Here's what Daniel wrote in this week. He starts from the Israeli telecom deregulation, the Kahlon reforms, where the state forced the incumbents to host virtual operators and made porting easy, and mobile prices fell sixty to eighty percent. And then he asks the obvious follow-up. If that worked, why not go further?
Further how?
He says, and I'm paraphrasing, if infrastructure is a public good, then the collective interest is in minimising the number of physical networks. Fewer antennas broadcasting the same signal, easier coordination with other utilities, less street disruption when things get dug up. And he pushes it to the limit. Maybe there should just be one network. With redundancy carved out only for government and critical users.
One network.
One. And the mechanism he proposes is interesting. Rather than grafting competition onto private owners by forcing them to share, the government provisions or acquires the network and becomes the sole operator. He names the downside himself. Privacy advocates would bristle at the government controlling the physical layer.
And the upside?
The government controls the physical layer. Streamlined coordination, a genuine level playing field where every retail provider is a virtual operator, and nobody can find imaginative ways to degrade the connection for a competitor. He also flags the obvious problem. The state would need to become the repository of talent for maintaining the network at the physical layer. And he admits privatisation has been the drumbeat for lower prices for so long that this would be a very hard political sell.
So the question is whether there are real examples.
Two questions, actually. Are there models where the state is both operator and seller, in telecoms, electricity, anything with monopolisation potential. And second, if you decouple physical maintenance from the commercial side, do you get the best of both worlds. The stability of a government operator plus the competition of the private market.
Let's define the model precisely first, because the version Daniel describes, state as both operator and seller, doesn't quite exist anywhere.
Nowhere?
Nowhere I could find. Every real national example decouples those two roles. The state owns the passive layer, the ducts and the fibre and the central offices, and retail is fully private and competitive on top of it. What exists is something adjacent. And honestly more interesting.
So there are two models on the table. The Israeli one, and this centralized one.
Right. The Israeli model is regulated wholesale access plus number portability, with private vertically integrated incumbents forced to share. The incumbent still owns the network and still retails to consumers. The regulator stands in the middle and polices the wholesale terms forever.
And the centralized model?
A single passive infrastructure owner, state or state-backed, that does not retail at all. Every retail provider is a virtual operator on equal terms.
So the distinction that matters is wholesale-only versus state-as-retailer.
It's the whole episode, really. Because the wholesale-only version delivers the anti-discrimination benefit Daniel wants without the state-retail-seller downside he's worried about. But it's not the model he described. And that gap matters.
Give me the strongest real case.
Liechtenstein. Forty thousand inhabitants. The state-owned power company, LKW, owns and operates the sole passive fibre network. Wholesale-only. It serves more than ten retail broadband operators. Symmetric eight gigabit at thirty-nine ninety a month in Swiss francs.
Ten operators for forty thousand people.
And the factsheet is almost smug about it. It says Liechtenstein is internationally unique, probably the only country where vertical separation via a passive open access network truly works and has created an exceptionally competitive market.
So let's go back to Israel first, because that's the baseline Daniel's working from. How did it actually work?
Kahlon came in as communications minister in two thousand nine. The big three, Pelephone, Cellcom, Partner, held ninety-five percent of subscribers the year before. The reforms forced them to lease their networks to mobile virtual network operators at regulated rates. They decoupled handset payments from service contracts, so the price of the phone stopped hiding the price of the service. They cut mobile termination rates, which is the fee one network charges another to finish a call, and that had been quietly inflating everything. And they structured the spectrum tenders so the new entrants, Golan and Hot Mobile, got rebates for hitting market-share and coverage thresholds.
Rebates for hitting thresholds.
The state was paying them to succeed. Deliberately.
What happened to prices?
Typical plans fell sixty to eighty percent by twenty eleven. Plans that had been a hundred and fifty to two hundred and fifty shekels a month went to twenty-nine to ninety-nine. Consumers saved an estimated six billion Canadian dollars within three years.
Six billion.
Industry revenue excluding equipment fell from twelve point three billion shekels in twenty eleven to ten point four billion in twenty twelve. That's a one point eight billion shekel collective saving. The incumbents' EBITDA margins dropped about ten percentage points.
