocation — and the incumbent equipment vendors have every incentive to slow-walk the standardization process.
Because they make more money selling dedicated hardware to each operator.
Of course they do. A shared infrastructure model for wireless would be the equivalent of what happened in fiber, and the Nokias and Ericssons of the world are not eager to see their hardware sales divided by the number of operators sharing each cell site.
The fiber model is proven and scaling. Energy is more complex but operational in multiple countries. Water is nascent but promising. And wireless is still a regulatory aspiration with no commercial reality.
That's the landscape. And the through-line in every sector is the same: you need structural separation, transparent cost-based pricing, and a regulator willing to enforce the rules when the infrastructure owner inevitably tests the boundaries.
The prompt asks how regulators ensure fair terms for virtual operators. The answer, across every case study we've looked at, is that they don't ensure it through goodwill. They ensure it through architecture. You design the system so that discrimination is hard, visible, and punished.
You keep designing. The infrastructure owner will always find new ways to advantage its own retail arm. The regulator's job is to stay close enough to the operational details to spot those moves and close the gaps before they become structural advantages.
Which requires a regulatory agency that's technically sophisticated, adequately funded, and politically insulated enough to make unpopular decisions.
That last part is the hardest. Infrastructure owners are usually large, politically connected, and skilled at framing their interests as the national interest. The regulator needs enough independence to say no to a company that employs thousands of people and sponsors the local football team.
Now: Hilbert's daily fun fact.
Hilbert: In eighteen thirteen, a French naturalist in the Solomon Islands discovered that fossilized tree resin from the region's ancient kauri forests produces a distinct acoustic ringing when struck — a property caused by the resin's unusually high degree of molecular cross-linking, which essentially turned the fossil into a natural glass that resonates at frequencies inaudible to the human ear but detectable by bats.
Bats can hear fossils.
That's a sentence I didn't expect to process today.
This has been My Weird Prompts. I'm Corn.
I'm Herman Poppleberry. If you want more episodes like this one, find us at myweirdprompts dot com or search for My Weird Prompts wherever you listen. Rate us if you feel like it.
Until next time.