So the money came out of the operators' pockets.
Mostly. Golan and Hot Mobile together captured twenty percent of the market, two million subscribers. And the government gave up about one point four billion shekels in licence fees. So the state paid too.
Rami Levi, right? The supermarket chain?
First MVNO in twenty twelve. Twenty agorot a minute and fourteen ninety a month line rental. Pelephone broke ranks in December twenty ten and agreed to host them, which is the moment the dam cracked.
So the mechanism of the Israeli model is competition grafted onto private infrastructure by regulatory force.
Force. The incumbent still owns the physical network and still retails. The regulator has to constantly police the wholesale terms to stop degradation.
Which is the problem Daniel names. Imaginative ways to degrade connectivity.
It's real. It's not paranoia. If you own the network and you also sell to consumers, you have every incentive to make your own service a little better than the one you're forced to wholesale. Not by breaking the rules. By finding the edge of them. Timing, provisioning queues, maintenance windows.
The slowness that's nobody's fault.
There's never a memo. There's just a queue that moves slower for some customers than others.
So the wholesale-only model solves that structurally.
This is the strongest theoretical case for the centralized version, and Liechtenstein's own factsheet makes it explicitly. LKW does not offer any active downstream services, and therefore has no incentive to discriminate. Very much unlike a regulated, vertically integrated incumbent, who has to provide passive infrastructure to downstream competitors as well as to its own retail arm.
Read that again. No incentive to discriminate.
It's not that they're forbidden. It's that there's nothing to gain. They sell one product to everybody at the same price. The discrimination problem doesn't get regulated away. It stops existing.
That's a cleaner answer than anything the Israeli model can offer.
And it's why I keep coming back to the electricity-grid synergy, because that's the part Daniel asked about that nobody talks about. LKW is a power company. It lays fibre because it already has the ducts, it already has the civil-engineering crews, and it's used to investment cycles of up to a hundred years. The factsheet says it expects utility returns. Not venture returns. Utility returns.
Easier coordination with other utilities. Minimal street disruption.
That's Daniel's argument, and it's not theoretical. It's the actual reason the network got built. When you're already digging for the grid, adding a fibre duct to the trench is marginal cost. When a telecom company digs on its own, it's the whole cost.
Seventeen optical distribution offices serving about twenty-six thousand premises.
National coverage including the alpine areas. Copper and cable TV networks mandatorily decommissioned within twelve months of fibre connection. No subsidies, no digital divide. LKW won the European Broadband Awards in twenty twenty-two.
Mandatory decommissioning is the part that surprises me. They didn't run parallel networks.
They refused to. Which is the minimum-networks argument Daniel makes, executed literally. One physical network, and the old ones get switched off.
So why isn't everybody doing this?
Here's where I have to complicate the story, because the Israeli model wasn't the clean market success it gets sold as. YouPhone's CEO, Ariel Schreiber, argued in twenty thirteen that the cellular reform had ended in failure. His claim was that the ministry retrospectively changed the tender terms to advantage Golan and Hot Mobile.
Retrospectively.
After the fact. The state picked winners. Which means the deregulation win required heavy state coercion of private incumbents, not pure market forces. The state didn't step back and let competition happen. It shoved.
So the deregulation model and the centralized model have more in common than either side would admit.
Both are the state deciding the shape of the market. One does it through the regulator's pen, the other through ownership. That's the honest framing.
Let's test the scale question, because every example you've given me is small.
Every strong one. Liechtenstein is forty thousand. Singapore is six million but it's a city-state with a single planning authority. Ireland, Utah municipalities. The open question is whether this survives contact with a large, politically contested market.
And the live test case?
Italy. FiberCop, twenty twenty-four. A wholesale open-access network carved out of Telecom Italia's fixed network, owned by a KKR-led investor group that includes the Italian state as a shareholder. The state frames itself as maintaining a role as guarantor in a sector intended to be protected as strategic for Italy.
Guarantor. That's a careful word.
It's a hybrid. State as minority shareholder, not sole owner. And there's a second Italian model running alongside it. Open Fiber, founded twenty seventeen, sixty percent owned by the state bank CDP Equity, forty percent Macquarie, building a wholesale-only ultra-fast network.
Two competing wholesale networks in one country.
Which is itself an admission that they're not sure which version works.
What about the national-scale examples that actually function?
Singapore's NetLink Trust is the cleanest. The government mandated structural separation. NetLink designs, builds, owns and operates the passive fibre infrastructure of the nationwide broadband network, and provides open wholesale access so retail telcos can focus on offering innovative products without incurring high fixed costs. It's a listed trust, and its returns are set by a five-yearly regulatory review.
Listed. So it has shareholders.
It does. And the regulator caps what it can earn. That's the trade. You get a stable, boring, predictable return, and in exchange you don't get to extract monopoly rent.
Ireland?
SIRO, an ESB and Vodafone joint venture launched twenty fifteen. Open-access, hundred percent fibre to the premises, nineteen operators reselling it nationwide. Sweden had municipalities doing this in the late nineties, among the first to deploy fibre to the home this way. And Carinthia in Austria is my favourite, because the state deliberately limited itself to the role of infrastructure owner and leases the network as an open network to all operators under equal conditions. The stated reason is that it did not want to mutate into a telecommunications operator.
It didn't want to mutate.
That's the quote. A regional government explicitly saying we will own the pipes and we will not become a phone company.
Which is the exact line Daniel's model crosses and this one doesn't.
And it's the line that matters politically. Owning pipes is a public works argument. Selling phone plans is a socialist argument. They're not the same sell.
What about the American examples? Because I assume there are some.
UTOPIA Fiber in Utah. Founded two thousand four by eleven cities, now twenty. It's an inter-local government agency that owns the fibre but cannot legally provide internet service. It wholesales to as many as nineteen ISPs.
Cannot legally.
It's in their own framing. Because of the municipal ownership, they cannot provide internet services, only the infrastructure. And their analogy is deliberately folksy. Similar to how your city provides and maintains roadways, your city partners with us to build and maintain fibre roadways.
Roads. That's the frame that sells it.
Their CEO, Roger Timmerman, makes a sharp point about the difference between real open access and what he calls fake open access. He points at the AT&T and BlackRock joint venture, Gigapower, and says it primarily supports AT&T. Nineteen ISPs versus one preferred tenant.
Kentucky?
KentuckyWired. A state-run open-access middle-mile network, thirty-four hundred miles, in every county in the state. Vermont went the other direction. Act seventy-one in twenty twenty-one transferred state fibre assets down to local communications union districts, which administer a three hundred and forty route-mile middle-mile network.
So the American version is municipal and state-level, not federal.
And it's mostly middle-mile. They build the long haul and let local providers do the last mile. Which is a smaller commitment than Daniel's proposal but a much easier political sell.
Let's go to electricity, because Daniel asked about it and the evidence there is older and messier.
Norway's Statkraft is state-owned hydropower, and Norway has among the lowest power costs in Europe, with profits returned to the treasury. Los Angeles has the largest municipal utility in the country and lower rates than PG&E serves. Germany's municipal waterworks are almost all publicly owned and nobody thinks that's strange.
And France is nationalizing EDF.
At a cost cited around nine point seven billion dollars. They had partially privatised it and they're buying it back.
Which is the opposite direction of the drumbeat.
And then there's the counter-example that I think is the most important one in the whole episode. PG&E in California. A private utility that made billions in profit while failing to maintain its own infrastructure. Wildfire survivors have been campaigning to nationalize it.
Billions in profit, failed maintenance.
Both at once. Which is the thing the privatization argument doesn't have an answer for. If the private operator captures the savings from deferred maintenance as profit, and the cost of the failure lands on everybody else, the incentive structure is backwards.
So the choice isn't public versus private.
It's which layer is public and which is private. That's the reframe. Nobody's arguing the state should sell phone plans. The argument is about who owns the trench.
There's a gap in the research here, and I want to flag it because Daniel raised it directly. The privacy concern.
It's under-covered. I couldn't find a single source that addresses it head-on. Everything emphasizes the anti-discrimination benefit of a wholesale-only owner and then stops.
So let's reason about it. What does a passive infrastructure owner actually see?
At the passive layer, it sees which premises are connected, when, and at what speeds. It sees physical metadata, not content. It doesn't see your retail relationship, because you buy from a virtual operator, not from the pipe owner.
Which is different from what a state retail operator would see.
Very different. A state that sells you your phone plan sees your billing, your usage, your customer service records, everything a carrier sees. A state that owns the duct sees that a fibre terminates at your building. That's a real privacy question, but it's a much smaller one.
It's the difference between the post office reading your mail and the post office owning the road the mail truck drives on.
And the road analogy is exactly the one UTOPIA uses, which tells you something about how they want people to think about it. Roads are public. Nobody thinks the government owning the road means the government reads your letters.
What about the talent problem? Daniel says the state would need to become the repository of talent for maintaining the network at the physical layer.
I think that concern is smaller than it looks, and here's why. Maintaining only the fibre line is a completely predictable cost structure. You're doing civil engineering and splicing. That's a workforce you can train and keep. Where costs become unpredictable is running the full network. Routing, data centres, customer premises equipment, retail operations. That's the part with the surprises.
So the state takes the boring half.
The state takes the half where the work is digging trenches and fusing glass, and it leaves the half where the work is marketing and billing and troubleshooting somebody's router to the private sector. Which is also the half where competitive pressure actually does something useful.
Which brings us back to the drumbeat. Privatisation for lower prices.
The evidence complicates it in both directions. Israel's win required the state to coerce private incumbents, not to unleash them. PG&E shows private profit coexisting with failed maintenance. And Liechtenstein's network was built by a state power company that expected utility returns over a hundred-year horizon.
No private investor has a hundred-year horizon.
No private investor can. The fund has to return capital to somebody. That's not a moral failing, it's a structural fact. And it's the deepest argument for public ownership of the passive layer. It's the only actor that can hold an asset for a century.
Which is a strange thing to hear on this show.
It's a specific claim about a specific layer. Not a general theory of government.
Alright. I want to know what actually happens on the ground, because we've been talking about the state as if it's one thing.
Hilbert: The trench was already open. That's the whole reason it happened.
Go on.
Hilbert: I spent a summer splicing for a municipal utility. Water department. They were replacing mains on a street, and the utility director had been in a fight with the phone company for about six years over a pole attachment, so when the trench was open he told the crew to drop a duct in it. Two-inch. Nobody approved it. It was in the ground before anybody thought to ask.
So the fibre went in because a man was annoyed.
Hilbert: The fibre went in because the trench existed and he didn't like the phone company. The planning document came later. I've seen the planning document. It's four pages and it's dated after the duct was buried.
And the duct was sized for water mains.
Hilbert: Sized for nothing. It was a spare. We pulled cable through pipe that was never meant to hold cable, and it fought us the whole way. Took three weeks longer than the estimate. But it's still in there. Fibre doesn't care what it's inside of.
That's the whole argument, isn't it. The infrastructure doesn't need to be purpose-built. It needs to be owned by somebody with a trench and a grudge.
Hilbert: The municipal utility and the city government were two different entities. Different budgets, different timelines, and a mutual grudge going back to a dispute over a parking lot. When you say the state, you're saying one word for four organisations that don't return each other's calls.
That's a fair correction.
Hilbert: I have to go. The pharmacy on the corner closes at six and I've got a prescription waiting that I've now missed twice.
So the coordination argument holds, but not because anybody coordinated.
Because the trench was already open and one man was annoyed. That's a much more honest version of the story than the one in the factsheets.
Which raises the real open question. The model works, and it works in a form Daniel didn't quite describe. The state is never the retail seller. It's the passive infrastructure owner. Does that distinction matter for the political sell, or is it just a detail that gets lost the moment somebody says the word nationalisation?
I think it matters enormously, and I think it's the thing to watch. If the model keeps proving itself at municipal and small-state scale, the privatization debate stops being public versus private and becomes a question about layers. Which is a much more useful argument than the one we've been having.
That's the episode. Thanks to Hilbert Flumingtop for producing. If you're enjoying the show, leave us a review. It helps other listeners find us.
This has been My Weird Prompts.
We'll be back soon.
